Craig Turnbull (Auth.) - A History of British Actuarial Thought-Palgrave Macmillan (2017)

Craig Turnbull
A History of British Actuarial Thought

CraigTurnbull
A History of British
Actuarial Thought
CraigTurnbull
Actuary, Edinburgh,
United Kingdom
ISBN 978-3-319-33182-9ISBN 978-3-319-33183-6(eBook)

DOI 10.1007/978-3-319-33183-6
Library of Congress Control Number: 2016955056
The Editor(s) (if applicable) and The Author(s) 2017

This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the
whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or
information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does
not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective
laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book are
believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors
give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions
that may have been made.
Printed on acid-free paper
This Palgrave Macmillan imprint is published by Springer Nature

The registered company is Springer International Publishing AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
To Lachlan and Leyla
Introduction
To paraphrase Winston Churchill, the title of this work doubtlessly requires

the reader to indulge the authors terminological inexactitude. It certainly
strives to tell a history, though it is told by an actuary rather than a historian.
It focuses on a history that unfolded in Great Britain, yet the start of the his-
tory pre-dates the formation of that nation (and at times during the writing of
this work it appeared as if its publication might post-date its demise).
It is intended to be a work on the development of actuarial thoughtthat
is, the intellectual basis for good actuarial professional practice. The story can-
not, of course, be told without reference to actuaries professional bodies and
the financial institutions in which they have worked, but this is not a history
of those organisations. One of the central themes of how actuarial thought
has developed through the centuries has been the way in which the profession
has (or has not) utilised relevant ideas that have been developed from outside
the profession. A proper examination of this theme requires some familiarity
with this external work and an understanding of how it developed histori-
cally. Thus a substantial portion of this history of British actuarial thought is
spent surveying the work of late-eighteenth-century French probabilists and
American financial economists of the second half of the twentieth century.
This perhaps poses the question of whether a history of actuarial thought
should be specific to any geographical region. This author would argue that
the development of actuarial thought, as loosely defined above, generally bears
little comparison with the progress of a universal, physical science. Good pro-
fessional practices are shaped by the society in which the group of profession-
als practice, as well as by theoretical or conceptual technical developments:
a societys laws, economic development, legal framework, tax regime and
the form of financial risk-sharing between individuals, generations and the
vii
viiiIntroduction
state that are acceptable to the societys social and political culture all help
to define the context and parameters within which the actuarial profession
must exercise its role. These factors mean that, for example, the intellectual
research interests of British and continental European actuarial professions
over the nineteenth and twentieth centuries have often been markedly differ-
ent (though perhaps increasingly less so by the end of the period).
A history is about the past rather than the present. Our story therefore fin-
ishes some years before todayat around the start of the twenty-first century.
In addition to providing an appropriate distance from contemporary actuarial
activities, this also represents a natural end-point. Around this time, each
of the key practice areas of the British actuarial professionlife, pensions
and general insurancehad, after some traditionalist resistance, undergone
intellectual revolutions of varying degrees that offered the prospect of better
coming decades than those experienced by the British profession at the end of
the twentieth century. The concluding section of the book, however, provides
some final reflections on the possible implications of the tumultuous finan-
cial years since the global financial crisis of 2007/8 for actuarial thinking and
practice.
This history attempts to span 350 years, the three core actuarial practice
areas and the most relevant external developments in the fields of probability
and statistics and financial economics. Inevitably, what has been striven for in
breadth has necessitated compromises in depth. For the reader whose inter-
ests have not been satiated, the author can heartily vouch for the pleasures of
reading some of the texts discussed in this work. Those with a deep interest
in the history of probability and statistics should read Hacking, Stigler and
Todhunter. For the history of financial economics, try Rubinstein. Walford
and Ogborn will provide the interested reader with a fascinating history of
eighteenth- and nineteenth-century British insurance. And for the most
inspirational, original British actuarial thought, the reader will not be failed
by the papers of Richard Price, John Finlaison, Frank Redington or Sidney
Benjamin.
Acknowledgements
Over the course of my career to date, I have had the great fortune of work-
ing with many actuarial thought-leaders who have provoked, stimulated and
inspired my own thinking. A few of these deserve particular credit for the
direct support, input and advice they have provided throughout the develop-
ment of this book. I would particularly like to acknowledge Seamus Creedon,
Peter McDade and Colin Wilson in this regard.
The librarian team at the Institute and Faculty of Actuaries were unfailingly
responsive to my seemingly endless stream of requests for obscure historical
journal papers. The library is a remarkable asset of the profession that is easy
to take for granted.
As an actuary whose only professional examination failure was in the writ-
ten communications paper, I never presumed to write a bookit just sort of
happened. It could not have happened without the patience, support, love
and understanding of my family.
ix
Contents
1 Probability andLife Contingencies, 16501750:

TheFirst One Hundred Years1
2 Revolutionary Developments Between 1750 and181037
3 Life fromtheNapoleonic Wars totheSecond World War71
4 A Brief History ofFinancial Economics forActuaries145
5 Life Offices After theSecond World War:

TheUnderwriting andManagement of
Financial Market Risk (19522004)187
6 British Actuarial Thought inDefined Benefit

Pensions (19051997)233
7 British Actuarial Thought inGeneral Insurance (18511994)275
Conclusions: Looking Back, Looking Forward311
Bibliography323
Index337
xi
Author Biography
Craig Turnbullis an Edinburgh-based Fellow of the Institute and Faculty of

Actuaries. He has published research papers in leading actuarial journals in the fields
of life and pensions and was the recipient of the Institutes Peter Clark Prize for best
research paper in 2013. He has served as a member of actuarial working parties
at both the institute and the Actuarial Association of Europe. He is a regular
speaker at actuarial conferences around Europe, and occasionally further afield,
most frequently on the topics of investment, capital modelling and asset-liability
management.
Craig was a Managing Director at Barrie & Hibbert, the leading finan-
cial risk modelling firm. In his role there he was based for several years in
NewYork as well as in the UK.He is now an Investment Director at a major
global asset management house. In his free time, he (clearly) likes reading
history.
xiii
1
Probability andLife Contingencies,
16501750: TheFirst One Hundred Years
Coming at the end of the Renaissance period and some 100 years after the
Reformation, the mid-seventeenth century is generally regarded by histori-
ans as part of the early modern epoch. We should not, however, infer from
this label that for Britain it was a time of stability in its political institu-
tions or sophistication in its financial institutions. England spent the middle
of the seventeenth century at war with the Netherlands, Scotland, Ireland
and itself. The country recovered quickly from the tyranny of Cromwells
Commonwealth; the arrival of King William III from the Netherlands in
1689 brought peace with the Netherlands but still more wars in Europe, this
time with France. In the second half of the seventeenth century, England
lagged behind some of northern Europe, and most notably the Netherlands,
in the sophistication of its financial systems. King William imported Dutch
practices in the raising of long-term government funding, and his costly wars
had much use for them.
Marine insurance was a well-established commercial activity in England and
continental Europe by 1650. European trade and early colonisation created
demand for the pooling or transfer of the risks associated with the transporta-
tion of valuable goods over long distances by sea. In contrast, very little life
assurance business was conducted at this time. There were no significant finan-
cial institutions in England in 1650 that existed for the primary purpose of
providing life assurance. The economy was predominantly agricultural. The
industrial revolution and the professional middle class that it would foster were
over a century away. Life assurance as insurance for a familys loss of a bread-
winners future salaried income was a concept whose time had not yet come.
The Author(s) 2017 1

C. Turnbull, A History of British Actuarial Thought,
DOI10.1007/978-3-319-33183-6_1
2 A History of British Actuarial Thought
This was also the midst of what is sometimes called the Scientific
Revolutionthe era of Galileo, Newton and Leibnitz. This title reflects the
exceptional intellectual progress that occurred in the seventeenth century,
which saw the development of the scientific method and the mathematisa-
tion of nature. Intellectual progress in the fields of probability and life con-
tingencies up to 1650 had been rudimentary. These disciplines, however, both
now started to attract some of the greatest mathematical and scientific minds
of this period of remarkable intellectual progress. The initial seeds of develop-
ment in each of these fields were planted side by side and followed strikingly
contemporaneous paths of growth over the next 100 years. The pricing of
life contingencies emerged as one of the most relevant empirical applications
of the evolving ideas of probability and statistics during this era and some of
probabilitys leading figures actively applied these new concepts to the simi-
larly nascent field of the pricing of life contingencies.
robability andLife Contingencies

P
intheMid-seventeenth Century
Probability emerged as a discipline of applied mathematics in the second half
of the seventeenth century. At this time, probability was focused on combi-
natorial problems and was usually applied to games of chance. The general
objective was to derive probabilities for the possible outcomes of well-defined
games where the underlying probability process was known (for example,
the distribution of the sum of n fair dice). Statistical inference is the inverse
problemthat is, the estimation of the characteristics of an unknown popu-
lation based on a limited sample of n data points. Statistical inference fol-
lowed as an application of probability but, as we shall see later, the transition
from probability to statistics involved many mathematical and conceptual
difficulties, and only fully emerged in second half of the eighteenth century.
Some antecedent discussions of probabilistic thought can be found
in historical literature stretching back at least as far as the ancient Greeks,
where gambling was an active pastime. There is some evidence of dice-based
gambling games as far back as ancient Egypt. These games prompted some
intellectuals to make probing observations in the field of randomness and fre-
quency, but no probabilistic thinking from those days contains any significant
mathematical content. There was nothing remotely close to a probabilistic
equivalent of Euclids Elements. The reasons for this are the subject of much
historical and philosophical debate.1
1
See Hacking (1975) Chapter 1 for a survey of ancient probabilistic thinking and a discussion of what
may have limited it.
1 Probability andLife Contingencies, 16501750... 3
We have to go forward another 2,000 years to the thinkers of the Scientific

Revolution of the sixteenth century before important precursors to modern
probabilistic thinking can be identified. Cardano and Galileo provide two
such examples. Gerolamo Cardano, a Dutch physician of high repute, wrote
Liber de ludo alae, a significant book on the application of elementary combi-
natorial calculations to games of chance in the middle of the sixteenth century.
The book analysed such problems as the probability distribution of the sum
of two and three dice throws, enumerating the results of the various equally
possible combinations of throws and hence outcomes for the sums. It was not
published, however, until 1663, 100 years after it was written. Therefore, as its
ideas had already been superseded by the time it was published, it was never
influential in the early development of probabilistic thinking.
Galileo Galilei was the first of the great mathematicians and scientists who
contributed to the development of probability, though his interest was no
more than passing. Like Cardano, he considered the distribution of the totals
of the throw of a number of dice. Apparently a friend or student noted that
three dice can total each of nine or ten through six distinct combinations.2 Yet
gamblers viewed a total of ten as more likely than nine. Galileo tabulated all
216 possible results of the throwing of three dice and showed that a total of
ten would be achieved in 27 of the cases whilst nine would arise in 25 cases.
QED. This is an interesting anecdote in a couple of ways. Firstly, it highlights
how tricky raw combinatorial logic can be to the human mind (the friend was
correct to say each total could be achieved by six different distinct combina-
tions of three dice scores, but he didnt allow for the fact that these six com-
binations were not equally likely to occura throw of 4, 3, 3 occurs in three
different ways (i.e. 4, 3, 3; 3, 4, 3; 3, 3, 4), whereas a throw of 3, 3, 3 can only
occur one way. Secondly, it highlights that small differences in probability
could be discerned through empirical observation (27/216 versus 25/216),
even though no systematic records of empirical frequencies are known to have
been made.
Prior to the 1650s, probabilistic thinking was limited to some elemen-
tary combinatorial analysis of games of chance such as those of Cardano and
Galileoprobability at this time might be considered pre-mathematical.
Conceptually, it was confined to calculating the ratio of favourable events to
a total number of equally possible events in well-defined circumstances. So
what of the practice of life contingencies during this period? There were two
forms of life contingency in Renaissance Europeshort-term life insurance
and life annuitiesand their histories are quite different.
2
A total of ten can be produced by six distinct combinations: 6, 3, 1; 6, 2, 2; 5, 3, 2; 5, 4, 1; 4, 4, 2; 4, 3,
3. A total of nine can be produced by the six distinct combinations: 6, 2, 1; 5, 3, 1; 5, 2, 2; 4, 4, 1; 4, 3,
2; 3, 3, 3.
As noted above, marine insurance and, to a lesser degree, fire insurance

developed into significant commercial activities much earlier than life insur-
ance. Marine insurance was established in major European ports such as
Genoa, Barcelona and Amsterdam from as early as the fourteenth century.
Some short-term life insurance policies were written in England from as early
as the sixteenth century, but such life insurance writing was essentially an ad
hoc offshoot of the core activities of the established marine insurance under-
writers. No underwriters appear to have existed in this period for the primary
purpose of providing life insurance products.
Approaches to the pricing of these life insurance contracts would have been
similar to those used by the underwriters in other areas of their business. Even
though significant volumes of marine insurance were being written in the
sixteenth century and earlier, no attempt was made to record the statistical
frequency of shipwrecks until the late seventeenth century. No intellectual
apparatus for statistical inference existed at this time, and even if it had done,
it would arguably have been of limited use: given the variability of the politi-
cal and economic environments of the times, the past may simply not have
provided much useful information about the future. Standardised tables of
maritime insurance prices did arise in various portsinsurers were clearly
risk-sensitivebut their methods were not derived from statistical analysis of
empirical data. Similarly, life insurance at the time would have been priced
based on rules of thumb and with regard to the specific known characteristics
of the life insured.
One of the main uses of early life insurance business was to provide protec-
tion from the risk that significant creditors died before repaying their debts.
It was also used to speculate on the potential impending deaths of famous
people of the time, such as the pope or monarchs. Such was its association
with gambling that life insurance was made illegal in most of continental
Europe (but not in England) by the end of the seventeenth century and this
remained the case until the nineteenth century.
Annuities were more established than short-term life assurance as a form
of life contingency contract in Europe in the Renaissance period. Whilst the
speculative abuse of life insurance resulted in it being outlawed, annuities,
on the other hand, were a useful instrument for borrowing in a lawful way:
the riskiness of the cashflows was a crucial characteristic in circumventing
the usury laws of the time. As a result, many European cities of the fifteenth,
sixteenth and seventeenth centuries issued annuities as a form of municipal
funding. No systematic allowance for age was made in setting these annuity
prices, though in some cases individual underwriting may have been used to
set the annuity price.
Pascal andFermat (1654)

Blaise Pascal and Pierre de Fermats correspondence in a series of letters over
the summer of 1654 represents the starting point for mathematical probabil-
ity. The two men were the pre-eminent mathematicians of their generation
(Newton and Leibnitz came a little later), and they achieved lasting fame
for their work in the physical sciences and number theory respectively. They
initiated a long line of seventeenth and eighteenth century mathematicians
and physicists of the highest order who made fundamental contributions to
probability and statistics whilst dedicating most of their time to other areas of
scientific research. Halley, De Moivre, Gauss and Laplace certainly fall under
this category. Leibnitz might also be added, though his contributions to prob-
ability were less direct.
Pascal and Fermats letters discussed solutions to what was known as the
problem of points. Suppose we have a game of chance between two play-
ers where the winner is the first player to reach a specified number of points
and the winner receives the two stakes of the players. The two players play
consecutive rounds until one player reaches the required number of points;
in each round one player wins one point, and we assume the players have
equal probability of winning any round. If the players agree to stop playing
part-way through their game, how should they split the stakes? To take the
specific example used in the letters: if the stake to play the game is 32 pistoles,
the winner is first to reach three points; the first player has two points and
the second player has one point, so how should the total 64 pistoles be split
between the two players?
The problem of points was not newit was discussed at length in fifteenth
century Italy and then again by Cardano in the sixteenth centurybut a solu-
tion had continually eluded each periods best mathematicians. It may appear
surprising that such a seemingly trivial problem caused the finest mathemati-
cal minds of the day such difficulty. After all, the first half of the seventeenth
century saw complex mathematical developments being applied in fields such
as astronomy with startling results. But the maths of the problem of points
was not the hard part: the type of thinking required to solve it was conceptu-
ally new.
Pascal and Fermat both independently arrived at the correct solution to
the problem of points, albeit in subtly different ways. Both approached the
problem by defining the mathematical expectation of each players pay-out
from the game as the fair settlement. This mathematical formulation of expec-
tation was the crucial conceptual breakthrough required for the problem of
points and it represented new territory. The Pascal and Fermat letters provide
the first recorded instance of mathematical expectation being explicitly calcu-
lated. But neither of them used the term expectation in their correspondence.
It was the Dutch mathematician, Christiaan Huygens, who first introduced
the term expectatio in the Dutch translation of his 1657 paper on why it
was appropriate to use mathematical expectation to price claims on uncertain
cashflows (he focused mainly on fair lotteries).3 Pascal and Fermats 1654 cor-
respondence was not published until 1679, but Huygens visited Paris in 1655
and, whilst he did not meet Pascal or Fermat, he was introduced to their ideas
by mutual acquaintances.
The idea of using mathematical expectations to value uncertain claims can
be viewed as a more technical treatment of the already established legal doc-
trine of equity, and its application to the treatment of aleatory contracts.4 The
term aleatory referred to contracts where there was a settlement of a fixed
payment today in exchange for uncertain future cashflows. Insurance and
annuities fell under this category, and so too did a wider array of contractual
arrangements such as settlement of inheritance expectations or, more gener-
ally, risky business investments. Like all contract law, a principle of fair and
equitable treatment for both parties existed. For aleatory contracts, a quali-
tative, heuristic notion of expectation was already part of established law in
the mid-seventeenth century. This did not make explicit use of probabilities,
but did recognise degree of likelihood as an important factor in determining
equitable contract settlements. From here, the development of mathematical
expectation can be viewed as a natural quantitative development.
Fermats solution to the calculation of mathematical expectation was to
tabulate all the possible (and equally likely) combinations of wins/losses that
could occur over the remainder of the game. He then worked out the prob-
ability of each player winning by enumerating the winning combinations and
dividing by the total possible number of combinations. The equitable split of
the stakes was then found as the winning probability multiplied by the total
stake, i.e. the expected pay-off from the game. This basic concept of aleatory
probability as a ratio of the number of equally possible favourable events to the
total number of equally possible events was not newas we saw above, it was
used by Cardano and Galileo a century earlier. The breakthrough was in the
conception of mathematical expectationthe probability-weighted value
and its use as a measure of fair value. But from a mathematical perspective,
3
Huygens (1657).
4
See, for example, Section 1.3, Daston (1988) for a scholarly discussion of the legal doctrine of equity
and its influence on the development of mathematical expectation.
Fermats approach is unexciting. Furthermore, the process of tabulating each

possible combination would be increasingly computationally cumbersome as
the number of players or the number of points required to win the game
became larger.
Pascals proposed method was essentially mathematically identical to
Fermats, but had a subtle difference in implementation which was argu-
ably quite profound. Pascals solution used a method of backward recursion
to calculate the mathematical expectation that applied at any point in the
game. Todhunters translation5 of Pascals explanation of his approach is worth
recounting in full:
Suppose that the first player has gained two points and the second player one
point; they have now to play for a point on this condition, that if the first player
gains he takes all the money which is at stake, namely 64 pistoles, and if the
second player gains each player has two points, so that they are on terms of
equality, and if they leave off playing each ought to take 32 pistoles. Thus, if the
first player gains, 64 pistoles belong to him, and if he loses, 32 pistoles belong
to him. If, then, the players do not wish to play this game, but to separate with-
out playing it, the first player would say to the second I am certain of the 32
pistoles even if I lose this game, and as for the other 32 pistoles perhaps I shall
have them and perhaps you will have them; the chances are equal. Let us then
divide these 32 pistoles equally and give me also the 32 pistoles of which I am
certain. Thus the first player will have 48 pistoles and the second 16 pistoles.
Next, suppose that the first player has gained two points and the second
player none, and that they are about to play for a point; the condition then is
that if the first player gains this point he secures the game and takes 64 pistoles,
and if the second player gains this point the players will then be in the situation
already examined, in which the first player is entitled to 48 pistoles, and the
second to 16 pistoles. Thus if they do not wish to play, the first player would say
to the second If I gain the point I gain 64 pistoles; if I lose it I am entitled to 48
pistoles. Give me then the 48 pistoles of which I am certain, and divide the
other 16 equally, since our chances of gaining the point are equal. Thus the first
player will have 56 pistoles and the second player 8 pistoles.
Finally, suppose that the first player has gained one point and the second
player none. If they proceed to play for a point the condition is that if the first
player gains it the players will be in the situation first examined, in which the
first player is entitled to 56 pistoles; if the first player loses the point each player
has then a point, and each is entitled to 32 pistoles. Thus if they do not wish to
play, the first player would say to the second, give me the 32 pistoles of which I
am certain and divide the remainder of the 56 pistoles equally, that is, divide 24
Chapter II, Todhunter (1865).

5
pistoles equally. Thus the first player will have the sum of 32 and 12 pistoles,
that is 44 pistoles, and consequently the second will have 20 pistoles.
This is the first record of a backward recursive method being used to evalu-
ate expectations through a path of stochastic steps. Its appeal to Pascal lay in
its relative computational efficiency and elegance, but, in retrospect, it offers
a great insight into the behaviour of the prices of assets with claims on uncer-
tain cashflows: his method provides insight not only into how to price the
claim at any given point in time, but also on how that price will change as
uncertain events crystallise. Those with a familiarity with standard option pric-
ing theory will immediately recognise the similarity between Pascals logic
and the binomial tree approach to option pricing6 (which is used both to
intuitively illustrate the mathematics of option pricing to the uninitiated and
to provide solutions to path-dependent option pricing problems that are too
complex for analytical treatment). But the path behaviour of asset prices was
not of interest to Pascalwhen he realised that the arithmetical triangle could
efficiently identify the binomial coefficients required by Fermats brute force
method, he advocated using this approach to solving the problem of points
and similar combinatorial problems.
In summary, Pascal and Fermat s solutions to the problem of points weaved
together a handful of concepts that were new or at best half-baked at the time
of their writing:
Mathematical expectation as the probability-weighted sum of uncertain

outcomes, where the probability is calculated by defining the set of exhaus-
tive and equiprobable outcomes and counting the relevant sub-set.
The principle that an equitable claim (price) on an uncertain cashflow
should be set by assessing the mathematical expectation of the cashflow.
This can be viewed as an explicit probabilistic rendering of the legal con-
cept of equity that had been an established part of the civil law relating to
aleatory contracts and which was applied using heuristic judgement.
Backward recursion as an efficient way of calculating expectations (and
hence prices) through stochastic paths.
The arithmetic triangle as an efficient computational means of producing
binomial coefficients. Pascal did not invent the arithmetic triangle ( Cardano
discussed it in the sixteenth century, and the Chinese and other ancient
civilisations are also thought to have known of it). But Pascal was the first
to identify its use in binomial expansions and how that could be applied to
probability problems.
Cox etal. (1979).

6
Together, these ideas and insights represent a fundamental breakthrough

in thinking, both with respect to mathematical probability and to the related
topic of valuing claims on uncertain cashflows.
J ohn Graunt, Johan de Witt andTheir

Prototypical Mortality Tables (16621671)
Amongst the small group of men who made profound contributions to prob-
ability and actuarial thinking in the seventeenth century, John Graunt stands
out for his relative mediocrity. He was not a world-leading mathematician of
his generation (Pascal, Fermat), nor was he an Astronomer Royal (Halley),
nor an international statesman of historical significance (de Witt). Graunt
appears to us as a middling merchant, a son of a tradesman without formal
academic training or any trace of earlier exposure to scientific analysis, armed
only with a peculiar interest in the London Bills of Mortality.
The London Bills of Mortality recorded the numbers of deaths and chris-
tenings in the parishes of London. They were produced weekly and continu-
ously from 29 December 1603. The Bills first appeared in 1592, but were
shortly discontinued before being resumed in 1603. Both 1592 and 1603
were years of heavy mortality arising from the plague. Part of the govern-
ments motivation for the production of the Bills appears to be to ease the
panic induced in the London population by showing that the impact of the
plague on human mortality was not as severe as rumoured.7
Graunts paper Natural and Political Observations made upon the Bills
of Mortality was published in 1662. It foreshadowed the use of statistical
analysis of population data to steer government social policy that would
emerge 150 years later with the development of government statistical offices
in Britain, France, Germany and Prussia. For example, Graunt observed that
the statistical rate at which beggars died from starvation was very low, and
suggested that, as they were already living off the wealth of the nation, they
might as well be paid a guaranteed wage to keep them off the streets. With
the use of statistical observations to form such policy arguments, it is arguable
that Graunt could be viewed more naturally as the antecedent of Condorcet,
Quetelet and nineteenth-century social mathematics rather than an as an
early pioneer of actuarial science.
As a source of statistical data to estimate mortality rates, the Bills of the
seventeenth century had serious limitations. The data provided no direct
Francis (1853).
7
information on the distribution of the population by age (so there was no

total exposed to risk available from the data). Even more fundamentally, the
information included in the records of deaths did not show the age of the
deceased. Inevitably, any table of mortality rate as a function of age derived
from this data would require some giant assumptions and extrapolations.
Undaunted, Graunt made several such assumptions to produce what can be
recognised as the first mortality table based on explicit statistical analysis.
Though the seventeenth-century London Bills of Mortality did not include
the age of the deceased, in the 1620s some parish records did start to include
a cause of death. The format of these records varied from parish to parish.
Some simply recorded whether the deceased had died of plague or not. By
1632, the parish of Westminster used a total of 63 categories to attribute the
causes of death (including, for example, bit with a mad dog). The most com-
mon causes were Consumption and Fever, and causes that were actually
references to the age of the deceased: Aged and Chrisomes,8 and Infants.
This data was recorded by Searchers, a form of local government official.
They were generally not medically trained and, for social and cultural rea-
sons, may not have been entirely unbiased in their reporting of the cause of
death. Graunt felt compelled to make rough adjustments to their records
for example, he decides that we shall make it probable, that in years of plague,
a quarter-part more dies of that disease than is set down.
The cause of death data was the key to the deductions Graunt made in
developing his mortality table. The table is not a major focus of his paper and
Graunt is not explicit in his workings. He identified the causes of death that
were most associated with death at a young age, and from this he estimated
that slightly more than one third of deaths occured by the age of six. He made
the assumption that, of every 100 births, there would be one survivor by age
76. In the intervening 70 years, he assumed there was an equal probability of
dying in each decade. This implies an annualised mortality rate of around 8 %
over the first six years of life and around 5 % for the next 70 years.9
Graunts analysis was conjectural and heuristic rather than statistically
rigorous. His focus was on providing sociological insight, not on making
improvements in the pricing of life contingencies. Nonetheless, his work on
the analysis of population mortality data broke new quantitative ground and
provided a starting point for the actuarial thinking that would later be pur-
sued by some of its greatest names. Whilst the publication of his paper had no
8
Infants that have not yet been christened.
9
The mortality rates implied by Graunts survival table increase after age 56, but this arises due to round-
ing of the number of remaining survivors from a pool that begins with the arbitrary total of 100 (for men
do not die in exact proportions nor in fractions) rather than by explicit design.
discernible impact on what limited life insurance business was practiced at the
time, it did highlight the latent potential in existing population records and
the possibilities of statistical analysis. His paper was reviewed by William Petty
in a Paris journal in 1666. In 1667, France started to collect statistical data of
a similar form to that found in the London Bills. The influence and impact of
Graunts work was confirmed when he was made a Fellow of the Royal Society
at the behest of King Charles II, who is said to have commended him to the
sceptical and snobbish fellows with the remark that if they found any more
such tradesman, they should admit them all.10
Johan de Witts life contrasts starkly with the humble background of John
Graunt. De Witt was born in 1825 as the son of an influential politician, and
by the age of 28 he had obtained the position of grand pensionaryroughly
equivalent to prime ministerof the States of Holland, and held it for the
following nineteen years. The Dutch Republic was one of the most significant
European powers throughout this period, and de Witt established himself
as one of Europes pre-eminent statesmen of his time. His political career,
and ultimately his life, was cut short in 1672 when 120,000 French troops
of Louis XIV invaded Holland. De Witt resigned but was nonetheless assas-
sinated, along with his brother, by a mob supportive of his political rival
Prince William of Orange, the future King William III of England, Ireland
and Scotland (William II in Scotland).
As a youngster, de Witt received tutelage from some of Hollands leading
thinkers of the time, from which he obtained a lifelong interest in mathemat-
ics. Between plotting geopolitical strategic alliances and fighting wars with
England and France, he also devoted some time to the theory of annuity
pricing. His interest in actuarial science was not driven entirely by mere math-
ematical curiosity. The Netherlands was advanced in the sophistication of its
approach to raising government funding and by the mid-seventeenth century
it had a long-established practice of using perpetuities and life annuities to
issue state debt. By contrast, England at the time had no facilities for raising
funded government debt until the final years of the century (it had experi-
mented with issuing life annuities in the first half of the sixteenth century,
but never in significant volume).11 De Witt was concerned that the state was
paying too much for its life annuity funding relative to the cost of perpetual
annuity funding, and he proceeded to produce the first rigorous analysis of
annuity pricing to make his point.
10
Francis (1853), Chapter 1, p.11, Hacking (1975), Chapter 12, p.106.
11
See Homer and Sylla (1996), p.112.
As a graduate of law and mathematics, de Witt was well equipped to apply

mathematical expectation as a form of legal equity. In a similar vein to Graunt,
de Witt made some sweeping assumptions about mortality rates. He weaved
together Graunts approach to basic mortality modelling with the mathemati-
cal expectations concept of valuing aleatory contracts pioneered by Pascal,
Fermat and others such as Huygens, so as to develop prices for life annuities.
In 1671 de Witt wrote a series of letters to the Estates General (the body
responsible for Dutch government funding) where he set out his analysis of
annuity pricing. He argued that with perpetual debt achieving a price of 25
times its annual interest, life annuities should be priced higher than the four-
teen times annual income at which they were being sold at the time of writing.
He first suggested that an uncertain cashflow should be valued at its expec-
tation: The value of several equal expectations or chances, a certain sum of
money or other objects of value pertaining to each chance, is found to be
exactly determinable by adding the money or other objects of value represented
by the chances, and by then dividing the sum of this addition by the number
of chances: the quotient or result indicates with precision the value of all these
chances.12 His demonstration of the reasonableness of this assumption borrows
liberally from his compatriot Huygens and showed how it was consistent with
the result obtained from a series of equitable exchanges of aleatory contracts.
In a similar heuristic style to Graunt, de Witt assumed that mortality rates
were uniform between the ages of three and 53. An important clarification
is required when describing what de Witt meant by a uniform mortality rate
assumption. He meant that, if we consider a starting pool of people of a given
age, a constant number from that pool would die every future year. The size of
the pool shrinks as people die, and so the mortality rate at which the survivors
consequently die is always increasing. Thus, de Witt supposed that, if we start
with a pool of 128 lives aged three, two would die every year over the next
50 years, leaving 28 remaining alive aged 53. This implies that the mortality
rate for a 53 year-old is more than four times higher than for a three year-
old.13 Such a progression in mortality rates would seem intuitively desirable
to modern eyes, but it does not occur in de Witts table by explicit design.
De Witt did not appear to appreciate the increase in probability of death that
the specification of a constant number of deaths from a fixed starting pool
impliedthe modern conception of a mortality rate had simply not yet been
realised. In contrast to Graunt, de Witt was happy to work with fractional
lives in his projected pool: he assumed that between ages 53 and 63 people die
from the remaining pool (of 28) at a rate of 4/3 per year; between 63 and 73
De Witt (1671).
12
128/30=6.7 %; and 1126/128=1.6 %.

13
they die at a rate of one per year; and between 73 and 80 they die at a rate of
2/3 per year. All 128 lives are therefore extinguished by age 80.
Unlike Graunt, de Witt offered no empirical basis for these mortality
assumptions. De Witt had some correspondence on these assumptions with
Johannes Hudde, a contemporary mathematically minded Dutch politician
who was mayor of Amsterdam during the period. Hudde developed an empir-
ical mortality analysis from the records of the annuities sold by the Dutch
government between 1586 and 1590. This qualifies as the first mortality expe-
rience analysis of annuity business. He grouped the data by age of the annui-
tant when they purchased the annuity and tabulated the number of years for
which each annuitant received their annuity payment. De Witt was satisfied
that this data was consistent with his assumptions, noting that the data sug-
gested that the proportion of 50 year-olds dying by age 55 was 1/6, and of 55
year-olds dying by age 60 was 1/5 (although de Witts assumptions implied
both these proportions would be very close to ).
Armed with the above mortality decrements and the 4 % interest rate
derived from the perpetual annuity price, de Witt then undertook the arith-
metic computation of mathematical expectations discounted by the time
value of money in order to calculate the price of an annuity for someone aged
three on purchase. He found that the annuity price was sixteen, thus support-
ing his initial statement that the price of fourteen, at which the annuities were
currently being sold, was too low. De Witt then went on to argue that the
selection effect of choosing a life, or person in full health, and with a manifest
likelihood of prolonged existence should significantly increase the price of
the annuity further. This is notable as perhaps the first published discussion of
the impact of selection on the price of life contingencies.
De Witts work can be viewed as a synthesis of the probabilistic and valu-
ation thinking developed by Pascal, Fermat and Huygens together with the
pragmatic mortality modelling introduced by Graunt. Its originality lay in
demonstrating how these emerging ideas could be used to rationally obtain
fair prices for annuities, and how such pricing must consider practical effects
such as anti-selection. The application of the then-recent developments in
probabilistic thinking to the valuation of life contingencies had begun.
Edmond Halleys Breslau Table (1693)

Some forty years after the ground-breaking correspondence of Pascal and
Fermat, another excellent mathematician dabbled in the emerging subjects
of probability and the valuation of life-contingent claims. Edmond Halley,
his interest piqued by the British governments plans for a significant issuance
of life annuities, presented a paper to the Royal Society in 1693 on mortal-

ity modelling and annuity pricing.14 It is little more than a footnote in the
published work of the famed Astronomer Royal, but the paper fundamentally
advanced the methods of mortality modelling and what we might now call
actuarial statistics.
Over the preceding 40 years, several seeds of actuarial thought had been
planted. Probability as a discipline of applied mathematics had been tenta-
tively established and had attracted some of the greatest mathematical minds
of the generation. This calculus of probabilities had been synthesised with
contemporary legal doctrine to provide an intellectual basis for the use of
mathematical expectations in valuing future uncertain cashflows, such as
those arising from life annuities. Graunts analysis of the empirical data pro-
vided by the London Bills of Mortality illuminated the possibilities of what
might be inferred from the analysis of existing datasets. De Witt had taken
similar mortality assumptions to those of Graunt and had applied them to
develop prices for life annuities.
Up until this time, however, empirical mortality data had not been used to
provide rigorous or granular insights into how mortality rates varied by age.
Graunt and de Witts mortality tables had relied on sweeping assumptions
and simplifications and were not based on what would be recognised today
as statistical fitting to a dataset. This was largely because of the limitations in
the empirical data. In particular, the lack of data on both age at death and
the numbers alive at each age severely limited the opportunity to conduct a
straightforward statistical analysis of probability of death by age.
In his Royal Society paper of 1693, Halley first identified the shortcomings
in the London Bills of Mortality data used by John Graunt. He noted that the
population numbers of each age were unknown; that the ages of the people
dying were unknown; and that the population size was unstable due to immi-
gration flows. Halley, apparently with the assistance of Leibniz,15 located bills
of mortality for the German town of Breslau for the years of 16871691. The
Breslau bills were the only such records known of at the time that recorded
age at death. Halley also argued that the town of Breslau was a small, sleepy
place that did not experience much immigration or emigration compared to
the growing metropolis of London. Thus, two of three defects in Graunts
London data were arguably addressed by the Breslau bills.
Halley used this superior mortality data to produce the first mortality table
with a granular description of how mortality rates varied as a function of age.
Halley (1693).
14
Hacking (1975), p.113.

15
However, his paper is silent on the quantitative steps he took to transform

the five years of raw data into this mortality table. Specifically, he does not
explicitly describe what steps he took to work around the remaining defect
of the datathe unknown total number alive at each age. Halley states that
he assumed a stable population. He most likely used this assumption to infer
the numbers alive at each age from the numbers that died: if the population
is stable, the numbers alive at any given age must be equal to the sum of all
those dying at that age and older. This observation allows the mortality rates
to be inferred without any explicit data on the numbers alive. The approach is
well-explained by Richard Price16 some 80 years after Halleys paper:
In every place that just supports itself in the number of its inhabitants, without
any recruits from other places; or where, for a course of years, there has been no
increase or decrease, the number of persons dying every year at any particular
age, and above it, must be equal to the number of living at that age. From this
observation it follows, that in a town or country, where there is no increase or
decrease, bills of mortality which give the ages at which all die, will show the
exact number of inhabitants; and also the exact law, according to which human
life wastes in that town or country.
Figure 1.1 compares the three seventeenth-century mortality tables dis-

cussed so far. Halleys table has a regularity that is absent from that of both
Graunt and de Witt, whose mortality rates are subject to jumps that result
from arbitrary assumptions. Halleys access to improved data permitted an
advancement in methodology that resulted in a mortality table recognisable
to modern eyes. To the extent there was such a thing, it remained the standard
reference mortality table until the second half of the eighteenth century.
Halleys paper went on to apply his mortality table to the valuation of life
annuities for both single and joint lives (and, in keeping with the time, these
include novel geometric representations!). In a similar vein to De Witt, Halley
concluded that government annuity prices appeared cheap relative to long-
term government bonds. Halley noted that with the long-term interest rate at
6 %, his table implied a fair annuity rate of less than 10 % for a 40 year-old,
yet the governments annuity pricing would offer an annuity income of 14 %.
Halleys motivation for his mortality and annuity investigation arose from
the English governments plans to imitate the Dutch and make more use of life
annuities as a means of government borrowing. We now turn to the English
government annuity issuance that occurred in the year of the publication of
16
Price (1772), Essay IV.
1.0
0.9 Graunt (1662)

0.8 De Witt (1671)
0.7 Halley (1693)

Mortality rate, q(x)
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0 10 20 30 40 50 60 70 80
Age, x
Fig. 1.1 Three seventeenth-century mortality tables
Halleys paper, and consider it alongside the other practices that were taking
place in the life contingencies business at the end of the seventeenth century.
L ife Contingencies at theEnd oftheSeventeenth

Century
The sale of life annuities is known to have occurred since late Roman times
and perhaps even earlier. The Roman juror Ulpian produced a table of annuity
prices in the third century AD17: a life annuity would cost a twenty year-old 30
times its annual income, whilst a 60 year-olds annuity would cost seven. The
need for these annuity prices was not driven, however, by an active market in
annuities. Rather, it was common for bequests to take the form of life incomes,
and a capital value for these income streams was required to assess estate duty.18
Further, there is nothing to suggest that a statistical approach to mortality mod-
elling was used in the derivation of these prices. In Tudor England, govern-
ment life annuities were sold at 7 for 1 of annual income (an annuity rate of
around 14 %), irrespective of age. Interest rates at the time were around 12 %,19
suggesting this annuity pricing was not particularly generous.
17
Hacking (1975), Chapter 13, p.111.
18
Ogborn (1953), Phillips in Discussion, p.196.
19
Chapter IX, Homer and Sylla (1996).
Following the glorious revolution of 1688, King William III was in much
need of funding for his European wars, and the English government looked
to raise 1 million through the issuance of life annuities in 1692. Interest
rates had fallen significantly since the Tudor annuity issuance: between the
sixteenth and late seventeenth centuries, English interest rates had dropped
from around 12 % to 6 %. Despite this substantial fall in interest rates, the
1692 annuity issuance was offered at exactly the same price as the one of the
1540s! Once again, the annuities were priced at a multiple of seven times
annual income with no variation by age or sex of the nominated life.
Halleys Breslau table implied that the governments annuity pricing was
extremely generous. He included a table of life annuity prices using an inter-
est rate of 6 %, a reasonable assumption for the long-term risk-free interest
rate prevailing at the time.20 The government also sold 99-year fixed-term
annuities at a price of 15.5annual income,21 again consistent with the 6 %
interest rate assumption. Using his table and this interest rate assumption,
Halley produced an annuity price for a ten year-old of 13.4annual income;
for a 40 year-old a price of 10.6; and for a 60 year-old a price of 7.6. Despite
the apparent generosity of the government pricing, the life annuities sold
poorlyless than 110,000 of the targeted 1 million was raised from the
sale of annuities on 1,002 lives.22 But many of the investors who did partici-
pate were well aware of the opportunity created by the age-insensitive annuity
pricing: more than half of the nominated lives were under eleven years of
age, and the annuities were therefore obtained at almost half of the fair price
implied by Halleys Breslau table.
Full records of the mortality experience of these annuitants have not been
found, but it is known that 503 of the 1,002 lives were alive in 1730; 175
lives were still alive in 1749 and that the last survivor died in 1783.23 This
mortality experience is quite similar to the ten year-old in Halleys table. The
government borrowed at a realised interest rate of 12 % when they could have
borrowed at a rate of 6 % by issuing long-dated gilts.
The historical literature offers different interpretations of the influence that
Halley had on the governments annuity pricing policy. Some modern histo-
rians have suggested that his annuity pricing recommendations were rejected
by the government.24 But Walford25 noted that the government subsequently
20
Homer and Sylla (1996) p.127.
21
Francis (1853), Chapter 3, p.55.
22
Leeson (1968), p.1.
23
Leeson (1968), p.1.
24
Daston (1988), p.139.
25
Chapter IV, Walford (1868).
16
Halley (1693)
14
Government (1692)
Annuity Price for Single Life Age x
12 Government (1703)
10
0
0 10 20 30 40 50 60 70
Age, x
Fig. 1.2 Government annuity prices and Halley
increased its single life annuity prices in a 1703 Act of Parliament from a mul-
tiple of annual income of seven to nine, and Francis attributes this revision
to the influence of Halleys paper.26 The conventional contemporary historical
position is that the early mortality tables had no discernible impact on the
practice of insurers or governments in their pricing of life contingencies.27
Figure 1.2 compares Halleys annuity prices with the government prices of
1692 and 1703.
Irrespective of its direct impact on contemporary government annuity pric-
ing policy, Halleys paper introduced a new standard of method for mortal-
ity modelling and the Breslau table remained the benchmark mortality table
until well into the second half of the eighteenth century.
We saw above how, despite the emergence of an actuarial approach to
annuity pricing, at the end of the seventeenth century the English govern-
ment did not feel compelled to move to an annuity pricing approach that
priced annuities as a function of age (or indeed any other variable). Virtually
no life insurance business existed in the private sector at this time. The growth
of this industry would gather pace in the following decades; but in the final
decade of the seventeenth century, there were only one or two examples of
businesses that were transacting life contingencies.
Francis (1853), p.59.

26
For example, see Daston (1988) Section 3.4, Hacking (1975), p.113.
27
The Mercers Company, established in London, was one such example. It

launched a form of reversionary annuity product in 1698. This product paid
a widow an annual income of 30 for the remainder of her life following
the death of her husband for every 100 of single premium. This annuity
rate did not vary as a function of the age of either the husband or the wife.
Mercers was ultimately unable to pay the annuities promised by this product.
It reduced the size of the promised annuity payments by one third in 1738
and finally required an Act of Parliament in 1747 to facilitate a partial govern-
ment bail-out.
Halleys Breslau table was published several years prior to the launch of the
Mercer reversionary annuity product. It is natural to consider what annuity
rates were implied by the Breslau table and how they compared to the prac-
tice of the Mercers Company. The Breslau table implies that the fair annuity
rate varies substantially as a function of the age of husband and wife. For a
husband and wife both aged 40, it implies that the Mercer products promised
annuity income was actually lower than the fair valueat 6 % interest and
before allowance for expenses, the Breslau table implies an annuity income
of around 45 per 100 of single premium. But for a husband aged 60 and
wife aged 30, an annuity income of only 20 can be provided for according
to the table. So the general level of the pricing is not obviously inconsistent
with Halleys table. But the fair pricing of the product is highly sensitive to
the ages of the husband and wife (and, in particular, to the difference in their
ages). The decision not to vary the pricing by age of husband or wife therefore
heavily exposed the company to selection effects.
Today we can only speculate as to whether it was this selection risk that was
the root cause of the businesss failure. The experience with the 1693 govern-
ment annuity issuance, where most of the nominated lives were under eleven
years of age, suggests enough of an understanding of annuity pricing existed
for investors to take advantage of this selection opportunity. Many decades
later, in 1772, Richard Price, of whom we will hear much more later, wrote of
the fall of the Mercers Company: The rapid fall of interest of money; their
admitting purchasers [i.e. husbands] at too advanced ages; and, particularly,
their paying no regard to the difference of age between husband and their
wives, must have contributed much to hurt them.28
The Mercers Company also offered one-year life assurance policies that
paid a lump sum on death. The lump sum, however, was not a fixed amount.
The total premiums (plus income less expenses) received from the cohort of
lives purchasing insurance for the year were divided at the end of the year
28
Price (1772), p.105.
amongst the policies of those who had died. No mortality risk was transferred
to the insurer, the company simply acted as an administrator of the direct
pooling of mortality risk amongst the policyholders.
In summary, at the start of the eighteenth century, the theoretical proba-
bilistic and actuarial developments of the preceding 50 years had not yet sig-
nificantly impacted on the life contingencies pricing practices of insurers or
governments. Relatively little mortality risk transfer took place in this era,
and the examples we do have are mainly of life annuities which resulted in
significant losses or failures for the recipients of the risk (whether government
or private sector company).
J acob Bernoulli andtheFirst Law ofLarge

Numbers (1692/1713)
Jacob Bernoulli (16541705) was the first in a veritable dynasty of Bernoullis
who made significant contributions to mathematics and physics in the eigh-
teenth century. At least five of the clan published on probability, but none
with greater influence than Jacob. His book, Ars Conjectandi, is a seminal
piece in the history of probability and statistics. It likely was written over the
1680s and 1690sand the most important part, the first limit theorem in
probability, is believed to have been written in 1692but it was never pub-
lished in Bernoullis lifetime. His nephew Nicolas, another famous mathemat-
ical Bernoulli (of whom more will be said later), edited and finally published
the book in 1713, several years after Jacobs death.
The first part of Ars Conjectandi considers combinatorial problems and
games of chance in a similar way to those dealt with by Pascal, Fermat and
Huygens over the preceding decades. This section of the book is most notable
for containing the first derivation of the binomial probability distribution.
But the most important section of the book is in Chap. 4, where Bernoulli
derives the first limit theorem in probability: Bernoullis Theorem, or what is
often now referred to as the weak law of large numbers. The theorem is the first
to consider the asymptotic behaviour of a random sample from a population
with known characteristics.
Bernoulli uses the analogy of an urn filled with black and white balls, where
the proportions of black and white balls in the urn are knownsay the pro-
portion of all balls that are black is p. n balls are randomly selected, with
replacement, from the urn. The number of black balls, m, in the n randomly
sampled balls is a random variable, as is the sampled proportion of black balls,
m/n. At the time of writing, the intuition that a larger sample size n would
result in m/n being a more stable and reliable estimate of p already existed.
Bernoulli writes: For even the most stupid of men, by some instinct of nature,
by himself and without any instruction (which is a remarkable thing), is con-
vinced that the more observations have been made, the less danger there is of
wandering from ones goal.
But no quantification of how the accuracy of a sample increased with sam-
ple size had been developed. Even more vitally, it had yet to be determined
whether there was some fundamental limit to the amount of certainty that
a random sample could provide, even for very large sample sizes. Bernoullis
treatment of these questions arguably signifies the moment where probability
emerges as a fully formed branch of applied mathematics.
Mathematically, Bernoullis Theorem, or the weak law of large numbers,
states:
limPr ( p m / n < ) = 1
n
where p, m and n are defined as above, and is an arbitrarily small number.

The proof of the theorem was mainly a matter of combinatorial algebra.
Bernoulli recognised that m would have the binomial probability distribu-
tion that he developed earlier in the book. He then used algebras Binomial
Theorem to expand the combinatorial terms arising in the probability distri-
bution and show how they behaved in the limit for n.
Bernoullis Theorem showed that a random sample of independent tri-
als, each with a known and constant probability of success, would provide
an unbiased estimate of the population success probability. Furthermore,
it showed that the noise in the sample probability asymptotically reduced
to zero as the sample size increased. This work also provided, again for the
first time, quantification of the sampling error associated with samples of
a given finite size n (under the assumption that the population probability,
p, is known). Bernoulli developed some quantitative examples. He consid-
ered the case where p=0.6, and he asked how big n must be in order for the
sample probability, m/n, to be between 31/50 and 29/50 with a probability
of 9999/10,000 (a probability level Bernoulli referred to as moral certainty).
He calculated that n needed to be at least 25,550. In his History of Statistics,29
Stigler noted that Ars Conjectandi abruptly ends with this calculation, and he
speculates that the magnitude of this sample size would have been disheart-
ening at a time when examples of large-scale empirical sampling had yet to
emerge.
29
Stigler (1986).
Bernoullis Theorem fundamentally requires the population probability, p,

to be known. The theorem tells us about the behaviour of a sample when the
characteristics of the population are already given. The inverse problem of
statistical inferenceof developing estimates of the characteristics of a popu-
lation given the observed sampleis not advanced by the theorem. It does tell
us that an infinite sample will provide an unbiased estimate of the population
probability, but it does not say anything about the behaviour of finite samples
from an unknown populationthe crux of statistical inference. For example,
if we do not know p, and we observe m=4 and from a sample of size n=10,
the theorem does not tell us anything about the interval we can infer for p
with some specified probability from these observation, or even that 0.4 is the
best estimate of p. Bernoullis writing seemed aware of this limitation, but
he also understood that a solution to the statistical inference problem was the
greater prize and he tried hard to find applications of his theorem to it.
Bernoullis efforts to wring some application to statistical inference out
of his theorem led to a series of letters between Leibniz and Bernoulli dur-
ing 1703. Leibniz disputed that Bernoullis Theorem could be meaningfully
applied to statistical inference. Bernoulli claims in a letter: Had I observed it
to have happened that a young man outlived his respective old man in 1,000
cases, for example, and to have happened otherwise only 500 cases, I could
safely enough conclude that is twice as probable that a young man outlives
an old man as it is that the latter outlives the former. But he was not able to
mathematically define safely enough, nor was he able to define in what sense
the 21 probability estimate in his example was the best estimate. Leibniz
used an analogy: for any finite number of sample points, an infinite number
of curves can be made to pass exactly through all of them, and there was no
means of establishing which one is best. Similarly, for a finite sample that
gives a particular sample probability m/n, there are many values of p that fea-
sibly could have created such an outcome, and so the choice of value remains
arbitrary. Concepts like least squares and maximum likelihoodfoundations
of statistical inferencewould take at least another 100 years before they
started to appear.
e Moivres Contributions toProbability

D
andLife Contingencies (17181724)
Abraham de Moivre is another mathematician of the highest calibre who
applied himself to probability whilst dedicating most of his energy to the
more conventional mathematical fields of his time. In de Moivres case, he
achieved particular renown in astronomy and complex analysis. He is perhaps

most well-known for his eponymous formula on complex numbers and trigo-
nometric functions, which is a precursor to the famed Euler formula (which
frequently tops surveys of the most beautiful formulae in mathematics). But
his contribution to probability is also highly regarded. Isaac Todhunter, that
most authoritative of nineteenth-century mathematical historians, writes of
de Moivre, it will not be doubted that the Theory of Probability owes more
to him than to any other mathematician, with the sole exception of Laplace.30
Born in France in 1667 to a Huguenot (Protestant) family, de Moivre moved
to London in 1687 to escape religious persecution. He lived in London for
the remainder of his life, and died in 1754. Despite being a recognised leading
mathematician and Fellow of the Royal Society, he was never able to obtain
a major position at one of the universities. He spent his entire life in relative
poverty and supported himself by working as a mathematical tutor.
De Moivre is notable in the history of actuarial thought for the publication
of two books that span the related fields of probability and life contingen-
cies: Doctrine of Chances,31 which was first published in 1718; and Annuities
on Lives,32 first published in 1724. The latter is the main focus of our inter-
est, but Doctrine of Chances is remarkable for including the first published
derivation of the normal probability distribution. De Moivre considered
Jacob Bernoullis binomial probability distribution and, like Bernoulli, he
considered the properties of samples from the distribution as the sample size
increased to very large numbers. The factorial calculations entailed by the
binomial distribution for large n were extremely unwieldy given the com-
putational limitations of the time. De Moivre was able use factorial algebra
to derive the normal distribution as a good approximation to the binomial
distribution for large n. (This calculation was further refined by James Stirling,
and now bears Stirlings name.)
At the time of publication of Doctrine of Chances, the wider implications of
the normal distribution were not yet understood. De Moivres work can now
be seen as a special case of the Central Limit Theorem that was to emerge at
the start of the nineteenth century. But it was still a very useful result at the
time of publication: it substantially reduced the computational burden of cal-
culating the probabilities associated with large sample sizes. It also provided
new conceptual insights. In particular, de Moivres work highlighted that the
variability of a sample probability calculated from a known population will
30
Todhunter (1865).
31
De Moivre (1718).
32
De Moivre (1724).
decrease in proportion to the square root of the sample size. The primitive statisti-
cal practices of the time, such that they existed in areas like the treatment of
discordant astronomical observations, had assumed that the sample variability
decreased in direct proportion to the sample size, rather than its square root.33
The section of Doctrine of Chances containing the derivation of the normal
distribution is entitled A Method of approximating the Sum of Terms of the
Binomial (a+b)n expanded into a Series, from whence are deduced some prac-
tical rules to estimate the degree of assent which is to be given to experiments.
De Moivre believed that his approximation result supported the inversion
of Bernoullis Theorem for use in statistical inference or inductionthat is,
to make statements about the likely behaviour of a population based on the
behaviour of a sample, as opposed to making statements about the behaviour
of a sample, based on knowledge of a populations characteristics. De Moivre
argued from Bernoullis Theorem that if a large sample could be observed
to converge on a given probability, then that must provide the unknown
population probability. This idea of convergence recognised that the sample
ratio is an unbiased estimator of the population probability, but de Moivre
did not suggest any means of determining how large the sample must be to
obtain convergence, or how much uncertainty is in a given sample when the
population probability is not known. De Moivres uncovering of the normal
distribution is arguably a profound moment in probability. The mathematics
of statistical inference, however, was not directly furthered by it.
De Moivres Annuities on Lives, published in 1724, was the most significant
work on life contingencies since Halleys seminal Breslau paper. De Moivre
did not seek to improve on Halleys mortality table. Rather, his objective was
to show how Halleys work could be more readily applied to the valuation of
annuities by reducing the computational burden associated with calculating
annuity prices from age-specific mortality rates. Halley himself had pointed
out in his Breslau paper that the calculation of annuity prices from the mor-
tality rates of his table will without doubt appear to be the most laborious
calculation and noted that the production of the annuity table in his paper
was the short result of a not ordinary number of arithmetical operations. De
Moivre specified a simple mathematical form for the behaviour of mortality
as a function of age that permitted a straightforward annuity pricing formula
to be obtained which required a much smaller set of arithmetic operations:
Consulting Dr Halleys Table of Observations, I found that the Decrements of

Life, for considerable intervals of time, were in arithmetic progression; for
instance, out of 646 persons of twelve years of age, there remain 640 after one
Stigler (1986), p.84.

33
year; 634 after two years; 628, 622, 616, 610, 604, 598, 592, 586, after 3, 4, 5,
6, 7, 8, 9, 10 years respectively, the common difference of those whole numbers
being 6.34
De Moivre derived a relatively simply pricing formula for a single life annu-
ity when the decrements of life followed this arithmetic progression:
1
(1 + r ) P
Annuity Price = n
r
where r is the annually compounded rate of interest, n is the number of years

until the number of lives is reduced to zero by the arithmetic progression, and
P is the price of an annuity certain payable for n years. To set the n parameter,
de Moivre assumed that the population would be reduced to zero by age 86,
and so n=86x, where x is the current age of the annuitant.
Of course, the whole of Halleys mortality table did not conform to de
Moivres assumption of a constant arithmetic progression of the decrements
of life. Figure1.3 shows the actual decrements produced by Halleys table.
This highlights that de Moivre was being somewhat selective in his quoting
of the stability of the decrement of life in Halleys table. It is indeed constant
for the ten years from age ten, as de Moivre points out (and also for fifteen
years from age 54), but there inevitably must be significant variationfirst
upwards as mortality rates rise with older age; and then downwards as the
population reduces to very small numbers.
How do the limitations of de Moivres approximation of Halleys mortal-
ity assumptions impact on annuity pricing? Consider the price of an annuity
on a 30-year-old life. De Moivres approximation assumes that 9.5 lives of a
starting number of 530 die every year for the next 56 years. Figure1.3 shows
that the actual decrement in Halleys table was eight per year at age 30, rising
to ten or eleven over the next few decades, before falling sharply from age
75 onwards. So de Moivres assumption is sometimes higher and sometimes
lower than the true values of Halleys table. The resultant approximate annu-
ity price is surprisingly accuratede Moivres price is 11.6, and the exact
price from Halleys table is 11.7 (at 6 % interest). Figure1.4 shows how de
Moivres approximation compares with the exact prices from Halleys table for
various ages of annuitants. It is less accurate for young ages, but impressively
accurate for annuitants of age 30 or older.
34
De Moivre (1724).
16 1000
Annual 900
14
Cumulative
800
Cumulative Decremnents of Life

12
Annual Decrements of Life
700
10
600
8 500
400
6
300
4
200
2
100
0 0
10 15 20 25 30 35 40 45 50 55 60 65 70 75 80
Age
Fig. 1.3 Decrements of life from ages 10 to 82 (from 1,000 births); Halleys Breslau
table
16
Exact
14
De Moivre's approximation
Annuity Price for Single Life Age x
12
10
0
0 10 20 30 40 50 60 70
Age, x
Fig. 1.4 A comparison of de Moivres annuity price approximation with exact

values calculated from Halleys Breslau table (at 6 % interest)
The remainder of Annuities on Lives is dedicated to the pricing of annuities

on multiple lives. He first focused on finding pricing formulae for joint life
annuities on two and three lives (where the annuity payments cease on the
first death). He again focused on finding computationally tractable formulae
by specifying an assumed mathematical form for the behaviour of mortality.
By doing so, he was able to find joint life annuity prices that were simple func-
tions of single life annuities such as:
M.P
Joint Life Annuity Price =
M + P rM.P
where M is value of the single life annuity on the first life; P is the value of the
single life annuity on the second life; and r the rate of interest.
The derivation of the above joint life annuity price relies on a different
assumption for the mathematical model of mortality than the one he used in
developing the single life annuity pricing approximation. Instead of assuming
that the decrements of life follow an arithmetic progression, he here assumes
that they follow a geometric progression. Expressed in modern actuarial termi-
nology, he assumes q is constant across age (though it does not need to be the
same q for each of the lives).
The assumptions used to derive M and P above, and the assumptions used
to derive the above joint life formula are not only different, they are incon-
sistent and mutually incompatible. But this does not necessarily matterde
Moivre was not deducing a theorem, he was attempting to find a compu-
tational shortcut that produced reasonable estimates. So how effective was
this formula against this criteria? Suppose we obtain the single life annuity
prices M and P above using de Moivres single life formula, and then plug
these prices into his joint life formulahow does such a joint life annuity
price compare with the exact value implied by Halleys table? Inevitably, the
additional layer of approximation amplifies the errors in the valuation. For
example, consider the joint life annuity price on two 30 year-olds. The exact
price according to Halleys table (and at 6 % interest) is 9.5. And whilst de
Moivres single life annuity price provides a very close approximation to the
exact single life price (11.6 versus 11.7), his joint life annuity price is 8.9,
whereas the exact price 9.5. Nonetheless, in the great scheme of approxima-
tions involved in obtaining an early eighteenth-century annuity price, per-
haps 5 % or so is not too bad.
L ife Contingencies intheFirst Half

oftheEighteenth Century (17001750)
We now turn again to developments in the British life assurance industry and
its institutions. At the start of the eighteenth century, insurance was under-
written by private individuals, partnerships and mutual associations. The vast
bulk of insurance activity at this time was in marine insurance. Life assurance
was essentially a spin-off activity of the individuals and businesses that were
focused on other forms of insurance.
In the early years of the eighteenth century, a number of new insurance
ventures emerged. Foreshadowing a trend that was to continue until the final
quarter of the century, many of these businesses were ostensibly life assurers,
but actually provided a vehicle for speculation on a wide range of contingen-
cies that could occur to third party individuals. The breadth of possible specu-
lations was constrained by an Act of Parliament of 1711 that made insurance
of events such as marriages, births and christenings illegal. But the legal con-
cept of insurable interest was not introduced until later in the century and, in
the meantime, life assurers continued to provide a vehicle for gambling and
speculation on the health or otherwise of (usually famous) other people. In his
review of life assurance in the early eighteenth century, Francis notes:
From 1720 much of the legitimate business had been usurped by it, policies
being opened on the lives of public men, with a recklessness at once disgraceful
and injurious to the morals of the country. That of Sir Robert Walpole was
assured for many thousands; and at particular portions of his career, when his
person seemed endangered by popular tumults, as at the Excise Bill; or by party
hate, as at the time of threatened impeachment; the premium was proportion-
ately enlarged. When George II fought at Dettingen, 25 per cent was paid
against his return.35
The most notable new insurer to emerge in the period of 17001720 was
the Amicable Society for a Perpetual Assurance Office (generally known sim-
ply as the Amicable), which was founded in 1706. Like the other businesses
that emerged around this time and prior to it, the Amicables life assurance
policies were essentially a form of mutual benefaction rather than insurance.
That is, rather than providing a fixed sum assured, the premiums received
from all policyholders in a given year were distributed amongst the arising
claimants in that year.
The joint-stock company emerged as an increasingly important form of
business corporation during this era. Such companies had been formed in
Britain in the previous century, mainly for the purposes of funding colonial
expansion. The East India Company is perhaps the most famous example. At
this time, joint-stock companies could only be established by a Royal Charter
of incorporation, and permission for such a charter would usually be granted
Chapter IX, Francis (1853).

35
by an Act of Parliament. Typically, the Royal Charter would grant the joint-
stock company some exclusive privileges in exchange for the company making
a form of contribution to the state. For example, the East India Company was
granted exclusive rights to trade with India in exchange for agreeing to loan
the government 1,662,000.
In 1720, an Act of Parliament gave permission for two new joint-stock
companies to be created for the specific purpose of writing insurance. Royal
Exchange Assurance and London Assurance were thus founded. The Act of
Parliament precluded any other joint-stock company from writing marine
insurance. Whilst this seemingly granted the two companies significant
market power, it did not prevent private individuals from continuing to
write marine insurance. As this was the long-standing practice in the marine
insurance market, it did not directly impact on the established marine
underwriters ability to compete with the new companies. But in return
for the exclusive right to be the only joint-stock companies permitted to
conduct insurance, the two companies agreed to pay 300,000 each to the
Exchequer.36 1720 was also the year of the South Sea bubble. The first half
of the year saw much activity in the forming of petitions for various new
joint-stock companies amid an atmosphere of stock market speculation.
Somewhat ironically, the Act of Parliament that provided the permission for
the Royal Exchange Assurance and London Assurance charters was primar-
ily focused on restraining the explosion of new joint-stock companies (it is
often referred to as the Bubble Act). The act went some way to pricking
the South Sea bubble, and, as a result, the new insurance corporations were
founded at the very peak of a stock market boom.
The total British life assurance market of this time is estimated to have been
no greater than a few thousand policyholders. Life assurance was not included
in the two new insurance companies charters as one of the forms of insurance
that the executive privilege applied to: initially, only marine insurance was
covered by the charter. In the months following the granting of the charter in
June 1720, the two companies major preoccupation was with the turbulent
stock market and the need to meet scheduled payments of the 300,000 gov-
ernment donation (Royal Exchange also committed, in addition, to buying
156,000 of government stock). Following some aborted equity capital issues
in the autumn of that year, Royal Exchange failed to meet the scheduled pay-
ment dates for their outstanding sum of 200,000. The companies argued
with the Treasury that their ability to pay their government obligations would
Part A of Supple (1970) provides an excellent account of the establishment and early growth of Royal
36
Exchange Assurance.
be materially improved by extending their charter to include fire insurance.

Life assurance was then tacked on to this application for an additional charter,
and this was granted one year later.
The life assurance business written by the two corporations over the fol-
lowing years was small. Between 1721 and 1783, Royal Exchange Assurance
generated an average annual premium income from life business of 550.
In contrast, the fire insurance business generated premiums of 6,000 in
1730 and 25,000 in 1780. The life assurance product offering was lim-
ited to a one-year term assurance. Though a simple product, it did repre-
sent a significant shift in practice from earlier times: the product provided
genuine insurance in the sense that a specified fixed sum assured was payable
on deaththe insurance company now bore the mortality risk. The Royal
Exchange Assurances pricing approach was very straightforward: a premium
rate of 5 % was charged, irrespective of age. This appears to have been the
norm of the era. There is no evidence of any market participant (government,
corporation, mutual association or individual) directly using Halleys or any-
one elses mortality tables to price any life assurance or annuity product in the
first half of the eighteenth century.
There has been much scholarly debate as to why life insurance practice in
the first half of the eighteenth century chose to ignore the developments in
mortality statistics of Halley and others that had emerged in the preceding
years. There are many practical factors that may have contributed to this:
The demand for the only life insurance product in existence at the time
(one-year term assurance) appears to have been so small that it failed to
provide any incentive to invest significant effort in refining its pricing.
In this era, age may not have been the most important factor in determin-
ing a fair one-year term assurance price. Individual risk factors such as class,
career, location and health may all have been considered by insurers when
deciding whether or not to write a life policy.
There would have been an intuitive understanding that for mortality statis-
tics to be useful as a set of forward-looking estimates, a temporal stability
in mortality behaviour was required that did not match the reality: the era
was one where mortality rates could vary significantly from year to year as
outbreaks of virulent diseases such as the plague came and went.
Even if such stability were present, it should be remembered that the lack
of an inferential statistical science meant that there was no understanding
of how much confidence could be placed in the estimates derived from
limited sample sizes. No one knew how to ask what the standard error was
in Halleys q(60), never mind how to answer it.
Some historians have also argued that the eras predominant use of life assur-
ance as a form of gambling and speculation was another important factor that
delayed the application of actuarial techniques to life insurance business.37
It is near impossible to apply conventional statistical methods to the analysis
of highly specific contingencies such as whether a prime ministers proposed
tax increases will result in him being the victim of a rioting mob, or whether
the king will lose his life on a foreign battlefield. Yet it may still appear odd
that no attempt was made to distinguish between the typical mortality rate
of a twenty year-old and 50 year-old in standard term assurance pricing. We
should not, however, lose sight of the fact that, at this time, quantitative sta-
tistical data had never actually been used to do anything. The first applications
of statistics to scientific problems such as astronomical observationwhich
arguably were more amenable to quantitative analysis as they had more statis-
tical regularity than social and demographic behaviourdid not fully emerge
until the nineteenth century.
aniel Bernoulli: TheBeginnings ofUtility Theory

D
andRisk-Adjusted Valuation (1713/1738)
We read earlier how, in the second half of the seventeenth century, Pascal,
Fermat, de Witt and Huygens developed and applied the concept of pric-
ing uncertain cashflows by assessing their mathematical expectation. The
discounting of expected cashflows to allow for the impact of the time value of
money was understood and well demonstrated by the annuity pricing work of
de Witt and Halley; in both cases, they explicitly calculated the mathematical
expectations of the future cashflows and then discounted them into a present
value using a government bond yield. Up until this time, this strand of think-
ing had not considered whether the valuation should make any allowance for
the riskiness of the future cashflows (except insofar as it impacts on the size of
the expected cashflows).
In 1713, a seemingly trivial valuation problem was introduced in a let-
ter from Nicolas Bernoulli (who we briefly met earlier as the editor of his
Uncle Jacobs Ars Conjectandi) to another eminent French mathematician
Pierre Montmort. This valuation riddle highlighted the possible need for a
more refined treatment of risk in the valuation of variable future cashflows.
The valuation problemknown as the St Petersburg Paradoxcan be
described as follows. Suppose we have a fair coin. The coin is tossed and if it
37
See, for example, Section 3.4.2, Daston (1988).
lands heads, player A pays player B 1. If it lands tails, then the coin is tossed
again. If it lands heads on this turn, then player A pays player B 2. If it lands
tails, then the coin is tossed again. If it lands heads on this turn, then player
A pays player B 4. And so on, with the amount player A must pay player B
when heads lands doubling each time the coin lands tails. An interesting dif-
ficulty arises when mathematical expectation is used to determine how much
player B should pay player A to play this game: the mathematical expecta-
tion of this cashflow is infinite. This was viewed as a paradox because it was
thought that no reasonable person would pay more than, say, 10 or 20 as
player B in the game, even though he would receive a pay-out that had an
infinite expectation.
The problem posed by Nicolas was eventually tackled by another Bernoulli
Daniel, another nephew of Jacob Bernoulli. Daniel was born in 1700 and
established himself as another Bernoulli mathematical genius, making last-
ing contributions to mathematics and physics, most notably in fields such as
fluid mechanics. In 1738, Daniel published a solution to the paradox in the
annals of the University of St Petersburg (from which the paradox derives its
name).38 Bernoullis solution proposed the use of a new form of expectation.
Instead of pricing a contract using mathematical expectation (i.e. by consider-
ing the product of the possible cashflows and their respective probabilities),
he proposed using what he termed moral expectation, which he defined as the
product of the utility the recipient gained from each possible cashflow and
their respective probabilities. This was the first time a quantitative concept of
expected utility was considered in the context of valuation of risky cashflows.
Bernoulli went on to show that an intuitive price could be obtained for
the St Petersburg game when moral, rather than mathematical, expectation
was used. To do this, he supposed that increases in wealth were directly pro-
portional to increases in utility (that is, a logarithmic utility function). He
noted that the risk-aversion embedded in this non-linear utility function
could rationalise demand for insurance and diversification practises (such as
dividing cargo amongst several ships) as well as help to solve the St Petersburg
Paradox.
To find the maximum price that someone would pay as player B in the
game, Bernoulli argued we should find the price that equates his expected
utility after playing the game with the utility he obtains from his wealth with-
out playing the game. If the utility function is linear, then the expected utility
from playing the game is infinite for any finite price. The non-linearity of the
logarithmic function places less value on the extremely unlikely upside pay-
Bernoulli (1738).
38
offs of the game. As a result, a finite value for the game is obtained. The risk-
aversion embedded in the utility function also implied that the maximum
value a player would rationally pay to play the game would be a function of
his current level of wealththe smaller his current wealth, the less he would
be prepared to pay to play, even though the expected pay-out from playing is
technically infinite.
Interestingly, Nicolas Bernoulli, who originally considered the paradox 25
years earlier, did not accept Daniels concept of moral expectation. Nicolas
was by this time a professor of law at the University of Basel. He maintained
that for a contract to be legally equitable it should be priced such that the
buyer and seller have an equal chance of winning or losing. Daniels more
sophisticated treatment had moved the pricing of uncertain claims from the
realm of jurisprudence to that of economics.
His cousin was not the only one to object to Daniel Bernoullis conception
of risk-adjusted valuation of stochastic cashflows. Jean DAlembert, another
contemporary French mathematician, argued that the introduction of moral
expectation was an ad hoc solution to the St Petersburg Paradox. He sug-
gested a simpler way of reducing the value that the mathematical expectation
attached to the extremely unlikely but extremely large pay-offs that arose in
the game: he argued that beyond a particular probability level, an event is not
physically possible, and it should make no contribution to the valuation. But
the choice of such a probability level is similarly ad hoc and arbitrary.
One of the implications of Bernoullis utility treatment is that the price a
player would be willing to pay to play the game increases as his starting wealth
goes up relative to the stake of the game. This implication is consistent with
modern economic thinking on pricing of risky cashflows. The pay-out from
the game is, in economic terms, a diversifiable risk. If the game can be played
an increasingly large number of times for increasingly small stakes (relative to
starting wealth), the utility function and starting wealth of the player becomes
increasingly irrelevant, and the value of the game will tend to its mathemati-
cal expectation. So Bernoullis solution to the paradox arguably only applies
when it is assumed that the game must be played for stakes that are a material
portion of the players current wealth. This line of argument was pursued by
the respected English mathematician Augustus De Morgan in the early nine-
teenth century. In his discussion of the St Petersburg Paradox he concludes:
The results of all which precedes shows us that great risks should not be run,
unless for sums so small that the venture can afford to repeat them often
enough to secure an average.39
39
De Morgan (1838), Chapter V, p.101.
The ultimate modern solution to the paradox, as argued by Samuelson,40

is that no rational person would accept playing as player A for any finite
price, and hence the game cannot be played and no paradox can arise. But
this doesnt really address the original point of the paradox, which was that, if
the game was offered, no reasonable person would intuitively be prepared to
pay more than ten or twenty times the pay-off from the first coin toss, even
though the total expected pay-off is infinite.
Concluding Thoughts
The 100 years between 1650 and 1750 saw remarkable progress across the
broad spectrum of human endeavour and societal development. The Scientific
Revolution was in full flow, with fundamental breakthroughs occurring in
mathematics, physics, astronomy and other branches of science, as well as in
the development of the scientific method itself. In Britain, the Royal Society,
founded during Charles IIs reign, had established itself as a significant arbiter
of intellectual endeavour by the early eighteenth century. The Enlightenment
and the age of reason brought important developments in philosophy and
economics that would further influence and challenge society.
From the perspective of British finance, London emerged particularly
strongly from this period. The City of London transformed itself from being
a laggard of financial sophistication relative to practices in cities such as
Amsterdam, Antwerp and Florence into a leading banking and insurance cen-
tre, with a respected Bank of England.
The developments in probability and life contingencies over the period can
be viewed as intrinsic parts of these broader scientific, economic and indus-
trial developments. Mathematical probability fully emerged as a branch of
applied mathematics over this period, and its early development fully repre-
sented its holistic nature, involving contributions from disciplines as varied
as jurisprudence and number theory. The use of statistical population data to
model mortality and analyse its implications for the pricing of life contingen-
cies developed contemporaneously. These early forms of statistical analysis
pre-dated a rigorous theory of statistical inference, and did not directly rely
on the theoretical developments in mathematical probability that were occur-
ring at the same time. The two disciplines overlapped, but were generally each
developed by a different cast of characters. Occasionally, however, titans such
as de Moivre straddled both these disciplines and made substantial and lasting
contributions to both.
Samuelson (1960).
40
Up until 1750, the developments in thinking in probability and life con-

tingencies were pregnant with possibility, but were still awaiting application
to real life. The tiny level of demand for short-term life assurance that existed
during the first half of the eighteenth century limited the incentive to develop
practical applications of the emerging actuarial science. The application of
this new discipline would require a societal change in demand for mortality
risk transfer, and a form of life contingency product that could meet the needs
of the growing, prudent and increasingly wealthy Victorian middle class of
the industrial revolution.
In the field of probability, there was an awareness of the need for a the-
oretical breakthrough to make it fully applicable to scientific endeavour.
Mathematical probability had been limited to finding probability distribu-
tions of random samples of populations with known properties. The new
scientific method would increasingly produce quantitative observational
data from which inferences would be made: modern thinking increasingly
called for inductive, rather than deductive, reasoning. The inversion of math-
ematical probability into statistical inferenceto make rigorous statements
about populations based on sample data observations, rather than to make
statements about sample probabilities based on known or assumed popula-
tion characteristicswas a problem whose solution had so far evaded all the
impressive group of thinkers who had tackled probability.
2
Revolutionary Developments Between
1750 and1810
The second half of the eighteenth century was a remarkable period of develop-
ment for both probability (and statistics) and actuarial thought (and practice)
more broadly. Two fundamentally important fields of statistical and actuar-
ial thought emerged over the period that can each now be seen as a natural
flowering of the seeds planted over the previous 100 years. First, a number
of major theoretical breakthroughs were made that created the permanent
foundations for the inversion of mathematical probability into inferential sta-
tistics. Second, the whole-of-life with-profit policy was conceived and success-
fully brought to market. This transformed life assurance from a short-term
insurance contract into a long-term savings vehicle that strongly resonated
with the emerging professional classes of late-Georgian Britain.
Over this period these two fields of mathematical statistics and actuarial
science followed increasingly distinct paths of development: the primary
application that drove statistical thinking over this period was found in the
physical sciences, and especially astronomy, rather than finance; similarly, the
success of new and more complex life assurance products meant that actuarial
thinking and practice had to wrestle with a broader set of challenges than
just making good statistical estimates of mortality rates. One man, Richard
Price, notably transcended both fields during this era, marking him as one of
the most important historical figures in early actuarial history.

DOI10.1007/978-3-319-33183-6_2
From Probability toStatistics, 17641810

Bayes andHis Application ofHis Theorem (1764)
Jacob Bernoullis binomial analysis of Ars Conjectandi had been driven by

the aim of making a breakthrough in the theory of statistical inference. He
ultimately failed in this objective, although he established some other fun-
damental probability laws along the way. Thomas Bayes built on Bernoullis
binomial analysis, and, with a novel alternative physical analogy of a table
instead of an urn, he introduced a new conceptual framework for inferential
methods.
Bayes was a Fellow of the Royal Society but by all accounts led a quiet life
as an English clergyman, largely detached from the intellectual fervour of the
period. His now famous paper An essay towards solving a problem in the doc-
trine of chances was unpublished during his lifetime.1 It, along with his other
papers and possessions, was bequeathed to a fellow clergyman, Richard Price.
Price was a significant mathematician and philosopher and he, like Bayes, was
also a Fellow of the Royal Society. As a thinker aware of the promise of statisti-
cal inference for empirical science, Price immediately recognised the value of
Bayes paper. He edited it and presented it to the Royal Society for publication
in its journal a year or so after Bayes death in 1762.
Bayes paper was both ahead of its time in concept and yet presented in an
anachronistic geometric style. Both these properties led to the paper being
somewhat impenetrable. The nineteenth-century mathematical historian
Todhunter wrote that the papers introductory discussion of the general laws
of probability was excessively obscure, and contrasts most unfavourably with
the treatment of the same subject by De Moivre.2 But the papers statement
of intent was quite clear:
Given the number of times in which an unknown event has happened and
failed. Required the chance that the probability of it happening in a single trial
lies somewhere between any two degrees of probability that can be named.
In essence, how can the population probability distribution be inferred

from the observation of a limited sample of outcomes from that population?
Richard Price viewed statistical inference as the rational solution to David
Humes sceptical problem of induction, in which it was argued that no
Bayes (1764).
1
Todhunter (1865).
2
2 Revolutionary Developments Between 1750 and1810 39
amount of empirical observation of the past could provide knowledge of what

would happen in the future.3 In Prices introduction to the paper, he argued
that the problem stated above by Bayes is necessary to be solved in order to
assure foundation for all our reasoning concerning past facts, and what is
likely to be hereafter it is necessary to be considered by anyone who would
give a clear account of the strength of analogical or inductive reasoning.4 He
believed that Bayes new approach to statistical inference provided the funda-
mental solution to this problem of inductive reasoning.
The paper is perhaps most well-known for its eponymous Bayes Theorem:
P ( B | A ) .P ( A )
P ( A | B) =
P ( B)

This theorem, when considered separately from its application to statisti-

cal inference, is a straightforward conditional probability statement that is
uncontroversial and unambiguously derived from the basic axioms of math-
ematical probability. In fact, it is so rudimentary and universal that scholars
are unsure whether Bayes was in fact the first to use it.5 But the structure
of the equationwhere probabilities of B conditional on A can be trans-
formed into probabilities of A conditional on Bopens the door to what was
then often referred to as inverse probabilities. Mathematical probability was
about making statements about the probabilities of observing specific out-
comes, given known population characteristics. Bayes Theorem alluded to a
potential means of inverting these probability statements into ones that infer
something about the probability characteristics of the population, given some
observed sample outcomes.
Bayes used a physical analogy to develop his application of this probability
theorem to statistical inference. Where Jacob Bernoulli had used an urn filled
with black and white balls, Bayes used a table and balls. Whilst his table was
two dimensional, his analogy only considers how the balls were distributed
along one dimension. The use of a table rather than a one-dimensional line
facilitated a geometric interpretation of his probability calculations.
He supposed a first ball was randomly thrown onto the table, and that this
was followed by a further n balls that were independently thrown randomly
onto the table. He considered how knowledge of whether each of the sub-
3
Hume (1739).
4
Bayes (1764).
5
Hacking (1965), p.190.
D C
n further balls
First ball
A B
w
Fig. 2.1 Bayes table
sequent n balls landed to the right or left of the first balls position could be
used to make a statement about the probability distribution of , the position
along the x-axis of the table of the first ball. This set-up can be summarised
simply in the Fig.2.1 above.
Bayes wished to infer the probability distribution of the position, , of the
first ball along the line AB, given only the number of times that the n subse-
quent random balls landed to the right or left of the first ball. For any given
value of , the probability of a subsequent ball landing to the right of the
first ball is simply the ratio of the line distances B/AB (as the balls position
on AB is uniformly distributed). For a given , the number of times, M, that
the n subsequent balls land to the right of the first ball is therefore binomially
distributed with parameters n and . So, we can write:
B n (n x)
= = x (1 )
Prob M = x
AB x

So far, this looks extremely similar to Jacob Bernoullis work on the distribu-
tion of the number of black balls chosen from an urn that has a known pro-
portion of black and white balls. But Bayes had merely laid the preparations
for his breakthrough step. He then noted that his theorem allowed him to
find an expression for the probability that is between two points b and f on
the line AB:
Prob ( M = x | ) Prob ( b < < f )

(
Prob b < < f M = x = ) Prob ( M = x )

By construction of his physical example, where each ball is independently and

randomly thrown onto the table, the position of every ball on the line AB is
uniformly distributed. This means that the unconditional probability of M =
x is 1/(n+1) for all x: the number of balls out of a total of n that lands to the
right of the first ball can take a total of (n+1) different values (0, 1, 2, n),
and if the balls are uniformly distributed then each of these outcomes are
equally likely. With this and the above equation, Bayes obtained:
( n + 1)! f x
(
Prob b < < f M = x = ) x ! ( n x )! b
(1 )
(n x)
d

Bayes had provided a rigorous mathematical solution for the probability dis-
tribution of the location, , of the first ball along the line AB, conditional on
the observed results of how many of the subsequent n randomly distributed
balls were positioned to the left or right of the first ball. Today, this distribu-
tion is known as the beta distribution. The integral is intractable for large
values of n and (n x), but is very straightforward for relatively small values.
For example, consider the case of n = 6 given in Fig.2.1 above. Without any
knowledge of the distribution of the subsequent balls, the probability of
being in the right-hand half of the line AB is clearly simply 1/2. If three of
the six balls were observed to land to the right of , the above distribution
implies that the probability would, intuitively enough, remain 1/2. However,
if, instead, four of the balls were found to have landed to the right of , the
probability of being in the right-hand half of AB would be reduced to
29/128 (slightly less than a quarter).
Thus far, Bayes work is a novel but still uncontroversial application of
mathematical probabilitywithin the physical framework set out, his results
are unambiguously rigorous mathematics. In Bayes physical model of the
table with randomly thrown balls, a uniform prior distribution for arises
by construction. It has an unambiguous physical interpretation that derives
from the set-up that the balls are randomly thrown onto the table.
Bayes argued that the inversion logic of his theorem was naturally appli-
cable to statistical inference. That is, to making conditional statements about
the properties of unknown populations (analogous to on his table) based
on observed sample data (the position of the n balls). The more general
a pplication of this approach to the problem of statistical inference required

treating the unknown parameters of the population that were being estimated
as random variables ( in the above example). The prior probability distribu-
tion would be a description of the information (or lack thereof ) known about
the parameter before the statistical observations were considered. The posterior
distribution was the parameters probability distribution after being updated
to reflect the information in the statistical observations. In his physical anal-
ogy, Bayes set-up had implied a uniform prior and he had derived a beta
posterior distribution.
This Bayesian approach to statistical inference involves two fundamental
steps that have invoked much philosophical and mathematical controversy,
particularly in the twentieth century when it was criticised by leading statisti-
cal thinkers such as Sir Ronald Fisher. The first issue is how to interpret the
idea of modelling the unknown (but fixed constant) population parameter as
a random variable. This made sense in Bayes analogy when the position was
physically distributed with a given probability distribution. But the standard
deviation, say, of a normal distribution is not distributed in this wayit is
simply an unknown fixed quantity. The use of a probability distribution to
represent the degree of knowledge or prior information available about the
parameter is a conceptual leap. To move from Bayes physical analogy to the
application of Bayes Theorem in more general statistical inference, the physi-
cal randomness of Bayes ball must be translated into an epistemological state-
ment of ignorance. This translation is in some sense philosophical rather than
purely mathematical, and as such its viability is open to some debate.
The second issue closely follows: in the absence of any knowledge or infor-
mation about the parameter, is it appropriate to assume the prior distribution
is uniform? Bayes appears to have thought deeply about this question and he
tackled it head on in the paper. He argued that for events where we absolutely
know nothing antecedently to any trials made concerning it then concerning
such an event, I have no reason to think that, in a certain number of trials, it
should rather happen any one possible number of times than another. This
assertion of equiprobability straightforwardly translates into the assumption
of a uniform probability distribution.
In modern times, this is sometimes referred to as the Principle of Insufficient
Reason and it may still be invoked as the logical basis for the uniform prior
distribution. It has been speculated that Bayes lack of confidence in the
soundness of this argument restrained him from publishing his paper dur-
ing his lifetime, but this is largely conjecture. It is equally possible that his
inability to find a workable mathematical approximation for his beta pos-
terior distribution was the main reason for his reluctance to publish. In any
case, this argument attracted little dissent until around 100 years after the
ublication of the paper. As we shall see in the next section, the Bayesian
p
uniform prior was adopted whole-heartedly by no lesser mathematicians than
Laplace and Gauss in the decades following Bayes paper. In 1854, George
Boole, an English mathematician and philosopher (and the father of Boolean
logic) was the first to publish a significant critique of the logic of Bayesian
statistical inference, arguing:
It has been said that the principle involved is that of the equal distribution of
our knowledge, or rather of our ignorancethe assigning to different states of
things of which we know nothing, and upon the very ground that we know
nothing, equal degrees of probability. I apprehend, however, that this is an arbi-
trary method of procedure.6
The philosophical argument that the assumption of a uniform prior dis-

tribution was in fact an arbitrary and therefore illegitimate general approach
to statistical inference gained further traction through the latter part of the
nineteenth century and the first half of the twentieth century. Sir Ronald
Fisher, who was one of the most important statistical thinkers of the twen-
tieth century, famously argued that, in the absence of any prior knowledge,
the assumption that the prior distribution was a squared sine function was
as logical a candidate for the prior distribution as the uniform distribution.7
In twentieth-century statistical practice, Bayesian statistical inference started
to be superseded by other methodologies such as maximum likelihood that
were argued to be less subjective or arbitrary. But the controversy reigns on,
and the Bayesian method has become increasingly popular once again in the
twenty-first century (so far!).
Laplaces Bayesian Analysis (17741781)
Pierre-Simon Laplace was the pre-eminent mathematician of the second half

of the eighteenth century. He made particularly important contributions in
the fields of astronomy and celestial mechanics. Over his lifetime, he also
sporadically committed significant time and energy to the study of probability
and statistics. He did more than anyone else of his time to develop the math-
ematics of Bayesian statistical inference (including Bayes). Laplaces employ-
ment of the Bayesian approach and the use of the uniform prior distribution
doubtless did much to add credence to the logic of the Bayesian approach.
Writing in the 1950s, Sir Ronald Fisher noted:
Boole (1854), p.370.

6
Fisher (1956), p.16.

7
The superb pre-eminence of Laplace as a mathematical analyst undoubtedly

inclined mathematicians for nearly fifty years to the view that the logical
approach adopted by him had removed all doubts as to the applicability in prac-
tice of Bayes theorem In spite of the high prestige of all that flowed from
Laplaces pen, and the great ability and industry of his expositors, it is yet sur-
prising that the doubts which such a process of reasoning from ignorance must
engender should begin to find explicit expression only in the second half of the
nineteenth century, and then with caution.8
Laplace produced two important papers on Bayesian statistical inference in

the two decades following the publication of Bayes seminal paper: in 1774,
Memoir on the Probability of Causes of Events9 was published in the jour-
nals of the French Royal Academy of Sciences; and Memoir on Probabilities10
was presented to the Academy in 1780 and published in 1781. Bayes paper
did not become known in continental Europe until around 1780the first
Laplace paper was published without any knowledge of Bayes developments.
But Laplace developed a remarkably similar line of attack to the problem
of statistical inference. He introduced his 1774 paper with the following
statement:
If an event can be produced by a number n of different causes, then the proba-

bilities of these causes given the event are to each other as the probabilities of the
event given the causes, and the probability of the existence of each of these is
equal to the probability of the event given that cause, divided by the sum of all
the probabilities of the event given each of these causes.
The above statement is Bayes Theorem with a uniform prior distribution!11

Laplace had independently developed the Bayesian framework for inverting
probability statements about the behaviour of a sample given a population
into statements about a population given sample data via the assumption of
a uniform prior distribution. Unlike Bayes however, who presented a proof
that was detailed to the point of philosophical obscurity, Laplace presents this
principle without any proof or even justification. Bayes had thought deeply
and philosophically about the fundamentals of probability and inference.
Laplace, on the other hand, was a mathematician whose primary interest was
in applying his renowned analytical capabilities to the mathematical problems
that probability presented.
8
Fisher (1956), p.20.
9
Laplace (1774).
10
Laplace (1781).
11
Stigler (1986), Chapter 3, p.103.
Laplace considered the same binomial problem as Bayes. That is, given a
sample of n binary observations of which x are successes, what is the prob-
ability distribution of the population probability of success, ? He obtained
the same result as Bayes for the posterior distribution of . We noted above
that Bayes was unable to find an analytical solution, or indeed a good approxi-
mation, for the integral when n and (n-x) were large. Laplace, however, used
his superior mathematical skills to manipulate the integral so as to provide
a much more accurate numerical solution than the crude limits that Bayes
(and Price) had been able to find. The posterior probability distribution that
could be inferred from a finite sample of binomial observations could now be
quantified quickly and accurately.
In his papers of 1774 and 1781, Laplace took a number of further impor-
tant steps beyond Bayes. Bayes paper had presented how to generate a pos-
terior probability distribution for the probability of the success of a binary
event, based on a sample of n observations and the assumption of a uniform
prior distribution. But he had not explicitly considered how to use this pos-
terior distribution to infer a single best estimate for the success probability.
Some measure of central tendency of the posterior distribution would be a
natural candidate for the estimate, but Bayes never explicitly discussed this
topic. Laplace addressed this directly, and suggested two possible approaches
for determining the best estimate: the value that makes it equally likely that the
true value will be larger or smaller according to the posterior distribution, i.e.
the posterior median; and the value that minimises the probability-weighted
sum of the absolute differences between the observed value and the best esti-
mate, where the probability weights were obtained from the posterior distri-
bution. Laplace then proved that these two approaches were mathematically
identical and hence would always produce the same value for the best estimate.
Laplace also moved beyond Bayes in terms of to what he applied the Bayesian
prior/posterior framework. Bayes paper had focused on binary events (success
or failure; a ball landing to the right or left of another). Laplace moved onto
variables that could take a continuum of sizes. His main motivation for this
was found in astronomical observation. Astronomers of the time found that
observations of an object in the sky, such as the planets Jupiter or Saturn, were
not perfectly consistent with each othersome empirical observation error
would inevitably arise from the physical, manual process of observing plan-
etary positions in the sky. How should these observations be combined to find
the best estimate of the position of the planet? This problem had bedevilled
some of the greatest mathematicians of the age, including Euler, who could not
find a satisfactory way of solving a system of inconsistent linear equations.12
Euler (1749). See Chapter 1 of Stigler (1986) for further discussion of Euler and the treatment of dis-
12
cordant astronomical observations.

Laplace wished to find a best estimate for some continuous variable V, based
on a limited sample of observations, say, v1, vn. He tackled this using the
Bayesian frameworkthat is, by obtaining the posterior distribution pro-
duced by assuming a uniform prior distribution for V, and then integrating the
conditional probabilities of observing v1, vn, over all possible values of V.
Laplaces best estimate would then be the median of this posterior distribution.
In Bayes specification of the problem, the conditional probabilities that he had
to sum across were already defined: because he was considering binary events,
the sum of observations had a binomial distribution. Laplaces more general
problem meant that he had to specify the conditional probability distribution
for observing v1, vn for a given value of V. But how could he determine a
universally useful form of distribution for these conditional probabilities?
In order to tackle this generic specification of the sampling distribution,
he made a subtle change of tack. Instead of considering the conditional prob-
abilities for v1, vn, he considered the probabilities of the differences or obser-
vation errors between vi and V: ei = vi V. This simple change immediately
allowed some intuitive general criteria to be established for the shape of the
sampling probability distribution. In his 1774 paper, Laplace specified three
criteria: the error probability distribution should be symmetric around zero
(as the errors should be as likely above as below); the error probability should
tend to zero as the error tends to infinity in both directions (small errors
should be more probable than large errors); and, of course, the area under the
error probability distribution must integrate to one. This still left an unlimited
number of possible functions that could be specified for the error probability
distribution and at the time no good reason existed to choose a particular
one amongst them. He considered some specific error distribution choices,
but was unable to find approaches that were amenable to analytical solution
for large sample sizes and that did not involve the introduction of additional
arbitrary parameters. A best estimate of a population parameter as a function
of sample data could not be mathematically defined without an explicit error
distribution. Laplaces journey to a general solution to statistical inference had
reached a dead end.
L east Squares, Errors andtheCentral Limit Theorem

(18051810)
New conceptual breakthroughs were required in order to make further prog-

ress in the general framework of statistical inference. Relatively little progress
was made in the twenty years following Laplaces 1781 paper. Laplace himself
appears to have put probability to one side and refocused on mathematical
astronomy, where he made terrific contributions. Two vital breakthroughs were

then delivered in succession in 1805 and 1809 by another two pre-eminent
European mathematiciansLegendre and Gauss. Laplace then returned once
again to the subject of probability and delivered the final fork of this trident
in 1810. The synthesis of the three new concepts that were introduced over
this five-year period represented a fundamental and permanent development
in mathematical statistics.
Like so many of his peers, Adrien-Marie Legendres primary focus dur-
ing his remarkable career had been in pure mathematics and astronomy. His
interest in probability and statistics arose through the contemporary problem
that had similarly attracted Laplace: finding a best estimate from a number
of discordant astronomical observations. In 1805 he published a short book
on modelling comets, Nouvelles methods pour la determination des orbites des
cometes.13 The book included an appendix entitled Sur la method des moindres
quarres'. In this appendix, he advocated choosing a parameter estimate by
minimising the squared errors that it produces for a given set of observations.
That is, using the notation developed above, if we have v1, vn observations
and we wish to make a best estimate of V, we would write ei = viV, and
find the estimator that minimises the sum of e12, en2. He noted that, in
this form of example, the estimator produced by least squares would be the
arithmetic mean of the sample (and this applied irrespective of the specific
form of the error probability distribution). He did not attach any profound
importance to this observation, but viewed it as a desirable property.
Legendres suggestion of the use of the sample arithmetic mean was clearly
easy to practically implement, and had a beguiling mathematical elegance.
But he appeared to have plucked the least squares fitting rationale out of
thin air. He did not provide any fundamental mathematical or metaphysical
rationale for why the least squares estimator property of the sample arithmetic
mean made it, in some deeper sense, best. Moreover, Laplace had already
established that, for an arbitrary choice of error probability distribution, the
arithmetic mean of the sample would not necessarily be best (in the sense of
being the median of the posterior distribution). Legendre had worked out
that the arithmetic mean of the sample was the least squares estimator of the
population mean, but for this observation to progress from being a curious
nicety to a pillar of a rigorous probabilistic approach to statistical estimation,
more was required.
The great Carl Gauss then took an important step towards an explana-
tion of why the least squares estimator could be mathematically described
13
Legendre (1805).
as a best estimate. Like Laplace and Legendre, he focused the bulk of his
energy and talents on pure mathematics and astronomical problems. In 1809,
he published a major paper on the mathematics of planetary orbits, Theoria
motus corporum celestium.14 Like Legendres paper on the orbit of comets,
he included as an appendix a piece on statistical inference. Gausss statistical
piece started where Laplace had finished. That is, he considered the Bayesian
approach to developing a posterior distribution based on an assumption of a
uniform prior distribution and the conditional distributions of the observed
errors. He showed that the choice of parameter value which was implied by
the median of the posterior distribution in this case was also the choice of
parameter value which maximised the value of the joint conditional probabil-
ity distribution of the errors. So, the parameter value could therefore be found
by differentiating the joint error probability distribution and setting it equal
to zero. Interestingly, this was very close to the maximum likelihood concept
that Fisher developed in the early twentieth century. However, without defin-
ing the form of the error probability distribution, Gauss had not practically
advanced beyond where Laplace had reached twenty years earlier.
To break the impasse, Gauss did something rather ad hoc. Instead of speci-
fying a form of error distribution and then deriving the optimal parameter
estimate that it generated, he specified a particular form of parameter estimate
(the arithmetic mean of the sample), and then derived the form of error dis-
tribution that implied this form of parameter estimate was optimal. He found
that the arithmetic mean of the sample maximises the joint distribution of the
errors only when the error distribution has the form:
h
z ( x) =
2 2
e h x

This is the normal distribution, though at this point in time no particular

importance was attached to it. A probabilistic rationale for Legendres least
squares method had now been produced: Legendre had shown that the arith-
metic mean was the least squares estimator, and Gauss had now shown that
the least squares estimator was the optimal estimator when the error distribu-
tion was normal. However, Gauss had not provided any good reason why the
normal distribution was the best choice of error distribution. For as long as
the normal distribution was just an arbitrary assumption for the error distri-
bution, a significant so what would hang over Gausss intriguing finding.
14
Gauss (1809).
This so what did not hang in the air for long. Laplace had returned to
the subject of mathematical statistics during the first decade of the nine-
teenth century, and, without any knowledge of the work of Gauss, published
in 1810 his Memoire sur les approximations des formules qui sont fonctions de
tres grand nombres et sur leur application aux probabilities.15 This paper con-
tained Laplaces most important contribution to probability and statistics:
the Central Limit Theorem. Speaking loosely, this said that the sum of any
independent variables would be approximately normally distributed when the
number of terms in the sum is large.
This suddenly propelled the normal distribution from obscurity to centre
stage. De Moivres derivation of the normal distribution as the limiting case
of the binomial distribution almost 100 years earlier could now be seen as
just one example of an all-pervasive phenomenon. In an instant, Gausss work
on normal error distributions went from being an interesting piece of ad hoc
analysis to being the defining statement of statistical inference. By Laplaces
Central Limit Theorem, error distributions could be assumed to be normal
(for large samples). In which case, Gausss optimal estimator for the popula-
tion mean was the arithmetic mean, which Legendre had shown was the least
squares estimator. An elegant and profound synthesis had been achieved that
would form part of the permanent foundation of inferential statistics.
The Emergence ofwith-Profits, 17561782

The previous section discussed how the work of Bayes, Laplace and others
during the decades following 1760 provided a fundamental breakthrough
in mathematical statistics. Contemporaneously, another small group of men
were revolutionising what life assurance could be and how it could be deliv-
ered in a sustainable and equitable way. Richard Price uniquely spanned both
these sets of men and their ideas, and for this he must be regarded as one of
the most important people in the history of actuarial thought. But before
Price, first came James Dodson.
James Dodson andWhole-of-Life Assurance (17561772)
James Dodson, born in England in 1710, was a friend and pupil of Abraham
de Moivre. It is reasonable to assume that his interest in life contingencies
15
Laplace (1810).
arose from de Moivres influence. Dodson published on mathematics, includ-

ing topics in annuity valuation, at a level that saw him inducted as a Fellow
of the Royal Society in 1755. He worked as a schoolmaster in London and
supplemented his income by consulting on financial matters relating to annu-
ities. The story, perhaps apocryphal, goes that he was refused a life assurance
policy by the Amicable in 1755 due to their then-policy not accept lives over
age 45 (a policy that was necessitated by their use of a flat 5 % premium rate
for annual life assurance for all permitted ages at the time). This motivated
Dodson to advocate developing rational age-dependent life assurance pricing.
In 1756, he published a treatise, First Lectures on Insurance, espousing his
ideas and he initiated a project to create a life assurer that would implement
them. This project ultimately led to the foundation of Equitable Life, though
he did not live to see it as he died prematurely in 1757, aged only 47 (perhaps
somewhat ironically, Amicable therefore did well to avoid insuring his life!).
Though Dodson was clearly an able mathematician, his permanent con-
tribution to actuarial science is not in mortality modelling or statistics, but
in his conception of a new life assurance product that could be appealing to
the emerging salaried and prudent middle class. Up until this time, the only
form of life assurance written was short-term (usually one-year) assurance.
As noted in Chap. 1, the demand for this business was very small, and was
largely limited to insuring the lives of business creditors and to speculative
gambling activities related to the mortality prospects of the celebrities of the
age. Dodson proposed a new form of life assurance product: instead of buy-
ing life assurance for one year, the policy would offer insurance over the entire
remainder of the policyholders life. Instead of a short-term insurance policy
that would only pay out in circumstances of extreme misfortune, Dodson
envisioned a policy that would pay out the fixed sum assured with certainty.
The only variable was the timing of the payment. In his Lectures, he went on
to show how this whole-of-life policy could be paid for by regular even premi-
ums. As mortality rates tend to increase with age, the regular premium in the
early years of the contract would be greater than the regular premium payable
for a single-year term assurance contract. This excess funded a reserve, which
was essentially a long-term savings element of the contract. The reserve would
be used to partly fund the claims that would arise many years later, when the
regular premiums payable in those years would not be sufficient to fully meet
the cost of that years cover. This conception of a whole-of-life policy trans-
formed life assurance into a form of long-term savings vehicle.
By paying for the contract with regular premiums, the policyholder was
providing a life annuity to the insurer. Dodson, as an expert in annuity pric-
ing, showed how the regular premium could be set such that their present
value equated to the present value of the sum assured benefit for a policy-
holder of a given age, according to a given mortality table and interest rate.
Dodson developed some example premium calculations using a mortality
basis derived from the London Bills of Mortality from 1728 to 1750. In 1728,
the London Bills had started to include age of death in its data (though even
then it was recorded only by decade rather than year of age).
Significantly, Dodson argued that the pricing basis for setting the whole-
of-life premium rate should be set by referring to the worst mortality that
was experienced in any given year in the data sample, rather than the average.
Thus he derived two sets of mortality rates: a mean deaths table and a great-
est deaths table. Dodson used the bills to show that the average number of
deaths over the sample period was 26,207, whilst the worst year experienced
32,169 deaths (this occurred in 1741, where severe weather resulted in high
food prices and near famine in parts of England). Thus the greatest deaths
basis had mortality rates that were almost higher than the mean deaths
basis. Dodson also noted that even the mean deaths basis would likely pro-
vide some margin over the experience of a well-run life assurer:
As the Bills of Mortality contain the deaths of all kinds of people healthy and
unhealthy and as care will be taken not to insure those lives which are likely to
be soon extinct therefore in all probability fewer of the persons insured will die
in proportion to their number than those who are not insured, which will also
contribute to the gain of the corporation since the premiums are proportioned
to the Bills.16
The use in the pricing basis of the greatest deaths rates experienced in any
one year of a 22-year period may appear today as a somewhat ad hoc and noisy
way of setting a pricing margin. But he provided an explicit rationale for this
margin, based on a distinction between those policyholders who would bear
the risks and those who would not. In his conception of a mutual life assurer,
there would be two classes of policies: one that would underwrite the guaran-
tees and participate in the profits of the business; and another that would not
participate in the risks or profits of the business. Furthermore, at this time lim-
ited liability was only applicable by exception to companies that were granted
it by royal charter. So the participating policyholders would have unlimited
liability, and in the event that claims were due which could not be met by the
assets of the corporation, they would face a call on their personal assets.
Dodsons view was that a non-participating policyholder should be charged
more than a participating policyholder for the same sum assured, so that the
16
Dodson (1756), Chapter 1.
participating policyholder was offered a compensatory expected return for

the risks they were underwriting. His proposal was therefore that the partici-
pating business be priced using mean deaths whereas the non-participating
premium basis would use the greatest deaths table: if the persons who shall
desired to be insured without being liable to such a call should be rated in
proportion to the greatest number of deaths that have happened within our
knowledge as I think they ought, then the latter ought to pay near part in
the premium more than the former.17 For those aged 4050, Dodsons great-
est deaths table would result in a non-participating pricing basis that was
quite similar to the 5 % assumption used by the Amicable.
In his Lectures Dodson produced tabulations of multi-year projections of
accumulated assets, premium payments and claim outflows to illustrate the
workings of his regular premium whole-of-life policy. Figure2.2 charts one
of these tables. This is a projection of a non-participating product. The pro-
jection is made on the mean deaths basis but the premium basis used is the
greatest deaths table.
The regular premium rate for a 40 year-old was calculated to be 4.625 %.
The starting annual regular premium income for the cohort of 8,165 lives
180,000
Assets
160,000
Regular Premiums
140,000 Claims
120,000 Interest Income
100,000
80,000
60,000
40,000
20,000
0
0 5 10 15 20
Year
Fig. 2.2 Dodsons twenty-year projection of a regular premium whole-of-life

policy: Cohort of 8,165 40 year-olds, sum assured of 100

17
with a sum assured of 100 each is therefore 37,763. By the end of Dodsons
twenty-year projection, 4,647 of the 8,165 lives in the cohort have died, leav-
ing a total of 3,518 alive (and hence the total sum assured outstanding is
351,800), while assets of 152,392 have been accumulated. Not all of these
assets represent a surplussome of it is required as a reserve for the heavier
mortality experience that is expected in the later years of the cohorts lives.
Dodson calculated that, at the end of year twenty, the present value (on the
mean deaths basis) of the future claims of the remaining 3,518 lives was
265,917; and that the regular premium annuity stream (again on the mean
deaths basis) from the 3,518 remaining 60 year-olds had a present value of
134,412. Thus, a surplus of assets over liability reserve of 152,392(265,
917134,412) = 20,887 had, so far, been generated by the pricing margin
implied by using the greatest deaths table in pricing.
Interestingly (in the context of actuarial debates to come much later),
Dodson presented this prototypical liability reserving calculation by explic-
itly appealing to the logic of valuing the liability with reference to the cost of
transferring the liabilities to a third party:
Now let us suppose that the corporation will contract with some other body,
take the insuring of these off their hands by a payment of a sum in hand, being
allowed discounts for the money so advanced them by computing the sum for
which 100 may be insured for a life of 60 on the manner already planned.18
Having established the principle that non-participating business should

be priced to generate a surplus for participating policyholders, the question
arose of how to distribute this surplus. The equitable distribution of the sur-
plus amongst the participating policyholders is discussed in Dodsons Lectures.
There he argued that the profit should only be distributed to participating
policyholders who have already themselves generated a profit for the assurer
by paying a total amount of regular premium that exceeded their sum assured,
and then in proportion to that excess:
It seems, therefore, unreasonable to divide any part of the profit [from non-
participating business] among those who have not paid as much as they receive
because they have evidently a profit by being paid their claims which profit is
very great in first years it seems reasonable therefore to divide the profit
amongst the persons to have paid more than claims and to do this in proportion
to the sums so overpaid.19
18
19
He suggested this surplus be distributed to the participating policyholders

in the form of a cash dividend. He did not consider the possibility of distrib-
uting the surplus in the form of an increase in their sums assured.
Dodson argued that it would be unnecessary for the business to hold any
initial cash reserve as a contingency for adverse experience. This may seem a
peculiar position for a prototypical actuary who understands risk well enough
to advocate charging on the basis of the worst single year in 22 years of experi-
ence. But he argued that the presence of participating policyholders unlimited
liability meant it was superfluous: even if a funded reserve was established, it
would not absolve the participating policyholders from potential calls on their
personal assets, and the security of the non-participating business would still
at least partly rely on such calls being made good. It should also be noted that
it was typical practice of joint-stock companies of the time to only require a
small portion of equity capital to be paid-up, with the remainder again being
funded by future calls on shareholders when the company required the cash.
So, in the broader context of the historical corporate practice, this position
was reasonably natural.
Dodsons work was less concerned with technical innovations, such as those
developed by Halley and de Moivre over the preceding 60 years. He was moti-
vated by a vision of how such techniques could be applied to the provision of
a new form of life assurance that was really quite revolutionary in the context
of the life assurance practices of the time. The summary features of his system
as set out in First Lectures in Insurance were:
The use of mortality tables to produce age-specific premium rates for life
assurance. At the time, a flat non-age-specific premium basis was standard
market practice.
The development of a whole-of-life assurance policy, available in single and
regular premium forms. At the time, only short-term (usually one-year
term) assurance was sold.
A with-profit system where participating and non-participating policies
were simultaneously provided. Non-participating policies would be more
expensive and would be priced to make it unlikely (relative to experience)
that losses would be suffered. Participating policyholders would nonethe-
less have an unlimited liability exposure to calls on their personal assets in
the event of losses exceeding the assets in the fund.
The surplus on non-participating business would be distributed equitably
amongst participating policyholders, in proportion to how much profit their
own policies had generated for the corporation by virtue of their longevity.
Contingency capital would be funded by calls on shareholders (the long-

term policyholders in a mutual scheme) rather than by cash funding of a
contingency reserve.
Everything advocated by Dodson as described above was implemented by

Equitable Life within 25 years of the publication of Lectures. The consequent
success of the Equitable meant that it would come to be regarded as the model
of life assurance practice. But this revolution would occur after Dodsons death
and the significant contributions of some talented others would be required to
bring his vision to fruition.
The Beginnings oftheEquitable (17571772)
It has been frequently said that the history of the Equitable Office is the history
of life assurance in England. If this be not literally true, it certainly is true that
the readiness with which the conductors of this Office have ever listened to and
embraced all real improvements, not only led it to the high position which it so
speedily attained, but has tended to advance both the theory and practice of Life
Assurance in this Kingdom.20
As Lectures went to press in 1756, Dodson organised a series of City of

London meetings to secure the financial and political support he would
require to bring his conception to fruition. Dodson secured significant back-
ing from influential City and political figures to create a new life assurer. His
City backers whole-heartedly bought into Dodsons vision with one impor-
tant exception: Dodson had envisioned the life assurer being established as
a mutual partnership with unlimited liability, but his backers believed that
the venture would be more marketable as a corporation that provided limited
liability to its participating policyholders. The share capital would still be
largely unfunded and so participating policyholders would still be on call to
inject capital if required, but this would be subject to a maximum amount.
Their first objective was therefore to obtain a royal charter for incorporation.
They presented a petition of application to the Privy Council in 1757.
The three corporations writing life assurance in Britain at the time
(the Amicable, Royal Exchange and London Assurance) lobbied for the peti-
tion to be rejected. These firms argued that the size of the life assurance mar-
ket was so small that it could not support another corporation. They also
20
Walford (1868), Chapter 4.
argued that it would be unsafe for the new insurer to be established without
the backing of a capital fund. The petition was rejected by government in
1760, and in the intervening time since the application, Dodson had died. A
report was published by the Attorney General and Solicitor General in 1761
explaining the reasons for the rejection of the petition. A few quotes from the
report can give a sense of the nervousness about the revolutionary life assur-
ance design being proposed by the petitioners.21
On the funding of risk capital through capital calls rather than a paid-up
fund:
Because it appears to us altogether uncertain whether this project will or can

succeed in the manner in which it is proposed, and if the success is uncertain,
the fund for supporting it, which is to arise from the profits of the undertaking,
will be precarious. This last objection is, in our opinion, a fatal objection to the
scheme; for, though an undertaking plainly calculated for the benefit of the
public may, in some instances, deserve encouragement, even where success is
dubious, yet, in such cases, the projectors alone ought generally to abide the
peril of the miscarriage. In the present proposal, therefore, whatever else may be
hazardous, the capital or fund to answer the losses ought to be certain and liable to
no casualty.
On the use of mortality tables to set premiums by age:
The success of this scheme must depend upon the truth of certain calculations
taken upon tables of life and death, whereby the chance of mortality is attempted
to be reduced to a certain standard: this is a mere speculation, never yet tried in
practice, and consequently subject, like all other experiments, to various chances
in the execution.
On the use of population data to set life insurance premiums:
The register of life and death ought to be confined, if possible, for the sake of
exactness, to such persons only as are the objects of insurance. Whereas the
calculations offered embrace the chances of life in general, the healthy as well as
unhealthy parts thereof, which, together with the nature of such persons occu-
pations, are unknown numbers.
21
The following quotations from the report of the Attorney General and Solicitor General are obtained
from Ogborn (1962), Chapter II.
On the prospect of the business being profitable:
We are more apt to doubt the event, because it has been represented to us that
all profit that has been received by the Royal Exchange Assurance Office, from
the time of its commencement to the present time (40 years) amounts only to a
sum of 2,651, the difference between 10,915 paid in premiums and the sum
of 8,263 disturbed in losses If then, this corporation, who are charged with
taking big unreasonable premiums have reaped no greater profit, we can hardly
expect a more considerable capital to arise from lower premiums, and the hazard
of loss will be increased in proportion as the dealing will be more extensive.
And their final advice:
If the petitioners, then, are so sure of success, there is an easy method of making
the experiment, by entering into voluntary partnership, of which there are sev-
eral instances now subsisting in the business of insuring; and, if upon a trial
these calculations are found to stand the test of practical experiment, the peti-
tioners will then apply with a much better grace for a charter than they can at
present, whilst the scheme is built only upon speculative calculations.
The promoters immediately took the advice of the Attorney General and
Solicitor General and produced a draft deed of settlement that would create
a new life assurer as an unincorporated partnership of participating members
(just as Dodson had always advocated). The Society for Equitable Assurances
on Lives and Survivorships was thus established in September 1762.
For the next ten years, the Equitable developed as an established life assurer
in a small life assurance market. Like most businesses, it had its fair share of
corporate politics and power struggles, but it stayed largely true to Dodsons
vision. By 1772, it had accumulated assets of around 30,000, with an annual
premium income of 7,558 and total sums assured of 174,282 on 567 assur-
ance policies. The policies were a mix of participating and non-participating
contracts, and whole-of-life and term assurance policies. There were also 20
annuitants receiving a total of 694 per annum. The further development
of Dodsons vision and indeed the next developments in actuarial thought
required impetus from a new character.
Richard Price andHis Observations (1772)
Richard Price is known for two particular achievements that made perma-
nent contributions to the history of actuarial thought and beyond. As we saw
above, he presented Bayes revolutionary work on statistical inference to the
Royal Society. Secondly, Price produced the Northampton mortality table,

which was the market standard for life assurance pricing in Great Britain and
North America for nearly a centuryover which period the industry finally
and substantially blossomed.
His contributions to actuarial thought, however, are even wider than these
two singular achievements. From 1768 through to the 1780s, the Equitable
consulted him as an actuarial trouble-shooter whenever challenging technical
issues arose. When his advice was given, it appears to have been consistently
taken and acted upon. More generally, his actuarial writing reveals a deep
understanding of both theoretical and practical issues in pricing and reserving
for life assurance. His work provides perhaps the earliest examples of a rounded
actuarial analysisquantitative and technical, but only as a means to an end,
and with the clear objective of developing pragmatic solutions to real and com-
plex financial problems, implemented with judgement and an awareness of the
limitations and complications of business reality and human behaviour.
Like Thomas Bayes, Richard Price was a clergyman, a mathematician, a phi-
losopher and a Fellow of the Royal Society. Price was also actively involved
in political debate, arguing passionately in favour of freedom of thought and
political reform in areas such as the extension of the voting franchise, the
reduction of the national debt and in favour of the revolution in America.
Politically, he was a prototypical liberal Victorian. Intellectually, he was a classic
late Enlightenment figure: an intense man who was passionate about rational
thinking and reasoned debate, and who obtained the highest regard of influ-
ential men across a diversity of intellectual circles. Price died in 1791, aged 68.
The philosopher and historian Ian Hacking ranks Prices Northampton
tables highly amongst the historical mortality tables. They were perhaps the
first statistical results to be taken seriously The Institute of Actuaries did not
do anything much better until 1869.22 The first version of the Northampton
table was published by Price in 1772 (along with tables based on population
data for Norwich and London) in the second edition of his Observations on
Reversionary Payments.23 This book addressed a wide selection of topics relat-
ing to life contingencies (and a number of financial topics outside the field
of life contingencies). The construction of mortality tables is addressed in
Essay IV of the book, which, in keeping with the style of the times, had an
exhaustively descriptive title: On the Proper Method of constructing Tables
for determining the rate of human mortality, the number of inhabitants, and
the values of Lives in any Town or District, from the Bills of Mortality in
which are given the numbers dying annually at all ages.
Hacking (1975), p.113.

22
Price (1772).
23
Halleys Breslau table had remained the standard reference table from its
publication in the 1690s until the arrival of Prices tables. No practical use was
made of Halleys table by life assurers, but it was used by other writers such
as de Moivre in their research on life contingencies. Halleys paper did not
include an explicit description of his assumptions and methodology. Price was
more forthcoming and provided a detailed discussion of exactly what steps he
believed should be taken in transforming raw data on deaths into a finished
mortality table.
In the earlier discussion of Halleys tables, we noted that Price had specified
how mortality rates could be derived from ages at death when the underly-
ing population is at a stable level and there is no immigration or emigration
(recall that the technical challenge here arose from the lack of information
about the number alive at each age, and this therefore needed to be inferred
from the data on numbers of deaths). In his essay, Price showed how mortal-
ity rates could be derived when there is a stated level of net immigration or
emigration, setting down the following general rule:
From the sum of all that die annually, after any given age, subtract the number
of annual settlers after that age; and the remainder will be the number of the
living at the given age.24
He noted that the effects of immigration would be particularly signifi-

cant in London, and he analysed its impact on estimated mortality rates by
calculating rates from the London Bills of Mortality 17591768 data with
and without his immigration adjustment. Naturally, the modelled impact of
immigration must be a function of assumptions about the ages at which immi-
gration occurs. Price noted that deaths significantly exceeded reported births
in London, even though London was generally reckoned to have increased in
population over the period. The available birth data of the period was some-
what unreliable and he expected it would underestimate the true levels of
births, but he still conjectured that around one quarter of the annual deaths
were of immigrants to the city. He assumed that all immigrants would be aged
20 at the time they entered.
Based on these assumptions, the mortality rates for ages less than 20 would
be understated, as the no-immigration assumption would result in an over-
estimation of the numbers alive at ages up to 20 (recall that the estimation of
the numbers alive at age x is based on the total numbers annually dying aged
more than x). With the assumption that all immigration occurs at age 20
24
Price (1772), p.246.
and none occurs beyond that age, no adjustment for immigration would be
implied for mortality rates from age 20 onwards.
Figure 2.3 compares the no-adjustment and with-adjustment results
obtained by Price for ages up to 20.
The chart shows that mortality rates at very young ages had, until this time,
been significantly understated by calculations using the Bills of Mortality (such
as those produced by Dodson for use by the Equitable). This did not have
much direct consequence for the pricing of life contingencies as most policies
were written on lives older than twenty years of age. It did, however, have
implications for the assessment of life expectancy at birth and for government
health policies. Great Britain underwent an unprecedented period of popula-
tion growth in the second half of the eighteenth century. The population is
estimated to have grown from around 6.3 million to 9.2 million between
1751 and 1801, and most of the growth over this period is believed to have
occurred form 1780 onwards.25 A sharp fall in infant mortality rates is gener-
ally believed to have been a significant factor in this population growth. Prices
estimates certainly highlight how shockingly high infant mortality rates were
at the start of this periodaccording to his calculations, most of Londons
newborn babies of this time would not survive to see their fifth birthday.
0.35
0.3
No adjustment
0.25 With adjustment
Mortality Rate
0.2
0.15
0.1
0.05
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Age
Fig. 2.3 Prices mortality rates derived from London Bills of Mortality 17591768
data, with and without adjustment for immigration
Deane and Cole (1967), Chapter 1, Section 2.

25
Price then identified Northampton and Norwich as towns that had kept
Bills of Mortality which included ages at death for many decades. He applied
his methodology, including his adjustment for immigration, to both these
datasets to produce tables for each town. He also compared these results with
Halleys Breslau table. The mortality rates of these two tables, together with
Prices London table and Halleys Breslau table, are compared below in Fig.2.4.
The consistency of these tables is quite striking, and Price noted, with an
unbridled satisfaction, there is a striking conformity between all the three
Tables [Norwich, Northampton, Breslau], which gives them great weight and
authority. The London table produced noticeably higher mortality rates than
the other three tables and this was unsurprising: John Graunt had noted a cen-
tury earlier the relatively lower mortality rates associated with country living.
As we might expect from Price given his history of involvement with sta-
tistical inference, he gave some consideration to the potential sampling error
in his data. He did not attempt to calculate an explicit standard error for his
mortality rates (no such concept existed at this time). Instead, he used an
empirical approach noting that whilst 30 years of data were used in each of
the Northampton and Norwich tables, the results for each were very similar
when any ten-year period within the 30 years was used instead. Based on this
analysis, he concluded: These Tables, therefore, are founded on a sufficient
0.25
London (1759 - 1768)
Northampton (1741 - 1770)
0.2
Norwich (1740 - 1769)
Breslau (1687 - 1691)
Mortality Rate
0.15
0.1
0.05
0
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86
Age
Fig. 2.4 Prices mortality tables for London, Northampton and Norwich and
Halleys Breslau table
number of observations; and it appears, that there is an invariable law which

governs the waste of human life in these towns.26
Prices Observations is most renowned for these mortality tables. But there
was much else in the book that had an important intellectual impact on the
developing actuarial thinking of the time. Observations was based on Prices
experiences as an advisor to a number of embryonic or would-be insurance
companies that were considering how to price a variety of forms of annuity
contract. The book begins with an exhaustive discussion of his experiences
in this area, giving annuity pricing examples for a number of variations of
reversionary and deferred forms of annuity in single premium and regular
premium forms. In these examples, he computed exact valuations rather than
relying on the sort of approximations that de Moivre had developed. In the
intervening 40 years since de Moivres Annuities on Lives, methods of arith-
metic computation had become somewhat more efficient, making the exact
calculations more feasible, if still highly laborious. Price included an essay
on de Moivres annuity pricing approximations that warned of the limited
accuracy of his joint life pricing formula. In particular, Price showed how
this could lead to substantial proportional errors when valuing a reversionary
annuity (which is valued as the difference between the single life annuity and
the joint life annuity).
Observations also included discussion of a number of annuity products that
had been launched by life and annuity businesses since the 1750s. In these
discussions, he repeatedly found that the products were substantially under-
priced and concluded it unlikely that policyholders would receive their full
promised benefits. This treatment was applied to the reversionary annuity
products of the London Annuity Society and the Laudable Society, and more
widely: There are in this kingdom several institutions for the benefit of wid-
ows, besides the two on which I have now remarked; and in general, as far
as I have had any information concerning them, they are founded on plans
equally inadequate.
The actuarial management of the life assurance companies was considered
too. Whilst the Royal Exchange and London Assurance were long-established
writers of life assurance with guaranteed sums assured, Price focused on the
Amicable, which wrote significantly higher volumes of business, though, as
was discussed earlier, without offering guaranteed sums assured. Price strongly
criticised this mutual benefaction model of life assurance that Amicable pro-
vided: It is obvious that regulating the dividends among the nominees by
the number of members who die every year is not equitable; because it makes
Price (1772), pp.257258.

26
the benefit which a member is to receive to depend, not on the value of his
contribution, but on a contingency; that is, the number of members that shall
happen to die in the same year with him.27 He also noted that charging all
members the same premiums irrespective of their age was not equitable.
He went on to contrast these imperfections with the Society for Equitable
Assurances on Lives and Survivorships (of which he was a paid consultant)
and provided some public advice to its managers:
First, on the need for rigorous underwriting to avoid anti-selection: I have

already more than once observed that those persons will be most for flying
to these establishments, who have feeble constitutions, or are subject to
distempers, which they know render their lives particularly precarious; and
it is to be feared that no caution will be sufficient to prevent all danger from
hence.28
Then on adverse variation in experience: The calculations only determine
probabilities; and agreeably to these, it may be depended on that events
will happen on the whole. But at particular periods, and in particular
instances, great deviations will often happen; and these deviations, at the
commencement of a scheme, must prove either very favourable, or very
unfavourable.29
And finally, with impressive prescience given the tumultuous debates that
would embroil the Equitable in the coming decades, on the prudent distri-
bution of surplus: the encouragement arising from the possession of a large
surplus (should) be led to check or stop the increase of its stock by enlarg-
ing its dividends too soon, (otherwise) the consequences might prove
pernicious.30
These pieces of advice highlight the completeness of Price as an actuar-

ial thinker. He strongly advocated a reasoned approach to pricing based on
the rigorous analysis of statistical data, and he furnished the profession with
much by way of methods and results of such analysis. But he also understood
the practical aspects that must attend the successful and sustainable actuarial
management of a life assurance business, and he anticipated some that had
not yet arisen at the time of his writing. He used both these technical and
practical insights to actively direct the development of new actuarial tech-
27
Price (1772), p.124.
28
Price (1772), p.129.
29
Price (1772), p.129.
30
Price, Observations, p.131.
niques at the Equitable, and these would be a critical part of that businesss
huge success in the following decades. Added to these contributions is his
vital role as the midwife of Bayes revolutionary work on statistical inference.
All this taken together, Richard Price stands as a titan of actuarial thought. If
an actuary is someone who provides rational and rigorous advice on the sus-
tainable long-term financial management of life contingencies business, then
Price was arguably the first actuary, and inarguably one of the most important
and influential actuaries in the professions history.
illiam Morgan andtheActuarial Management ofwith-

W
Profit Business (177582)
Several developments important to the nascent life assurance industry

occurred over the dozen years following the Equitables founding in 1762.
Prices Observations, with its new mortality tables and sound actuarial
insights, brought actuarial thinking on the pricing of life contingencies to a
new level of competence. The Equitable had established itself and its whole-
of-life assurance product as successful features of the British life assurance
industry. It had become the leading provider of life assurance, but the size of
this market remained modest relative to other forms of insurance business.
The Life Insurance Act of 1774 further improved the conditions for future
growth in the industry. This act introduced the legal concept of insurable
interest. Henceforth, life assurance could not be used as a means of specula-
tive gambling, but only as a form of risk mitigation where the buyer had a real
and legitimate exposure. This focused life assurance business on products that
were generally most amenable to actuarial and statistical analysis. Thus the
scene was set for the next step in the development of life assurance business
and the actuarial practices required to make it sustainable.
William Morgan, nephew of Richard Price, and the actuary of the Equitable
for a full 55 years from 1775 to 1830, played a most significant role in shaping
this next stage of actuarial practice. Morgan was appointed assistant actuary
in 1774, serving under John Pocock, who was elected actuary a month earlier
following the death of John Edwards, the previous holder of this post. John
Pocock had been in office less than a year when he died unexpectedly in early
1775. Morgan had distinguished himself sufficiently well during his short
term as assistant actuary for him to be elected actuary in an uncontested elec-
tion at a general court of the society in 1775. Morgan did not attend univer-
sity but did receive some medical training as an apprentice at an apothecary
prior to joining the Equitable. He was tutored in life assurance mathematics
by Price in 1773 before being appointed assistant actuary. At this time, the
actuarys role in a life assurance business was not yet well defined. Edwards
and Pocock had been book-keepers rather than statisticians. It would not have
been viewed as essential for Morgan to demonstrate advanced mathematical
learning, and the tutelage under Price would likely have carried some weight
with the directors of the business.
At the start of Morgans tenure as actuary, there was an emerging inter-
est in the distribution of the capital that was evidently accumulating in the
business. A satisfactory way of assessing the appropriate level of reserves for
in-force business had yet to be fully implemented (actuarial methodology
and the computational burden of policy-by-policy valuation both had to be
addressed). A rigorous and accepted means of determining how much of the
visible asset accumulation could be considered as surplus available for distri-
bution was therefore wanting. With Morgan still establishing himself in his
new role, the directors looked to Price for guidance and he duly provided
them with a number of recommendations. Price suggested the society should
pursue three approaches to analysing the current adequacy of the accumulated
capital fund: a comparison of the actual mortality experience to date with
the mortality assumptions used in the premium basis; a comparison of the
amount of premiums being received with the claims being paid (he s uggested,
as a rule of thumb, that around two thirds of premiums received in the early
years of a whole-of-life policy would be required to adequately build up the
reserve for the claims that would arise in its latter years); and a comparison of
the accumulated assets of the society with a valuation of the liabilities.
In anticipation of a surplus being identified by the above methods, Price
also provided some advice on the form of distribution of surplus to the par-
ticipating policyholders. In particular, he advised against the payment of any
cash dividends. Instead, he argued, if any distribution of surplus was deemed
appropriate, it should be made by either increasing the sums assured or reduc-
ing the future regular premiums of the participating policies. Prices advice
was acted upon by Morgan and, like virtually everything else Price did in the
field of actuarial science, his recommendations became part of standard actu-
arial practice for decades (or, as in this case, centuries).
The market value of the assets of the society at the end of 1773 amounted
to 33,000. Price surmised that a valuation of the liabilities might be around
22,000, implying a surplus of 11,000. It was Morgans job to perform the
first rigorous valuation of the societys liabilities and hence determine a reli-
able measure of the surplus. He and his small team of assistants valued each
of the 922 policies in-force individually. Four months and several thousand
manual calculations later the full results were ready to report. The results were
even better than anticipated by Price. At the start of 1776, the assets were val-
ued at 41,928 (their market value) and the liabilities were valued (using the
premium basis) at only 16,786, implying a surplus of 25,142. This startling
result raised two fundamental questions: how did such a large surplus arise?
And what should be done with it?
The surplus primarily arose from the much lighter mortality experienced
to date than that assumed in the premium basis. This was especially true in
the non-profit business, where, as discussed earlier, the premium basis was
more expensive than the premium basis applied to with-profit policies. But
even the with-profit experience was very substantially better than assumed in
its premium basis. Morgan himself noted that the 1776 investigations had
concluded the probabilities of life in the Society had been higher than those
in Mr Dodsons Table, from which its premiums were computed, in the pro-
portion of three to two.31
This substantial mortality margin had several contributory factors. The pre-
mium basis used Dodsons recommended London Bills of Mortality basis.
This was based on mortality data from 50 years prior at a time when life
expectancies were improving significantly due to medical developments and
improvements in economic and social conditions. The period on which the
bills were based included 1740, a year of weather-induced food shortages that
resulted in mortality almost as severe had been experienced in bouts of the
plague. The bills were based on the experience of London, which was probably
the least healthy place in the country; and they were based on the full breadth
of the London population, not the emerging middle class that was buying the
Equitables whole-of-life assurance policies. Furthermore, the business had an
active approach to underwriting, with every new life assured being reviewed
and approved by a director.
There were other factors beyond the mortality basis that also contributed
significantly to this accumulation of surplus. A number of somewhat arbitrary
additional loadings were applied on top of the premiums implied by Dodsons
tables. These included loadings for larger assurances, young lives and females
(sex-specific mortality tables were not in use at this time). And then a final
6 % margin was additionally applied to all premiums. Furthermore, wars in
Europe and America had resulted in significant increases in government bond
yields, and much new money was therefore invested at yields well in excess of
the 3 % assumed in the premium basis.
These profits were amplified by the gearing effect whereby the surplus gen-
erated by non-participating short-term business was left to be shared amongst
the remaining participating policyholders. A decade later, in an analysis of the
Morgan (1829), p.20.

31
surplus that had emerged between 1768 and 1787, Morgan concluded that
more than half was due to profits made on temporary (and non-participating)
assurances.
Finally, there was another source of surplus that was important during this
period. Of the whole-of-life assurances written during the societys first twenty
years, less than half ultimately resulted in claims. The vast majority of those
policies that did not result in claims were forfeited without any compensation.
The society did have a policy of offering fair surrender values, but it may have
been that many policyholders never actually applied to surrender their policy
but merely forfeited them when they chose not to make any further premium
payments. Furthermore, the surrender basis used by the Equitable was based
on a form of net premium valuation rather than an asset share-based calcula-
tion. This prospective approach to assessing the surrender value was highly
sensitive to the mortality basis used in the valuation. When the Equitable
moved its premium basis to the Northampton table in 1780 (discussed further
below), it meant that many of the regular premium whole-of-life contracts
written on the old premium basis were worth very little on this prospective
basis, or even had a negative value to the policyholder!32 So the surrender
values were much lower than the accumulated value of the assets that the
surrendered policies had contributed (even after deducting the cost of the life
assurance that had been provided over the duration of the policy to date).
What to do with the surplus? A series of meetings of the general court of
the society in 1777 determined that a 10 % reduction in all future premiums
should be effected. Moreover, the in-force with-profit policyholders would
have this reduction back-dated: it would be calculated for all the premiums
they had paid to date, and this would be offset against their next regular
premium. To the extent that the refund was greater than the next regular
premium, the remainder would be paid to the policyholder in cash. So the
back-dated element was essentially a one-off cash dividend.
The premium reduction and refund of 1777 was a temporary stop-gap
measure that alleviated the building pressure from participating policyhold-
ers. But a more fundamental review of pricing and surplus distribution was
called for. As usual, the intellectual impetus was provided by Richard Price.
In 1780, he advocated a change in the underlying mortality tables that were
used in the premium basis. Dodsons London Bills of Mortality premium
basis was still in use, and this was clearly generating premiums that were much
larger than was reasonable. Indeed, it is interesting that the business could
continue to generate significant new business at these rates. This perhaps
32
Morgan (1829), p.38.
reflects the lack of (credit-worthy) competition in the life assurance market

of the time. It may also suggest that policyholders increasingly viewed the
whole-of-life with-profit policy not merely as an insurance product, but as an
investment product (with likely future participation somehow in the surplus).
Price recommended using a mortality basis derived from the Northampton or
Norwich tables that he published in Observations. The directors, as usual, fol-
lowed this recommendation and decided upon the Northampton table. Price
updated this table to include Northampton Bills of Mortality data up to 1780
(so the table was based on data from 1735 to 1780). The directors opted to
add a 15 % loading to the premiums implied by this table. Even Price, whose
advice to the society tended to be universally prudent and frugal, referred to
this loading as exorbitant.
This new premium basis would be applied to new business. It did not
address if and how to distribute the surplus that had accrued to date to exist-
ing policyholders. How much of the surplus should be distributed to the
existing with-profit policyholders, and what form should this distribution of
surplus take? The directors recommended that an addition should be made to
with-profit policyholders sums assured of 1.5 % of the sum assured for each
annual regular premium paid (hence a policy that had been in-force for the
full 19 years since the founding of the society would obtain a 28.5 % pro-
portional increase in their sum assured). The proposal was approved by the
members, and the first with-profit reversionary bonus was thus paid in 1781.
No explicit historical explanation is available, but it is likely that the size of
the reversionary bonus was chosen to distribute a targeted portion of the sur-
plus. What drove Prices preference for distributing surplus through increases
in sums assured rather than cash dividends? It was perhaps because retaining
the assets within the society for as long as possible fitted with his vision of a
long-term growing business of increasing influence and scale. In Ogborns
outstanding history of the Equitable,33 he suggested that the increase in gov-
ernment bond yields that occurred in the period of 17701780 may also
have played a part. The American War of Independence had placed consid-
erable strain on the British Governments finances, and long-term govern-
ment bond yields had increased from around 3 % in 1770 to around 6 % by
1780.34 Most of the assets of the society were invested in long-term govern-
ment bonds, and there may have been a reluctance to be forced to sell at these
deflated price levels to fund a cash distribution (the assets, however, were val-
ued at these market values in Morgans surplus analysis). The society had also
Ogborn (1962), p.110.

33
Homer and Sylla (1996), pp.156157.

34
recently started to invest a significant portion of its assets in mortgages. The

illiquidity of these assets might also have tempered the directors enthusiasm
to distribute significant amounts of cash (particularly if it created the expecta-
tion that more might be follow).
A further reversionary bonus of the same form was made in 1792, this
time of 2 % of the sum assured for each regular premium paid. This same
level of bonus was added again in 1800 when the surplus was calculated to be
484,000 (and the bonus was calculated to reduce this surplus to 225,000).
The bonus system established by Morgan proved highly attractive to prospec-
tive policyholders. Furthermore, it was not a system that could be quickly
replicated by competitors.
Despite spending much of the period at war, between 1781 and 1800 the
British economy grew rapidly as the early stages of the industrial revolution
gathered momentum. Increases in overseas trade and substantial popula-
tion growth fuelled impressive economic growth.35 For the emerging sala-
ried middle class, the whole-of-life with-profit policy was a highly compelling
composite of a short-term life assurance and a long-term savings vehicle. The
Equitable was perfectly positioned to take advantage of this demand. The
business of the Equitable exploded in scale over this period: between 1776
and 1800, the assets of the Equitable grew from 42,000 to over 1 million.36
The greatest challenges facing Morgan in the following decades were related
to managing the consequences of this remarkable success. In particular, he
was occupied with determining how to limit the number of new policyhold-
ers being admitted to the society so as to avoid carpet-bagging; ensuring an
equitable distribution of surplus between the old and new members; and,
more generally, managing the expectations of policyholders as to the amount
of surplus that it was reasonable to distribute over time. In 1800, the soci-
ety introduced two safeguards to over-distribution of surplus: bonuses would
only be distributed once every ten years; and the present value of the size of
the bonus should not exceed two thirds of the surplus assessed at that point in
time. A further, rather more controversial safeguard was put in place in 1816:
bonuses would only ever be applied to the oldest 5,000 policies of the society
at the date of the bonus distribution. At the time of its implementation there
were around 9,500 policies in-force.37 This created a waiting period for new
policyholders before they were entitled to participate in any bonus distribu-
tion. Morgan expected at the time of the rules implementation that, based
35
Deane and Cole (1967), Chapters I and II.
36
Ogborn (1962), pp.104 and 123.
37
Ogborn (1962), p.155.
on historical discontinuance rates, this waiting period would be around five

years. Whilst controversial at the time of its inception, this rule remained in
place until 1893.
Thus, by the start of the nineteenth century, much of the actuarial ethos of
with-profit management had fully emerged: Dodson, Price and Morgan had
envisioned, analysed and implemented a financial productthe whole-of-life
with-profit policythat had radically reshaped life assurance as a product,
business and important element of the financial sector. With William Morgan
as the first role model, the actuary had emerged as a person of considerable
power in a life assurance company: his considered judgement and detachment
from short-term commercial pressure was required to determine equitable and
prudent distributions of the surplus of a with-profit fund. Morgan was almost
a caricature of this role: in later years, he would face much criticism for his
near pathological prudence, resisting pressure from managers and members to
reduce premiums or increase surplus distributions. Indeed, he felt compelled
to defend his actions in a book that he published a year before retiring from
his position as actuary in 1830, explaining in its Introduction:
It is indeed much against my inclination that I have engaged in the present

work; but my anxious desire to justify the conduct of a Society for whose welfare
I have laboured so many years, has impelled me to notice some of the misrepre-
sentations with which it has been assailed, and, as the best refutation of them,
to give a short history of the Society.38
Morgans time was many years before the introduction of financial regula-
tors and statutory capital standards. There was perhaps less need for them
when men like him were in the chair.
Morgan (1829), Introduction.

38
3
Life fromtheNapoleonic Wars
totheSecond World War
When the romantic formative period of a science is over, it gradually settles

down into one of ordered progress, which while supremely useful has less about
it to attract the imagination.1
So said Samuel George Warner in his 1917 presidential address to the

Institute of Actuaries. And yet, with the benefit of 200 years of hindsight, this
statement might be more applicable to the beginning of the nineteenth cen-
tury. In Warners era, actuarial thought and practice was largely reactive in the
context of a series of unprecedented events and circumstances: global war, high
inflation, interminable economic depression, low interest rates, more global
war, government-induced ultra-low interest rates. At the start of the nineteenth
century, life assurance practices still had to navigate the political and economic
challenges of their time, but the actuaries of the nascent life assurance indus-
try were arguably freer to plot the course of their profession and its practices.
The trail had been blazed in the final half of the eighteenth century and a
revolutionary vision of life assurance as a long-term savings vehicle for the
middle classes had been realised with the remarkable success of the Equitable
in the final two decades of the century. Many new life offices would follow
in Equitables wake. The pioneering work of Dodson, Price and Morgan now
had to be systematically developed into robust best practices for an embryonic
profession that was invested with substantial powers and discretion in a fast-
growing and increasingly significant element of the financial sector.
Warner (1918), p.16.

1

DOI10.1007/978-3-319-33183-6_3
The modelling of mortality and its implications for the pricing of life con-
tingencies continued to dominate actuarial thought in the first half of the
nineteenth century. Whilst important, even fundamental, developments in
mortality modelling continued through to the end of the nineteenth century,
a sense emerged by the middle of the century that mortality was increasingly
well-understood. Likewise, it was recognised that a greater amount of profes-
sional energy should be invested in the further development of thought in other
important aspects of actuarial responsibility within a life assurer such as reserv-
ing methods, surplus distribution approaches and the assurers investment pol-
icy. The forming of the Institute of Actuaries in 1848 (and Faculty of Actuaries
in 1856), and the arrival of their professional journals and sessional meetings,
was a catalyst for more collaborative and co-ordinated thought-leadership
amongst actuaries. It is perhaps no coincidence that the 25 years following
the formation of the institute was a period that saw the crystallisation of many
important ideas and principles that acquired permanence in actuarial thought.
The technical considerations involved in the valuation of liabilities, and
hence surplus, emerged as a major field of actuarial research in the mid-
nineteenth century, together with the related topic of determining how the
surplus should be distributed amongst the with-profit policyholders in a way
that was just and equitable. To a lesser degree, the second half of the eigh-
teenth century also saw actuaries give greater consideration to the asset side of
the balance sheet, and the first actuarial papers on investment strategy for life
assurance liabilities were produced. But just a handful of investment papers
emerged in the nineteenth century, and this field was only to become an area
of real focus for the profession in the twentieth century.
The economic turmoil of the first half of the twentieth century and its
implications for the management of the British life offices kept valuation,
surplus and investment strategy at the top of the thought-leadership agenda.
New mortality tables continued to be developed, but the first half of the
twentieth century did not witness any notable fundamental breakthroughs in
thinking on the construction of the tables.
The historical developments in these three broad areas of actuarial thought
mortality modelling, valuation and surplus, and investment strategy for life
assurance liabilitiesover the nineteenth and first half of the twentieth cen-
tury are discussed in turn below.
Developments inMortality Modelling, 18081881

Chapter 2 noted that Richard Prices Northampton table has been described
as a mortality benchmark that stood for the next 100 years and that according
to Hacking the Institute of Actuaries did not do anything much better until
3 Life fromtheNapoleonic Wars totheSecond World War 73
1869. The truth is more nuanced. The next significant development in mortal-
ity modelling to follow Price was made by Joshua Milne in 1815. Milne was
then actuary at the Sun Life Assurance Society, one of the many new life assur-
ance offices established in the early 1800s following the success of the Equitable.
Milnes two-volume tome A Treatise on the Valuation of Annuities and
Assurances on Lives and Survivorships largely consisted of mathematical book-
work on the valuation of various forms of annuities and assurances.2 But the
inclusion of his Carlisle mortality table made it a historic actuarial publica-
tion. The Carlisle table was, like Prices Northampton table, based on observa-
tions from local Bills of Mortality data (the Carlisle table was based on data
from 1779 to 1787 inclusive). However, Milnes Carlisle data was superior
to the data used in any prior mortality table in one key respect: it included
population data categorised by year of age (it was also categorised by sex,
and whilst Milne did analyse and discuss some of the differences in mortality
behaviour observed for the two sexes, he did not publish separate male and
female mortality rates in his Carlisle table). The Exposed to Risk element
of the mortality calculation therefore did not need to be estimated with the
use of assumptions about population stability and immigration/emigration
patterns. It could now be directly observed. This naturally supported a more
refined and accurate estimation of experienced mortality rates.
The Carlisle data had another advantage over Northampton: it was some-
what less out of date. The Northampton data was based on mortality data
from 1741 to 1770, whereas the Carlisle table used data from 1779 to 1787.
Life expectancy had improved at a significant rate over the second half of the
eighteenth century, particularly for infants. Milne noted that, even over his
observation period, infant mortality rates were rapidly changing, largely due
to the impact of social health policy initiatives such as smallpox inoculation.
The Carlisle data included some data on cause of death, and Milne noted that
the number of deaths from smallpox in the Carlisle Bills was 90in 1779 and
only 141 for the remaining eight years from 1780 to 1787. Dr Heysham, a
local doctor who prepared some of the Carlisle data noted that in 1779 that
several hundreds were inoculated in the neighbourhood of Carlisle, and it is a
pleasing truth that not one of them died.3 Almost all of these smallpox deaths
would have occurred at age five or under. In the Carlisle data, only 712 deaths
of age five and under occurred over the full nine-year data period. If the small-
pox death rate had continued at its 1779 rate, more than 700 would have died
from smallpox alone. This reduction in smallpox mortality therefore had a
substantial impact on the overall infant mortality rates of the period.
Figure3.1 compares the Carlisle and Northampton tables.
Milne (1815).
2
Milne (1815), p.735.

3
0.3
0.25 Carlisle (1779-1787)
Northampton (1741 - 1770)

0.2
Mortality Rate
0.15
0.1
0.05
0
0 10 20 30 40 50 60 70 80
Age
Fig. 3.1 The Carlisle and Northampton tables
The Carlisle table produced lower mortality rates than the Northampton
table for all ages from birth to aged 80. This reduction in estimated mortality
is attributable to two effects: the later data period of the Carlisle table and
the improvements in life expectancy that occurred over the second half of
the eighteenth century; and Prices assumption of a stable population from
Northampton in estimating Exposed to Risk, when in fact its population was
rising (as noted above, Milne did not require any such assumption for Carlisle
due to the availability of population data).
The Carlisle table was undoubtedly an improvement on Prices Northampton
table, and it was widely recognised as such in the actuarial and academic cir-
cles of the period. For example, Professor Augustus De Morgan, an academic
mathematician and influential writer on life contingencies (and the great
grandson of James Dodson!), wrote of the Carlisle table in 1838 they are to be
considered the best existing tables of healthy life which have been constructed
in England.4 Yet life offices never applied it to the pricing of life contingen-
cies to the degree that they used Prices table. This was at least partly because,
as we shall see later, life offices increasingly focused on the use of mortality
data directly observed on assured lives rather than the general population.
Meanwhile, at the Equitable, an aging William Morgan, increasingly set in
his ways, would never abandon his uncles Northampton table. As we shall see
De Morgan (1838), p.167.

4
in the next section, Morgans intransigence was set to become an increasing

source of contention and controversy within actuarial circles and beyond.
J ohn Finlaison andtheBritish Governments Life Annuity

Issuance (18081829)
The aftermath of the British Governments life annuity issuance of 1808 pro-
vides a colourful and historically important example of how the eighteenth-
century foundations of mortality thinking were built upon and extended by
some new and highly able actuarial thinkers in ways that prompted some dis-
agreement with the earlier actuarial generation. This particular example of the
changing of the guard of actuarial thought leadership involved some public
controversy, prompted Cabinet-level political reaction and was accompanied
by increasing personal animosity.
Some 120 years earlier, the British Government had experimented with the
use of life annuities as a form of debt funding for its European wars. It found,
somewhat peculiarly, that the scheme was both quite unpopular with investors
and yet highly unprofitable for the government issuer, relative to simply issu-
ing fixed long-term bonds. The government subsequently tried on occasion to
use life annuities to raise debt funds over the course of the eighteenth century,
but the issuances were very small in volume. For example, a life annuity was
issued in 1779 with the aim of raising up to 7 million, but only 133 annui-
ties were sold raising a sum of only around 150,000.5 In 1808, in the midst
of the Napoleonic wars and with Britain under serious threat of invasion, the
government tried again with renewed vigour. The profitability to the issuer
was similarly dismal to the 1692 issuance, but this time the market was better
prepared to spot a bargain.
William Morgan, as the actuarial leader of the pre-eminent and dominant
life assurance company of the age, advised the government on the mortality
basis to be used for the pricing of these life annuities. His advice was to do
what he did: use the Northampton table. The exact form of the advice that
Morgan provided is not available, but with the benefit of hindsight his recom-
mendations might appear surprising, or indeed rather odd. Morgan was well
aware at this time that the Equitables mortality experience was significantly
lighter than that implied by the Northampton table. He would have known
that if the annuity purchasers had experienced mortality rates similar to those
experienced at the Equitable, life annuities priced with the Northampton table
Leeson (1968), p.9.

5
would prove highly profitable for the annuitant investor and unprofitable for
the issuer. However, Morgans view was steadfast and consistent throughout
his entire career: he believed that the Northampton table was a fundamentally
accurate description of the law of human mortality; and the lighter mortal-
ity experienced by the Equitable on its assured lives was due to the selection
effects of its rigorous life assurance underwriting practices.
Even if Morgans belief was in fact correct, it ignored the self-selection effect
that is inevitably to be observed in annuity experience: only the healthiest of
lives would be nominated for annuity contracts, and this selection effect was
likely to be even stronger than the impact of the underwriting safeguards that
Equitable applied to its assured lives. The potential for annuity selection was
exacerbated by the ability of the annuity purchaser to nominate the life on
which the annuity is writtenannuities were not considered as insurance in
the context of the 1775 act that created the concept of insurable interest. In
Morgans defence, however, it should be noted that the Equitable had been
open for annuity business for many years and used the Northampton table as
the premium basis throughout. This had hardly stimulated an overwhelming
public demand for annuities (they consistently sold in tiny volumes compared
to assurance business). Indeed, the Equitable had managed to sell such small
volumes of annuities on the Northampton basis that there does not appear to
be any record of an analysis of its profitability.
And so, in 1808, the government offered life annuities at prices based on
the Northampton table, and this offer remained open for many years to fol-
low. Ten years later, several million pounds had been raised by the annu-
ity scheme, requiring annual annuity payments of 640,000.6 Around this
time, a civil servant at the Admiralty named John Finlaison wrote a letter to
the Chancellor of the Exchequer expressing his concern at the losses that he
believed were arising from the overly generous annuity pricing basis. Finlaison
estimated that losses of 8,000 per month were being incurred as a result.
Finlaison would, in 1848, become the first President of the Institute of
Actuaries. For an actuary of the period, he had an unconventional career and
background. He never worked in a life office, and instead spent his entire
career as a civil servant. He first made his mark in 1805 as the second clerk of
the Commission of Revising and Digesting the Civil Affairs of the Navy. As
noted in a recent history of the role of Britains civil service in the Napoleonic
Wars: Finlaison, born in Caithness, the son of a fisherman, had come to
London after an education in Edinburgh; at this time only twenty-three, but
already demonstrating a formidable logical brain and an immense capacity for
Francis (1853), p.203.

6
work.7 He worked at the Admiralty from 1809 to 1822, where he joined the
newly established Admiralty Record Office and was responsible for a much-
improved parliamentary reporting of the naval accounts and expenditure. He
then moved to the Treasury and was appointed the Actuary of the National
Debt, a position he held for the next 29 years.
No action was taken by the chancellor in 1819 when Finlaison first raised
his concerns about the pricing of the governments life annuities. Empowered
by his appointment as Actuary of the National Debt, he was eventually asked
by the Treasury to provide a full analysis and he delivered a parliamentary
report on 31 March 1829.8 Finlaisons report contained a comprehensive
analysis of the mortality experience of the annuities and tontines written by
British (or English) governments since 1693. His work was ground-breaking
in many ways. Most pertinently, it showed that the view he had expressed
to the chancellor ten years earlier was right: the mortality experience of the
government life annuities was much lighter than the Northampton table that
had been used in pricing them for the previous 20 years. This is illustrated by
Fig.3.2, which also shows Milnes Carlisle table for comparison.
0.14
Government Annuities Experience (1808-1826)
0.12
Price's Northampton Table (1741-1770)
0.1
Milne's Carlisle Table (1779-1787)
Mortality Rate
0.08
0.06
0.04
0.02
0
40 45 50 55 60 65 70 75 80
Age
Fig. 3.2 The government life annuity mortality experience and Northampton
and Carlisle tables
Knight (2013), p.325.

7
Finlaison (1829).
8
Finlaisons analysis evidenced the profound differences between the gov-

ernment annuity mortality experience and the mortality rates assumed in
the Northampton table: for example, at age 60, the Northampton table was
implying a mortality rate that was fully double that experienced by the gov-
ernment annuitants! The Carlisle table provided a significantly closer guide,
but still overestimated the mortality rate experience (which could be explained
by annuity anti-selection effects and by life expectancy improvements in the
decades between the two sets of observations). Finlaisons analysis implied
that the government annuity prices should be increased by around one quar-
ter. His report had an immediate impact: an Act of Parliament was passed in
1830 which upwardly revised the government life annuity prices in accor-
dance with his recommendations.
Even under the revised annuity pricing basis, scope for anti-selection was
still significant, especially at older ages: a 90-year-old life could obtain an
annuity rate of 62 % (payable in quarterly instalments). Francis 1853 book
provides amusing tales of the creative anti-selection speculation that this
generated:9
The shrewd gentlemen of the Stock Exchange immediately saw and seized the
advantage. Agents were employed to seek out in Scotland and elsewhere robust
men of ninety years of age, to select none but those who were free from the hard
labour which tells on advanced life Wherever a person was found at the age
of ninety, touched gently by the hand of time, he was sure to be discovered by
the agents of the money market The inhabitants of the rural districts of
Scotland, of Westmoreland, and of Cumberland, were surprised by the sudden
and extraordinary attention paid to many of their aged members. If they were
sick, the surgeon attended them at the cost of some good genius; and if they
were poor, the comforts of life were granted them. In one village the clergyman
was empowered to supply the wants of three old hale fishermen during the win-
ter season, to the envy of his sick and ailing parishioners. In another, all the
cottagers were rendered jealous by the incessant watchful attention paid to a
nonagenarian by the magnate of the place. It was whispered by the less favoured
that he had been given a home near the great house; that the cook had orders to
supply him with whatever was nice and nourishing; that the laird had been
heard to say he took a great interest in his life, and that he even allowed the doc-
tor twenty-five golden guineas a year, so long as he kept his ancient patient alive.
Clearly there was still much to learn about annuity market behaviour!
Nonetheless, the new pricing basis undoubtedly brought the cost of ser-
vicing new government life annuities closer to that of long-dated gilts.
Francis (1853), Chapter XII.

9
William Morgan resented and resisted the implication that his advice had
unnecessarily cost the government huge sums of money, commenting:
By the assistance of these tables (Finlaisons tables of the government annuity

mortality experience), the notable discovery has been made of the loss sustained
by the public of many thousands every week With the least knowledge of the
subject, it might easily have been seen that the use of the Northampton table
could not possibly be attended by such a loss, and that the extent of any loss (if it
really existed), instead of six years, did not require as many hours to ascertain it.
Finlaison and Morgan also adopted opposing views on other actuarial

debates of the time. Whilst Morgan argued that selection was the primary fac-
tor explaining the discrepancy between the Equitables experienced mortality
and the Northampton table, Finlaisons position was that there is very little if
any advantage at all in favour of selection.10 Whereas Morgan argued for the
use of statistical rates of sickness in friendly societies, Finlaison argued that
life and death are subject to a known law of nature, but that sickness is not,
so that the occurrence of the one event may be foreseen and ascertained, but
not so the other.11 On these other points, history proved Morgan the more
accurate. The following section will discuss how selection effects were mea-
sured by actuaries using techniques developed over the nineteenth century
that showed them to be highly material. The estimation of rates of sickness
was also developed into an increasingly reliable science by actuaries over the
course of the nineteenth century, culminating in the Manchester Unity sick-
ness tables that were published in 1903 and remained in use through much of
the twentieth century.12
Beyond the impact it had on government policy, Finlaisons report was
an important milestone for actuarial thought. It was noted in Chap. 1 that
a government annuity mortality experience analysis was first performed by
the Dutchman Johannes Hudde in the 1660s. Finlaison noted that Kersse
(again in Holland) and Deparcieux (in France) had performed analysis of
government annuity and tontine experience in the 1740s. But the scale and
rigour of Finlaisons analysis was of a different order to these earlier works. His
study also provided the eras most comprehensive analysis of how British mor-
tality behaviour varied by sex. The chronological span of Finlaisons analysis
provided insight into how mortality rates had decreased over the eighteenth
century. It was commonly understood that life expectancy had improved
10
Finlaison (1829), p.19.
11
Parliamentary Papers (1825).
12
Watson (1903).
s ubstantially over this period, but Finlaison was able to provide greater detail
of how mortality rates had changed differently across ages. He also noted that
this had implications for how historical mortality observations should be used
in constructing forward-looking estimates:
If the diminution (in mortality rates) had been doubtful, or very slight in degree,
it would follow as a matter of course, that a combination of all the facts observed
upon would have formed the proper basis for life annuities and insurances; but
this being quite otherwise, there was no choice but a combination of two or
three observations, the latest in point of time.
Finlaison also applied a smoothing or graduation technique to the fitting

of the tables derived from mortality experience. Graduation was starting to
emerge as an important topic in mortality table construction at this time and
Finlaison was a pioneer of its real-life application. More generally, Finlaisons
detailed work with experience data rather than population data anticipated
and perhaps partly motivated the next important development in life office
mortality modelling.
Life Office Experience Tables andSelection (18291881)
Despite John Finlaisons scepticism, by the 1820s there was a strengthening

conviction amongst life office actuaries that the selection effects of life under-
writing practices were likely to have a material impact on the mortality rates
that their business experienced. But no quantification of the impact of selec-
tion had been made. How could selection effects be measured? If selection
effects existed, did they have a permanent impact on the mortality behaviour
of policyholders relative to the broader population, or a more temporary one?
If temporary, for how long did the effect persist? These questions could only
be answered through an analysis of the life offices mortality experience data.
The first analysis of mortality experience at a British life office was con-
ducted by William Morgan in 1776 at the behest of Richard Price. As noted
earlier, this investigation led Morgan to conclude that the probabilities of
life in the Society had been higher than those in Mr Dodsons Table, from
which its premiums were computed, in the proportion of three to two. Fifty
years later, as the life assurer with by far the most data on the mortality rates
experienced in its business, the Equitable was naturally positioned to take the
lead on further exploring selection effects. But William Morgan resisted the
development or application of mortality tables based on Equitables mortality
experience for as long as possible. Like most of Morgans later resistance to
change, it was at least partly motivated by reasoned argument as well as mere

intransigence. Morgan held a firm belief that the Equitables lighter mortality
experience relative to the Northampton table was due to a transient selection
effect which would be imprudent to pass on to policyholders via the premium
basis. In 1828, Morgan also argued that moving the valuation of Equitables
business onto the much-lighter experience mortality basis would motivate the
release of an inequitable amount of surplus to older policyholders relative to
younger policyholders.13
For once, Morgans resistance proved futile. Charles Babbage and Griffith
Davies, both leading actuaries of the time, each published their own versions of
Equitable experience mortality tables based on the limited statements Morgan
had made about the Equitable experience relative to the Northampton table.
An Equitable member formally proposed that the Equitables liability val-
uations (and hence assessments of surplus) should be based on the experi-
enced mortality rates. Whilst the Equitables legal counsel found in favour
of maintaining the status quo, the pressure from Equitable policyholders was
sustained and Morgan recognised that it might be more counterproductive
to continue to resist. He finally published an Equitable experience mortality
table in 1829, though this was clearly done through gritted teeth and with
heavy caveats: The following tables are still deduced from documents,
particularly in the earlier parts of life, much too defective to be depended
upon, and have been constructed merely for the purpose of effecting some
calculations directed to be made by the last General Court of the Society.14
Morgans table confirmed the substantial extent to which the Equitables
experience was lighter than the Northampton table that it used in its pricing
and valuations. However, the information provided by Morgan could not
provide much new insight on the source of this differential in mortality. Was
it simply that the segment of society that purchased Equitable life assurance
tended to be richer and healthier than the population at large? Or had mortal-
ity significantly improved since the time of Prices data due to improvements
in economic and social conditions? Or were the Equitables rigorous under-
writing processes and the rejection of lives with above average mortality rates
the main reason for the lighter mortality experience? To answer these ques-
tions, a more comprehensive analysis of the behaviour of the mortality experi-
ence would be necessary (particularly if and how mortality experience varied
as a function of the duration of policy). But the reluctant Morgan would not
sanction any further exploratory work on the Equitable experience data.
13
See Ogborn (1962), p.200.
14
Morgan (1829), Appendix.
William Morgan retired in 1830 and passed away in 1833, aged 83.
His son, Arthur, who was a member of the Equitables actuarial staff when
Morgan senior retired, was elected actuary. Arthur Morgans first major task
was to construct mortality tables of the Equitable experience on a more thor-
ough basis than his fathers somewhat half-hearted effort of 1829. His analysis
included all 21,398 lives who had been members of the Equitable during its
first 67 year years since inception (17621829). The data included a total of
5,144 deaths. This work appears to have been regarded highly by his actuarial
peershe was elected as a Fellow of the Royal Society in 1835 primarily for
his work on the experience tables, and his Royal Society sponsors included
leading actuaries of the day such as Benjamin Gompertz and Griffith Davies.15
Arthur Morgans Equitable experience analysis was the first statistical
analysis of mortality rates that considered the behaviour of mortality rates
as a function of the duration of membership as well as age of member. This
demonstrated that the Equitables experienced mortality rates had varied sig-
nificantly with duration of policy. For example, a 40-year-old who had been
admitted to the society within the previous five years experienced a mortal-
ity rate of around 5 %, whilst a member of the same age who had been a
member for between 15 and 20 years experienced a mortality rate of around
7.5 %.16 This strongly supported the hypothesis that the selection effect of
underwriting was a significant driver of the gap between the mortality experi-
ence and the pricing basis, and that the selection effect, whilst substantial, was
temporary.
Arthur Morgan made his experience analysis publicly available, providing
the opportunity for other actuaries to contribute to the investigation of the
data. T.R. Edmonds, a widely published economist and demographer who
had been appointed actuary of Legal & General in 1832, published an impor-
tant paper in The Lancet in 1837 that discussed the implications of Morgans
results.17 Edmonds argued that the bulk of Equitables mortality profit had
likely arisen from this selection effectthe mortality rates of long duration
policies were not significantly different to the premium basis. This had been
William Morgans consistent position. Edmonds also warned about expecting
this effect to be a permanent feature of the future of life assurance:
15
Ogborn (1962), p.215.
16
Edmonds (1837), p.159.
17
Edmonds (1837).
At the present day, such is the competition among various life offices, that a
large proportion of lives proposed for insurance are now accepted, who would
formerly have been rejected.18
In 1841 Thomas Galloway, the actuary of the Amicable, followed Morgans

example and published the mortality experience analysis of his office. This
data covered the period 18081841.19 The Amicable was the second-largest
British life office of the time and the volume of the data was approximately
one third of that available to the Equitable (1,792 deaths). Galloway per-
formed some innovative analysis that suggested that the selection effect was
not unique to the Equitable. He divided his dataset into two parts: one for
policyholders who were already members at the start of 1808; and another
for those who joined in 1808 or thereafter. His reasoning for this was that
the offices pricing and underwriting standards were fundamentally altered
in 1808. Galloway wanted to measure the selection impact of this change in
underwriting practice. His two mortality experience tables are shown below
in Fig.3.3.
0.15
0.14
0.13
0.12 Amicable (pre-1808 members)
0.11
Amicable (post-1808 members)
0.1
Mortality Rate
0.09
0.08
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0
0 10 20 30 40 50 60 70 80
Age
Fig. 3.3 Amicable mortality experience (18081841)
18
Edmonds (1837), p.162.
19
Galloway (1841).
The post-1808 members experienced materially lower mortality rates

than the pre-1808 members, particularly between the ages of 50 and 70.
Galloways experience analysis provided a measure of the impact of changes in
underwriting practices on mortality experience. This measure of the effect of
selection would, however, have been somewhat over-stated if life expectancies
had improved over the 33-year period, as the average calendar year of experi-
ence of, say, a 60 year-old would have been earlier for the pre-1808 members
than for the post-1808 members.20 Galloway did not note this possible effect,
which itself highlights that no systematic allowance for mortality improve-
ments tended to be considered during this era.
The next important step in the development of experience mortality tables
was initiated at a meeting held at the London Coffee House on Ludgate Hill
on 19 March 1838. There, the actuaries of many of the leading British life
offices met and agreed to jointly share their experience data for the purposes
of producing mortality tables based on their pooled experience. A committee
of senior actuaries including Benjamin Gompertz, Joshua Milne and others
was appointed to oversee the effort. Seventeen life offices submitted data that
totalled 83,905 policies and 3,928 deaths. The Equitable and Amicable were
amongst the contributors to the combined experience, but they did not con-
tribute the full data that they used in their own individual experience studies,
partly because the Seventeen Offices dataset only used policies written on
single lives (to avoid potential double-counting of lives). The combined analy-
sis was eventually published in 1843 and the mortality tables became known
as the Seventeen Offices tables.21
The average duration of the policies in the combined data set was only eight
years. The committee decided that this data could not support an analysis
of mortality rate as a function of duration of policy as well as age of policy-
holder. This had two important implications for the results: given the pub-
lished results of the Equitable and Amicable experience, it was reasonable to
assume that the relative immaturity of the business being analysed meant that
long-term mortality rates experienced by life offices were likely to be higher
than those estimated from this dataset; and, secondly, it restricted the ability
to quantify selection effects. The committee had also set out to analyse cause
of death but found that the returns were deficient in so many of the Lists,
that is was not considered desirable to make any classifications of them.22
20
I am grateful to D.C.E.Wilson for highlighting this point to me.
21
Ansell etal. (1843).
22
Ansell etal. (1843), p. xi.
These limitations meant that no new fundamental insights were obtained

from the combined data. The reports main conclusion was that the combined
experience was very similar to that published by the Equitable a decade earlier.
Nonetheless, the Seventeen Offices table is notable for representing the begin-
ning of a long and important tradition of actuaries pooling experience data
for the purposes of combined mortality studies. The table itself was also used
widely in life office premium bases both in Britain and overseas.
The established life offices naturally accumulated more experience data as
the years moved on, and by 1869 the available data could facilitate a much
richer combined experience analysis. A committee of some of the eminent
actuaries of their generationBrown, Sprague, Bailey, Woolhouse and oth-
erssuperintended the development of a new set of combined tables.23
This time 20 life offices contributed their experience, providing data on
160,426 lives and 26,721 deaths. The 1869 tables are most notable for their
treatment of selection. As we saw above, Galloways Amicable analysis of 1843
had statistically demonstrated that rigorous underwriting, and the selection
of healthy lives for assurance, could generate a significantly lighter mortality
experience. But questions remained. In particular, would this selection effect
be experienced for the entire duration of the policy, or would its effect be tran-
sient, only impacting on mortality for the first few years of the policys dura-
tion? Arthur Morgans Equitable experience analysis suggested the latter, but
no other evidence had yet been produced to corroborate this. Samuel Brown,
the President of the Institute at the time, even worried that the mortality rates
of select lives may ultimately be poorer than other lives24:
It is essential to the prosperity and good management of an office to know not

merely whether the selection of lives affords unusual profits within a short
period after admission of the assured, or how long that period continues, but
also whether it may not require some reserves out of those profits to meet a
permanent excess of mortality hereafter.
When the Seventeen Offices tables were constructed, actuaries were ask-
ing these questions, but the volume and spread of data could not support a
rigorous answer. By 1869 the British life offices had been in existence long
enough to generate the experience data required to build a two-dimensional
mortality table that recorded mortality as a function of the age of the poli-
cyholder and the number of years the policy had been in-force. This data
23
Institute of Actuaries (1869).
24
Institute of Actuaries (1869), p.19.
analysis yielded significant new insights and quantifications on the duration

of selection effects: For all practical purposes, the benefit of selection may
perhaps be said to be lost after the fifth year of assurance.25 It also highlighted
how substantial the selection effect was in the early years of the policy. For
example, the mortality rates of 4044 year-old healthy males in the first year
of assurance were found to be only 37 % of the mortality rate of the same
category in their fifth year or later.26
By the mid-nineteenth century, significant life assurance businesses were
developing outside Europe, particularly in the British colonies and the United
States of America. In the USA, a small number of life assurers had existed
since the second half of the eighteenth century, but it was legislation in
NewYork State in 1840 that provided a catalyst for rapid growth in the sec-
tor. It was estimated that 50,000 life assurance policies were in-force in 1859,
and that this number increased to 800,000 by 1872.27 During this period,
in the absence of any local data, American actuaries used British mortality
tables, most typically Prices Northampton table, Milnes Carlisle table or the
Seventeen Offices table.
In 1859, the first US life office experience table was produced by Sheppard
Homans, based on the experience of the Mutual Life Insurance Company
of NewYork, of which he was the actuary. This table was updated in 1868
and was commonly referred to as the American Experience Mortality Table.
This table was used widely by US life offices over the following fifty years.28
In 1881, American actuaries set a new world record for the amount of experi-
ence data used in a mortality table when the Thirty American Offices Table
was published.29 This table was based on a 30-year experience up to the end
of 1874, and, as its name suggests, 30 life offices contributed their mortal-
ity experience. It included experience data on over one million policies and
46,543 deaths. It claimed to represent a full three quarters of all experience of
life offices in the United States over the 30-year period considered.
Like the Institute of Actuaries table that came before it, the Thirty American
Offices Table provided strong evidence of the impact of underwriting and
selection on mortality experience. For example, for males aged 4045, the
mortality rate experienced in the first year of the policy was found to be around
two thirds of the rate experienced for policies with durations of five years or
25
26
27
Meech (1881), Preface, p.4.
28
Maclean (1948), p.284.
29
Meech (1881).
more.30 In general, the results were consistent with the Institutes conclusions
that the selection effect was evident for the first five years of policies duration
and was not significant thereafter.
By 1881, the practice of pooling experience mortality data was well-
established in developed life assurance markets around the world. Actuarial
thinking moved on from the quantification of selection effects to considering
the implications that this should have for pricing and reserving.
Graduation ofMortality Tables (18251867)
The use of pooled mortality data increased the sample sizes available for use
in the development of mortality tables. But statistical noise was inevitably still
found in mortality rate estimates, and this was especially true when rates were
estimated as a function of both age of policyholder and duration of policy.
Visual inspection of the Carlisle table highlighted that it could also be true
for the contemporary tables based on population data. Some back-of-the-
envelope calculations could easily highlight why. A total of 1,840 deaths were
observed over the nine-year span used in the Carlisle data set. 173 of those
deaths were at ages between 60 and 70.31 So there were less than 20 deaths
observed at age 60: a single observation of one more death of a 60-year-old
within the nine-year period would result in a 5 % proportional increase in the
mortality rate estimate. In the Carlisle table, the mortality rate of 6070-year-
olds increases by an average proportion of 4 % per year of age. So the noise-
to-signal ratio was high. Data samples of this size would unavoidably result
in material irregularities in the observed progression of mortality rates as a
function of age.
Actuarial thinking on the smoothing, or graduation, of age-dependent
mortality rate estimates started to develop in the 1820s. Broadly speaking,
two types of approach emerged: a non-parametric smoothing approach that
set the mortality rate of a given age as a weighted average of the raw mortal-
ity rates estimated across a range of ages centred around the given age; and a
parametric approach that specified a functional form for how mortality rates
varied as a function of age (and potentially policy duration), and then fitted
the parameters of that function to the observed mortality rates. These two
approaches were focused on the same objective, but were philosophically dif-
ferent. The latter approach aimed to provide an explanatory law of mortality
30
Meech (1881), p.31.
31
Milne (1815), p.405.
that was consistent with the data, whilst the former merely tried to remove
the noise in the statistical samples of mortality rates. In a wider context than
actuarial thought, this was an era of exploration in the application of sta-
tistical approaches to social science, and the notion of scientific laws that
could explain social phenomena was in keeping with the zeitgeist of the first
half of the nineteenth century. Perhaps partly for this reason, the explanatory
approach caught the actuarial imagination and became the dominant method.
As we shall see below, it also had more grounded actuarial advantages.
To gain an understanding of these approaches and how they were imple-
mented, we need to recall the overall state of development of statistics during
this era. As we know from Chap. 2, the method of least squares had been
developed and placed in a statistically rigorous context by Legendre, Laplace
and Gauss a decade or so earlier. But this method had not yet been used
to develop a best fit for a function that describes how a dependent variable
(in this case, the mortality rate) behaves alongside an independent variable
(in this case, age). Perhaps surprisingly in retrospect, it took another half-
century before a statistical theory of regression was developed and applied
when Francis Galton considered the hereditary dependencies of plants.32 And
so, lacking a statistical framework to consider how information from neigh-
bouring estimates could statistically improve the estimate of a given point, the
methods proposed for fitting the smoothed rates were inevitably somewhat
ad hoc. There was also some debate about what should be smoothed. Efforts
were focused on the smoothing of the mortality rates, but some thinkers of
the time argued that it would be preferable to apply the smoothing process
directly to the ultimate variable of interest: the value of a life contingency.
Writing in 1838, the influential Augustus De Morgan argued:
The events of single years are subject to considerable error, and generally present
such varieties of fluctuation, that it has become usual to take some arbitrary and
purely hypothetical mode of introducing regularity. This practice cannot be too
strongly condemned, since the tables thereby lose some of their physical facts,
without any advantage ultimately gained. For if by using the raw result of exper-
iments, tables of annuities were rendered unequal and irregular, it would be as
easy, and much more safe, to apply the arbitrary method of correction to the
money results themselves, than to introduce it at a previous stage of the
process.33
See Stigler (1986) Chapter 8.

32
De Morgan (1838), p.162.

33
This perspective did not prevail: virtually all actuarial thinking of the time
focused on smoothing the observed mortality rates. John Finlaison, in his
parliamentary report of 1829 on government annuity pricing, employed a
couple of non-parametric smoothing formulae in his experience mortality
tables, such as:
5Px + 4 Px 1 + 4 Px +1 + 3Px 2 + 3Px +2 + 2 Px 3 + 2 Px +3 + Px 4 + Px+4

Px ( smoothed ) =
25
where Px is the observed probability of a life aged x surviving one year.

In Finlaisons approach, each smoothed mortality rate is set as a weighted-
average of the raw mortality rate observations found for ages up to four
years older and younger than the age of the rate being estimated. The weights
used vary inversely with the distance of the age of the observation from the
age being estimated. The above formula and its logic are quite similar to a
modern-day local or non-parametric regression method.
In 1839, W.S.B.Woolhouse, in the early years of a long and distinguished
career as an actuary and mathematician, suggested another non-parametric
smoothing approach in his paper on the observed mortality rates of the Indian
army.34 Woolhouses approach made adjustments to the observed mortality
rates such that the time progression of the lives remaining in a closed popula-
tion would have a regular pattern. Statistically, he aimed to ensure that the
pattern of the fourth order of differences in the progression of lives remained
as smooth as possible. He devised an iterative arithmetic process that could
be implemented to obtain this objective.35 This approach entailed more arith-
metic manipulation than Finlaisons smoothing approach, but had a more
explicit statistical objective.
Finlaison and Woolhouse appeared satisfied with the performance of these
methods. The Woolhouse method had a greater impact on actuarial practice
and it was applied in the development of the Seventeen Offices tables of 1843.
But others had less success with these approaches. In his Amicable experience
analysis of 1841, Galloway attempted to use a smoothing formula of the kind
proposed by Finlaison but found that considerable anomalies remained.36
He therefore reverted to the use of a parametric function, and this increas-
ingly became the standard actuarial practice.
34
Woolhouse (1839).
35
Woolhouse (1839), p.7.
36
Galloway (1841), p. ix.
The parametric function approach had another advantage over the non-
parametric smoothing methods. Since the time of de Moivre, it was recog-
nised that specifying a particular mathematical form for the behaviour of
mortality could provide an annuity pricing formula that involved signifi-
cantly less arithmetic operation than the explicit calculation of expected cash-
flows directly from a given set of mortality rates. De Moivres assumption of
arithmetic decrements in deaths could be viewed as the first parametric form
of mortality graduation. For de Moivre, the object of this assumption was
entirely motivated by the improvement in the efficiency of the annuity pric-
ing calculation. One hundred years later, enough progress had been made in
arithmetic computation to make full calculation of single-life annuity prices
accessible. But the calculation of the prices of annuities that were written on
two or three lives was still very challenging. So in the early nineteenth century
actuaries such as Gompertz hoped that parametric functions for mortality
rates could kill two birds with one stone: providing an appropriate way of
smoothing out the noise in sampled mortality rates, and providing efficient
formulae for the pricing of complex annuities.
Gompertz started a revolution in graduation thinking in 1825 when he
developed a parametric function that was intended to be consistent with the
fundamental characteristics of how mortality should behave as a function of
age (a law of human mortality)37:
It is possible that death may be the consequence of two generally co-existing

causes; the one, chance, without previous disposition to death or deterioration;
the other, a deterioration, or an increased inability to withstand destruction.38
This suggested there were two forms of exposure to mortality: one that was
constant across all ages; and the other that increased with age. To model the
behaviour of the age-varying component of the mortality rate, he assumed
the average exhaustions of a mans power to avoid death were such that at the
end of equal infinitely small intervals of time, he lost equal portions of his
remaining power to oppose destruction which he had at the commencement
of those intervals.39
The above statement implied that the age-dependent component of the
mortality rate increased exponentially with age. Gompertz then considered
how well this assumed mortality behaviour could fit to standard mortality
37
Gompertz (1825).
38
Gompertz (1825), p.517.
39
Gompertz (1825), p.518.
tables such as Prices Northampton table and Milnes Carlisle table. Curiously,
when Gompertz came to the application of his formula, he chose to omit the
age-independent mortality component that he had earlier described. Hence a
Gompertz function only includes an age-dependent exposure, even though
he expressly identifies a constant age-independent source of mortality (chance
without deterioration).
Gompertz originally expressed his function in terms of the number living
at age x, lx, rather than the force of mortality. T.R.Edmonds, writing a few
years later in 1832,40 defined the force of mortality, x, as the continuously
compounded rate of mortality and expressed the formula in those terms. The
modern form of the Gompertz function would typically be written:
x = e x
It was implemented by Edmonds without recourse to the exponential func-

tion as:
x = x
Today, such a function would typically be fitted by finding the parameters that
minimise the squared errors between the observed and fitted mortality rates.
As discussed above, this regression-style approach to function fitting had not
yet been applied to statistical problems, and Gompertz used a more heuristic
approach. He considered the change in mortality rates that applied over the
ten-year gap between ages 15 and 25, and over the ten-year gap between ages
45 and 55. These observations could be used to solve two simultaneous equa-
tions that uniquely determined the two parameters of the function.
Whilst the parameters would fit exactly to those equations, they would
not produce an exact fit to all the mortality rates of the table. However, he
was pleased with the quality of fit he obtained when he compared his func-
tion to the observed rates of the Northampton table between ages fifteen
and 60, noting: This equation between number living and the age is deserv-
ing of attention, because it appears corroborated during a long portion of
life by experience.41 Figure 3.4 shows a Gompertz function fitted to the
Northampton data for ages 1560.
40
Edmonds (1832).
41
Gompertz (1825), p.519.
0.05
0.04 Northampton Table

Mortality Rate
0.03
0.02
0.01
0
15 25 35 45 55
Age
Fig. 3.4 Gompertz mortality function fitted to Northampton table (ages 1560)
The above chart illustrates how the Gompertz function can provide a very
useful graduation of the Northampton table over this range of ages. However,
the limitations of the function become apparent when a broader range of ages
is considered. Figure3.5 fits to the 1580 age range and again shows the fitted
function values and observed mortality rates.
Clearly, the two-parameter function is unable to provide an adequate fit over
this wider age range. The same conclusion was reached when Gompertz applied
the function to other standard mortality tables. Nonetheless, the Gompertz
law of human mortality was widely celebrated by actuaries and medical think-
ers of the time. De Morgan wrote of Gompertz: As this ingenious paper con-
tains a deduction from a principle of high probability, and terminates in a
conclusion which accords in a great degree with observed facts, it must always
be considered a very remarkable page in the history of the enquiry before us.42
It was clear, however, that as a practical actuarial tool, improvements
were required. William Makeham, another senior actuary of his generation
(and also a noted mathematician), proposed a natural generalisation of the
Gompertz law in papers published in 185943 and 1867:44 he suggested includ-
ing the age-independent parameter that Gompertz had described but chosen
to ignore in his formula. The force of mortality could then be written as:
42
De Morgan (1839).
43
Makeham (1859).
44
Makeham (1867).
0.15

Mortality Rate
0.09
0.06
0.03
0
15 20 25 30 35 40 45 50 55 60 65 70 75 80
Age
Fig. 3.5 Gompertz mortality function fitted to Northampton table (ages 1580)
x = x +
This was, of course, exactly in keeping with Gompertzs original description

of the effects of mortality on age. We can only speculate as to why Gompertz
did not proceed directly to this functional form in his paper of 40 years earlier.
It clearly would have resulted in a more complicated fitting process, but his
heuristic fitting approach could be extended to the three-parameter case in a
straightforward way. Makeham found that the additional degree of freedom
permitted significantly more accurate fits to mortality tables, as illustrated
below in Fig.3.6.
Whilst the additional degree of freedom naturally resulted in better fits
to mortality data, Makeham was at pains to emphasise that his extension
was not an arbitrary additional degree of freedom but rather something that
captured the characteristics of a fundamental law of mortality. His extension,
he wrote, in no way interferes with the philosophical principle upon which
Mr Gompertz has shown his theory to be based: a feature which distinguishes
his formula from all others which have hitherto been proposed, and which
doubtless accounts for the favourable reception it has met with from the high-
est scientific authorities.45
The three-parameter Makeham function could provide a good fit to mor-
tality data within the range of ages of primary importance to life assurers,
45
Makeham (1867), p.333.
0.15

Gompertz fit (ages 15-80)
Makeham fit (age 15-80)
Mortality Rate
0.09
0.06
0.03
0
15 20 25 30 35 40 45 50 55 60 65 70 75 80
Age
Fig. 3.6 Makeham and Gompertz mortality functions fitted to Northampton

table (ages 1580)
but Makeham and Gompertz both recognised that such functions could
not provide a satisfactory fit across the entire spectrum of ages in mortality
tables, particularly at very young and very old ages. They each proposed vari-
ous extensions. Gompertz proposed using piece-wise functions with differ-
ent parameterisations of Gompertz functions applied to specified age bands.46
Makeham proposed extending the function further by adding polynomial
terms.47 Elsewhere in Europe, other mathematically inclined actuaries devel-
oped similar functions for graduation purposes. For example, the leading
nineteenth-century Danish actuary T.N.Thiele proposed a seven-parameter
exponential function in 1871.48
The significant developments in statistics that occurred in the early twenti-
eth century provided a further fillip to actuarial research in mortality gradu-
ation. Elderton49 and Ogborn50 each considered the application of Pearson
frequency curves to mortality table graduation, but neither was able to dem-
onstrate any significant advance in performance relative to the Gompertz-
Makeham framework. It remains today an important and well-used part of
the toolkit of mortality actuaries and demographers.
46
Gompertz (1871).
47
Makeham (1889).
48
Thiele (1871).
49
Elderton (1934).
50
Ogborn (1953).
ctuarial Thought onInvestment Strategy,

A
18581952
British actuarial thought on investment strategy for life offices began to
emerge, somewhat tentatively, in the third quarter of the nineteenth century.
Barely a handful of investment papers were published in the Journal of the
Institute of Actuaries and Transactions of the Faculty of Actuaries in the second
half of the nineteenth century. These papers, however, tended to be written by
senior and influential actuaries. They were leading the profession into a new
field that would become ever more significant in twentieth century actuarial
thought and practice.
There were several drivers for the increasing interest in this topic in the
mid-nineteenth century. The experience of the previous decades had imbued
actuaries with a growing sense of confidence in their modelling of mortality
rates. A long track record of success in predicting mortality rate behaviour
and profiting from it had been established. The plagues and epidemics of the
seventeenth and eighteenth centuries were, by then, largely distant memories.
Mortality behaviour appeared to be governed by laws that were well under-
stood by the actuarial profession. With this mastery came the realisation that
the increasingly competitive life assurance landscape could not yield the level
of mortality profits that had been earned on assurance business over the previ-
ous 100 years. Mortality protection was becoming commoditised.
This was accompanied by an increasing sense that canny investment
decision-making could be an important source of profit and differentiation
for life assurers. In 1858, actuaries would read in the Journal:
The two principal sources of profit to an Assurance company are the selection of
lives and the accumulation of the excess of the premiums at a higher rate of
interest than is assumed as the basis of the Companys operations With
regard to the [selection of lives], it is impossible not to feel that the margin of
profit has very much diminished of late years This makes the question of the
rate of interest obtained from investments so much the more important: it is, in
fact, the principal source of profit.51
And in 1862:
Of the two main elements on which all life assurance transactions depend the
rate of mortality and the rate of interest the latter, I think, affords more scope
for the exercise of judgement and skill than the former there can be no doubt
51
Brown (1858), pp.24142.
that the amount of interest realised on the assets can be materially influenced by
the degree of judgement and knowledge brought to bear upon the subject.52
These statements could be viewed as the very first steps on a long slip-
pery slope that led to twentieth-century British actuaries taking actions which
suggested that they regarded geared exposure to financial market risks as an
equivalent and equally sustainable form of risk taking to the underwriting
of largely diversifiable mortality risk. That was not, however, the vision of
the nineteenth-century British actuarial profession. Rather, these early con-
siderations of the asset side of the balance sheet can be viewed as a natural
broadening of the actuarial horizon as the skills, experience, confidence and
influence of the profession grew. It was also perhaps inevitable in the context
of the long-term trend in British life assurance whereby the products invest-
ment element became ever more significant relative to the protection element.
In the early nineteenth century, British life offices invested in two main
asset classes: long-term government bonds (usually consols, which were per-
petuities) and mortgages on freehold property (including agricultural lands).
Long-term government bonds served two purposes for life offices. First, whilst
the emergence of concepts such as duration and interest rate immunisation
were many years away, there was an intuitive understanding that the long-
term nature of the liabilities demanded that some funds should be invested in
assets that paid similarly long-term cashflows. Second, in this era, the possibil-
ity of an epidemic or plague, whilst viewed as increasingly unlikely, was still a
credible contingency. Funds needed to be readily convertible to cash in such
circumstances, and so the liquidity offered by government bonds was valued.53
The main attraction of mortgages was that they offered higher yields than
government bondshigh-quality mortgages typically yielded between 50 and
100 basis points more than long-term government bonds.54 The mortgages,
however, offered neither liquidity nor long-term fixed cashflows, and these
characteristics constrained the amounts that were invested in the asset class.
By the middle of the nineteenth century, the concerns about the possibility
of epidemics had receded and with it so did the requirement for asset liquid-
ity. The amounts invested in mortgages correspondingly rose. This investment
strategy stands up well to modern scrutiny: it was a strategy that paid respect
to the liabilities interest rate duration and liquidity characteristics and which
aimed to obtain additional yield through reducing excess asset liquidity rather
than by materially increasing market or credit risk.
52
Bailey (1862), p.143.
53
Deuchar (1890), p.451.
54
Brown (1858), pp.245, 25354.
The first actuarial paper on life assurance investments was published by

Samuel Brown in 1858.55 Brown was a senior figure in the actuarial profes-
sion of the time. He was actuary of the Guardian Assurance Company, a
well-established British life office, and he would go on to be President of the
Institute of Actuaries between 1867 and 1870. Brown did not propose a doc-
trine for investment strategy. He restricted himself to providing a commentary
on the main asset classes that life assurance funds invested in, and undertook a
small survey of the asset allocations of four major British life assurance offices
of the time (Eagle Life, London Life, Rock Life and Metropolitan Life). The
survey showed that, across the four offices, the amounts invested in mortgages
varied between 46 % and 79 % of their total assets.
By the 1850s, low interest rates had rendered long-term government bonds
increasingly unattractivethe consol yield had fallen from almost 5 % in
1810 to barely above 3 % by 1855.56 The 3 % yield level was important to
actuaries because virtually all with-profit business had been written using a 3
% interest rate in the premium basis. If funds could not generate a 3 % return,
the life offices profitability, and ultimately their solvency, was threatened. This
fall in yield triggered significant reductions in allocations to consolsby 1858
the allocations had been reduced from perhaps as much as 50 % earlier in
the century to between 5 % and 15 % across the four offices. The funds
were mainly switched into high-quality bonds (such as colonial sovereign
bonds) and debentures (mainly on railways and docks). These asset classes had
increased in size and liquidity over the first half of the nineteenth century with
the continuing progress of the industrial revolution and the British Empire.
Debentures typically offered between 50 and 150 basis points more than con-
sols.57 They also provided more liquidity and more duration than mortgages,
but were accompanied by a more material default risk. This is arguably the
first example of a phenomenon that reoccurs in actuarial history: long peri-
ods of falling long-term interest rates (for example, in Britain in 19301945)
tended to compel institutions to bet their way out of their potential unprofit-
ability by increasing investment risk rather than locking in a loss.
This brief sketch of the nineteenth-century asset allocation practices of
British life offices highlights a number of asset-liability management consid-
erations that were the natural domain of the actuary: how much asset liquid-
ity was required by life assurance liabilities? How much investment and credit
risk should be sought by the asset strategy and how did this choice interact
55
Brown (1858).
56
Homer and Sylla (1996), p.182.
57
Brown (1858), pp.25354.
with pricing, reserving and bonus policy? Did the very long-term nature of
the liabilities create particular forms of risk that would help determine a desir-
able asset profile? The history of actuarial consideration of questions such as
these is explored below.
Actuaries andInvestment Liquidity (18621933)
A second paper on life office investments followed Samuel Browns survey

a few years later in 1862. Arthur Hutcheson Baileys On the Principles on
which the Funds of Life Assurance Societies should be Invested was the first
actuarial paper that analysed investment strategy in the context of life assur-
ance liabilities.58 At five pages in length, it also has the distinction of being
one of the shortest ever actuarial papers. It became a classic paper and was core
reading in British actuarial examinations for at least 50 years following its pub-
lication. Bailey was the actuary of London Assurance, a significant life office
of the period (and one of the very oldest, having been established by Royal
Charter back in 1720). Like Samuel Brown, Bailey too went on to become
President of the Institute of Actuaries, in Baileys case from 1878 to 1882.
Bailey considered investment strategy in the context of the characteristics
of life assurers liabilities. The nature of these liabilities would inevitably place
some constraints on the features of an appropriate strategy. Bailey, however,
also recognised that these liabilities could create some unique comparative
advantages for life assurers relative to other financial institutions:
They (life assurers) engage to pay fixed sums of money at periods generally long
distance from the time when the contracts are entered into the probable
amount of demands on their resources can be calculated from time to time
within not very wide limits. Life assurance Societies, unlike banks and commer-
cial enterprises generally, are not exposed to sudden or unusual demands on
their resources in times of panic and financial difficulty.59
Put simply, life assurance liabilities are long-term, predictable and illiquid,
especially when considered relative to those of banks. Bailey took it as a given
that illiquid assets offered additional compensation to investors for sacrificing
liquidity. He hence argued that life assurers should take full advantage of their
unusual capacity to bear asset illiquidity:
Bailey (1862).
58
Bailey (1862), p.143.

59
The much larger proportion [of life office assets] may safely be invested in secu-
rities that are not readily convertible; and it is desirablethat it should be so
invested, because such securities, being unsuited for private individuals and
trustees, command a higher rate of interest in consequence [emphasis added].60
Mortgages, which Bailey defined as every species of loan secure on tan-

gible property, were the main asset class he identified for investment in secu-
rities that are not readily convertible. He recognised that the size of these
asset classes may be exhausted by the growing funds of life assurers and hence
that alternative illiquid asset classes would need to be found. He identified
loans for land development under recent Land Drainage Acts as a natural
candidateessentially, nineteenth century infrastructure debt.
Bailey viewed the illiquidity premium as the way in which life assurers
should look to enhance their investment yield. He did not believe that life
assurers should attempt to increase investment returns by taking equity risk
or substantial credit risk. His starting investment principle was That the
first consideration should invariably be the security of the capital,61 and the
notion of life assurers investing in equity investment was anathema to Bailey:
It is, I think, generally admitted that the ordinary stock and shares of trading
and other companies are not eligible for Assurance Societies investments, as
being too speculative.62
Baileys thinking stands the test of time strikingly well. To the modern actu-
arial reader there are clear similarities with Baileys arguments and contempo-
rary ideas on investment policy for long-term non-profit life business such as
fixed annuities. The concepts of competitive advantage through unique long-
term illiquid funding and the availability of asset illiquidity premia are increas-
ingly central to modern actuarial thinking on investment strategy for illiquid
liabilities such as these. Asset classes such as mortgages and loans remain the
natural asset classes for life assurers intending to access illiquidity premia.
Baileys investment doctrine was largely consistent with the prevailing mid-
nineteenth-century asset allocation practices described above. By 1890, life
office allocations to mortgages were around 52 % and approximately 22 %
was invested in various forms of illiquid loans. Another 6 % was invested
directly in property. So by 1890, fully 80 % of funds were invested in illiquid
asset classes. The remaining assets were invested in a range of asset classes such
as British, foreign and colonial government securities (8 %), debentures (5 %)
and cash and deposits (2 %).63
60
Bailey (1862), p.144.
61
Bailey (1862), p.143.
62
Bailey (1862), p.145.
63
Mackenzie (1891), p.195.
The final quarter of the nineteenth century was the high watermark for
illiquid asset investment by British life assurers. In subsequent years, British
life offices started to tilt their asset allocation towards more liquid assets. The
final decade of the nineteenth century marked a decisive shift away from mort-
gages. Allocations to more liquid stock exchange securities such as long-term
debentures and government bonds correspondingly increased. Between 1890
and 1900, the percentage of life assurance assets invested in stock exchange
securities increased from around 18 % to 39 %. By 1910, this percentage had
reached 49 %; by 1920 it was 60 %.64 A similar trend was observed amongst
the US life offices, and indeed started a decade earlier following the US bank-
ing crisis of the early 1870s. The US offices, however, retained a larger mort-
gage allocation than the British offices after the switch. Between 1874 and
1920, US life offices investments in mortgages had been reduced from 54 %
of assets to 34 %.65
This great rotation from illiquid to liquid assets that occurred between 1890
and 1920 was not solely due to a desire for greater asset liquidity, though that
was a factor. Other drivers were also important. There was a substantial fall in
the value of British agricultural land during this era (perhaps as much as 30
%). This may have awakened life offices to the potential risks inherent in the
mortgage asset class. It also reduced the value of the land upon which new
mortgages could be offered. Increasing competition amongst mortgage lend-
ers emerged, with legislative changes making the provision of mortgages more
accessible to other investors such as trust funds. Mortgage yields fell.66 In
summary, there was a reduction in the illiquidity premium on offer, a reduc-
tion in the size of the market, and perhaps a re-appraisal of the risks associated
with the asset class. All whilst the size of funds available for investment by life
offices was increasing substantially.
The First World War also had a direct and significant impact on asset allo-
cations. In particular, it resulted in a very substantial increase in life offices
allocations to British government securities. Life offices were not legally
compelled to buy government debt, but they reached a form of patriotic
gentlemans agreement with the government where they agreed to buy gilt
issues during the war.67 As a result, investment in British government bonds
increased from 1 % to 35 % of life office assets between 1915 and 1920.
However, this was not the main driver of the move from illiquid to liquid
assets: much of the increase in gilt allocations was financed by reductions in
64
Murray (1937), p.259.
65
C.M.Gulland, in Discussion, Murray (1937), p.274.
66
67
Whyte (1947), p.223.
liquid debenture securities (the asset allocation of which was reduced from
30 % to 12 %). Nonetheless, the scale of the gilt allocation necessarily further
reduced illiquid asset holdings. Mortgage allocations fell from 21 % to 15 %
between 1915 and 1920.68
Changes in the liquidity profile of life assurance liabilities played a signifi-
cant part in motivating this fundamental shift in asset allocation and liquidity
profile. The liquidity requirements of the liabilities were growing by the end of
the nineteenth century. Unlike 100 years earlier, these liquidity requirements
were not driven by the risk of mortality spikes arising from epidemics (whilst
the influenza epidemic of 1918/19 did generate an unexpected concentra-
tion of claims, its impact was still relatively small in comparison to premium
incomes).69 Instead, the liabilities liquidity profile was being changed by sur-
render practices. Surrender rates had increased. In the fast-growing industrial
assurance sector of the 1920s, double-digit surrender rates were the norm.70
Historical anecdotes from the USA also highlighted how life liabilities may
unexpectedly experience sudden liquidity demands. In 1873, a US banking
liquidity crisis had created a surge of demand for the surrender of life assur-
ance policies as these policies became the policyholders most direct route to
cash. In British life assurance, there was an increasing practice, driven by com-
petitive forces, of offering guaranteed surrender values (legislation passed in
1929 may also have encouraged guarantees in surrender values on some types
of business).71 A 1933 paper by William Penman noted that if very large
sums can be withdrawn, virtually without penalty and at very short notice, it
becomes necessary to revise materially our views of Life Assurance finance
it seems to me that the investment position, to the extent that guaranteed sur-
render values are made excessively liberal, approximates to that of a bank.72
These changes in surrender practices materially impacted on the desired liquid-
ity profile of the assets. With-profit life policies were never again thought of as
illiquid. Today, actuaries illiquid investment strategies are largely focused on lia-
bility types such as non-profit fixed annuities that do not offer surrender values.
Actuaries andInterest Rate Risk (18581952)
Actuarial awareness of interest rate risk was first heightened by the economic
experience of the early nineteenth century. Interest rates increased significantly
68
Dodds (1979), Table A.3.
69
Clayton and Osborn (1965), p.65.
70
Clayton and Osborn (1965), p.64.
71
Laing in Discussion, Penman (1933), p.421.
72
Penman (1933), p.391.
during the Napoleonic Wars, and then experienced a long and substantial
decline for the following next forty years (Fig. 3.7). In 1838, Augustus De
Morgan wrote:
I consider the fluctuations of mortality as very little to be feared, compared with

those of the rate of interest the rate of interest has been halved within the
memory of man, and a heavy war might double it again.73
Whilst Baileys thinking on illiquid investment strategies strikes a contem-

porary chord, nineteenth-century actuarial thought on interest rate risk was
several steps removed from the sophisticated and quantitative techniques that
abound in twenty-first-century yield curve risk management practices. But
from the time of Richard Price and William Morgan, there was a general appre-
ciation that funds backing long-term fixed liabilities should include assets that
deliver long-term fixed cashflows. The perpetual format of the standard long-
term British government securities of the time (consols) confounded any pro-
totypical matching method. It also created a strong distinction in the mind of
the actuary between income and capital: with a risk-free perpetuity, income is
guaranteed, but the capital is only ever returned through the sale of the bond
at the then-prevailing market price. In his 1858 paper, Samuel Browns discus-
sion of consols noted that the certainty of the income is dearly purchased at
6
Long-term Government Bond Yield (%)
0
1750 1800 1850 1900 1950
Year
Fig. 3.7 UK long-term government bond yield (17501950) (Mitchell (1988))
De Morgan (1838), p.260.

73
the price of the uncertainty of the return of their capital and so many events
cause fluctuations in the Funds [consols], that he never knows what amount
of his capital he may be able to recover when he needs it.74
Baileys analysis of long-term government bonds as an asset class for life
assurance liabilities continued in the vein started by Brown, and was even
more emphatic:
I believe it to be true that the English Funds [consols] are altogether unsuited
for life assurance investments. For income they offer probably the best security
the world has yet seen; but with us that is a secondary consideration; the capital,
the security of which is our first object, is subject to very inconvenient fluctua-
tions in value.75
There was no substantive development in technical actuarial thinking on

interest rate risk beyond Brown and Baileys observations until the 1930s,
when several papers emerged following the economic turmoil of the First
World War and the Great Depression. However, it has been argued that an
increasing awareness of the reinvestment risk created by the potential for fur-
ther falls in interest rates may have been another motivating factor in the
switch from mortgages to bonds during the 18901920s period.76
British long-term interest rates had fallen from 5 % at the end of the
Napoleonic Wars to 2.5 % by the start of the twentieth century. These falls
were almost fully reversed over the traumatic first quarter of the twentieth
century, peaking at 4.5 % in 1925. Rates then started to fall sharply again
as government policy attempted to stimulate economic growth during the
depression era. This volatility motivated the first wave of actuarial ideas on
what would now be called asset-liability management. In 1933, Charles
Coutts, then President-Elect of the Institute, opined that an attempt should
be made to marry the liabilities and the assets as far as possible.77
There was a growing recognition amongst the profession that the investment
strategy for assets backing long-term life liabilities was more than merely inci-
dental to actuarial work. But no actuarial framework for investment strategy
existed beyond Baileys 70-year-old investment principles. The profession rose
to the challenge and a number of important papers were presented on interest
rate risk and asset-liability management over the following 20 years. In the
1930s, there was a lack of precision both in the objectives and the d esirable
74
Brown (1858), p.244.
75
Bailey (1862), p.144.
76
See Murray (1937), p.249.
77
Coutts in Discussion, Penman (1933), p.425.
features of strategies that would successful marry assets and liabilities. But
clarity gradually emerged. In the Institutes sessional meeting that discussed
an investment paper by A.C.Murray in 1937, J.D.Binns submitted:
Regarding the office as a closed fund, owing to the incidence of future income
it will be found that assets should be made to mature at rather later dates than
the corresponding liabilities. Then should interest rates fall the future income
will be accumulated at a lower rate of interest, but this would be compensated
by the appreciation shown by the assets at the time that the corresponding lia-
bilities fall due (and vice versa if interest rates go the other way).78
This is a non-mathematical anticipation of Redingtons immunisation the-

ory (which would be presented to the Institute fifteen years later). Redington
would provide a revolutionary degree of clarity, but the crucial essence of his
immunisation idea was bubbling under within the actuarial profession for
decades prior to his paper.
Actuaries were not the only ones thinking about interest rate risk at this
time. Economists of the period were also making important strides in their
measurement of interest rate behaviour. The most notable contribution was
from Frederick Macaulay, the Canadian-born economist. He spent much of
his professional career in the USA as a researcher at the National Bureau of
Economic Research. He was the son of T.B.Macaulay, the influential actuary
of Sun Life of Canada (and son of a Scots immigrant).
Frederick Macaulay undertook ground-breaking empirical analysis of the
long-term behaviour of yield curves in the USA since the time of the US Civil
War. In his statistical analysis of the relative behaviour of bonds of different
maturities, he recognised the inadequacy of a bonds term to maturity as a
measure of its longness. He saw that an accurate measure must take account
of the size and timing of its coupon payments as well as its maturity. Macaulay
introduced the term duration to measure the longness of the bond and
defined it as the weighted average of the time of the cashflows where the
weights are the present values of the cashflows.79 He did not deduce this defi-
nition from a set of first principles or objectives, but arrived at it in an ad hoc
way, noting it was an intuitive and well-behaved measure:
It would seem almost natural to assume that the duration of any loan involv-
ing more than one future payment should be some sort of weighted average of
the maturities of the individual loans that correspond to each future payment.
J.D.Binns in Discussion, Murray (1937), p.278.

78
Macaulay (1938), pp.4546.

79
Two sets of weights immediately present themselves the present and the future
values of the various individual loans Future value weighting seems clearly
inadmissible. It gives absurdly long durations The argument for present
value seems strong.80
Macaulay did not recognise that his duration measure was also the first
derivative of the bond price with respect to changes in the interest rate. He
did, however, find that it was extremely useful in describing the empirical rela-
tive behaviour of different bonds: The concept of duration throws a flood of
light on the fluctuations of bond yields in the actual market.81
Macaulays duration measure had no discernible impact on economic
research or financial market practice for over 30 years. Then, in the early 1970s,
the confluence of the rapid growth in fixed income securities with important
breakthroughs in quantitative financial economics research suddenly placed it
firmly at the centre of fixed income risk management. The actuarial profession
appears to have been wholly unaware of Macaulays work in the years follow-
ing its publication. Macaulay himself had no appreciation of its implication as
a risk and asset-liability management tool. His discussion of duration covered
a handful of pages in a 600-page tome on nineteenth-century US interest
rate behaviour. In these circumstances, it is perhaps unsurprising that the
latent usefulness of Macaulays duration measure in asset-liability manage-
ment passed by the actuarial profession (and everyone else).
Following the hiatus of the Second World War, the British actuarial profes-
sion embarked on its most fertile period of thought leadership activity since
the third quarter of the nineteenth century. Two important papers on inter-
est rate management were published in 1952Redingtons famous paper on
immunisation theory82 and a paper by Haynes and Kirton83that tackled
similar subject matter. Whilst these papers would have been somewhat con-
troversial at the time of publication or at the least outside the norms of con-
temporary thinking, with historical hindsight they can now be viewed as the
natural culmination of the developing interest rate and asset-liability manage-
ment thinking of the previous 20 years (in particular, the Penman and Murray
papers of the 1930s and their respective Staple Inn Discussions).84
Frank Redington spent his entire working career at Prudential Assurance,
a leading UK life assurer, joining in 1928 after graduating in mathematics at
80
Macaulay (1938), p.47.
81
Macaulay (1938), p.50.
82
Redington (1952).
83
Haynes and Kirton (1952).
84
Penman (1933), Murray (1937).
Cambridge. He became chief actuary there in 1951 and remained in that posi-
tion until his retirement in 1968. He was President of the Institute of Actuaries
from 1958 to 1960. He is most famed, both in actuarial circles and in the
broader financial community, for his work on duration and immunisation. His
contribution to actuarial thought-leadership and practice, however, was much
broader than immunisation theory. He published widely throughout his career
on a range of actuarial topics, most notably on with-profit reserving, and valua-
tion and bonus policy. Indeed, the paper in which he introduces his immunisa-
tion theory was entitled Review of the Principles of Life Office Valuations, and
the section on immunisation was a last-minute addition to it.85
Redingtons immunisation theory took the amorphous actuarial intuition on
interest rate matching that had gradually developed over the previous quarter
of a century and crystallised it into some clear mathematical statements. In
this work, Redington produced the same measure of duration that Frederick
Macaulay published some 25 years earlier. Redington was not, however, aware
of Macaulays work, and Redingtons approach was the polar opposite to that
taken by Macaulay: where Macaulay had been ad hoc and empirical, Redington
was mathematical and deductive. Macaulay had been interested in explaining
the historical price changes of different bonds. Redington was interested in
matching assets and liabilities, where matching meant the distribution of the
term of the assets in relation to the term of the liabilities in such a way as to
reduce the possibility of loss arising from a change in interest rates.86
It is often the hallmark of a great idea that it appears disarmingly simple
after someone has thought of it. Redingtons immunisation theory belongs in
that category. He was aware that the results of the mathematical treatment
he was about to introduce were intuitive or even inevitable: in the broadest
sense, it is apparent without mathematical proof that if the liability-outgo and
asset-proceeds are to be equally sensitive to changes in the rate of interest they
must have roughly the same mean terms.87 Writing 30 years after the publica-
tion of the paper, Redington explained how his fundamental development of
the theory arose:
As is so often the case once you look at a problem straight between the eyes and
with undivided attention you find yourself staring the answer in the face. The
actual assets and liabilities were known facts if their present values were equal
at one rate of interest and remained equal on a shift in the rate of interest this
85
Redington (1982), pp.8485
86
Redington (1952), p.289.
87
Redington (1952), pp.28990.
was only the laymans way of saying that the differentials of the present values
with regard to the rate of interest must be equal. The basic equation followed
immediately as an elementary piece of differential calculus.88
Redington set out the Taylor series expansion of the change in assets and
liability values for a given change in the interest rate i (which, importantly,
is assumed to be constant as a function of the term of bond). From this
expansion he was able to argue that a satisfactory immunisation policy is one
where two conditions hold:
dVA dVL
=
di di
and
d 2VA d 2VL
> 2
di 2 di
where VA is the value of the asset portfolio and VL is the present value of the
liability cashflows.
Assuming asset and liability cashflows are fixed, we can write the asset and
liability values as:
VA = v t At

VL = v Lt
t

where v=1/(1+i) and At and Lt are the asset and liability cashflow respectively
due at time t. Taking the derivatives of the above asset and liability values with
respect to i, the first immunisation condition implies:
tv A = tv L
t
t
t
t

Redington noted that this condition is simply that the mean term of the value
of the asset-proceeds must equal the mean term of the value of the liability
out-go.89 The weights used in the calculation of the mean term are the ratios
88
Redington (1982), pp.8485.
89
of the present value of the cashflow due at that time to the total present value.
He did not know it at the time, but he had shown that Macaulays duration
measure was the first derivative of the bond price with respect to the interest
rate, and that immunisation therefore required equating the duration of assets
with liabilities (the term duration never appears in Redingtons paper).
Redingtons second immunisation condition implies that a successful
immunisation strategy should satisfy the secondary condition:
t v A > t v L
2 t
t
2 t
t

He intuitively describes this condition as the spread of the value of the asset-
proceeds about the mean term should be greater than the spread of the value
of the liability-outgo.90
The major limitation of Redingtons immunisation analysis was its assump-
tion of a uniform interest rate applicable over all maturities (that is, a flat yield
curve). This was a natural modelling assumption to make and Redington
noted the practical complication that Yields are not uniform for all terms
of assets, and the differentials are not stable in time.91 But he did not ponder
much on the limitations that this assumption placed on the implications of
his analysis. This is doubtless easy to see in hindsight after several intervening
decades of quantitative yield curve modelling development, but it is none-
theless true that the direct implications of Redingtons equations could lead
to dangerously misleading insights. Redingtons two immunisation equations
imply that the optimal asset strategy would entail equating asset and liability
duration whilst maximising the spread of asset cashflows around the liability
duration. This implies that the best asset strategy would be the mix of cash
and perpetuity (or, even better, very long-dated zero-coupon bond) that has
the same duration as the liability cashflows. Such a strategy, however, is merely
maximally exploiting the limitations of the uniform yield assumption. If we
consider the possibility that movements in yields may be imperfectly corre-
lated over the term structure, or that long-term yields will be less volatile than
short-term yields, we can see that the cash/perpetuity strategy would gear up
on risks that are not recognised by Redingtons analysis. This is not a point
that he directly addressed.
Haynes and Kirton independently published a paper on cashflow matching
only a month prior to Redingtons immunisation paper.92 Their paper did not
90
91
92
Haynes and Kirton (1952).
develop the duration or immunisation concepts, but they did find that asset
portfolios that used combinations of cash and perpetuity (dead-short and
dead-long assets in their terminology) were optimal across a range of parallel
yield curve stresses amongst the bond portfolio choices they considered. This
result was entirely consistent with Redingtons pair of immunisation equa-
tions. Haynes and Kirton attached the explicit caveat that at this stage we
would emphasise that any such combination of dead-short and dead-long
investments offers no protection against a change in interest rates which is not
uniform for all terms.93
Redingtons immunisation theory delivered a thunderbolt of much-needed
clarity to the management of interest rate risk inherent in long-term liability
business. Thirty years later, he wrote It is really extraordinary that the elemen-
tary matching principle to which I have given the rather impetuous title of
immunization should have escaped the profession for so long.94
For life offices, however, it did not resolve the question of which liabilities
it ought to be applied to. In particular, in the context of with-profit business,
should an immunisation strategy be pursued that immunises liabilities inclu-
sive of anticipated future bonuses, or only the guaranteed liabilities that have
accrued to date? In his paper, Redington explores, without explicitly advocat-
ing, the strategy that immunises both accrued liabilities and future bonuses
(and future regular premiums could be treated as negative liability cashflows).
This strategy of attempting to lock in all future bonuses at a with-profit
policys inception met with the disapproval of many of his peers. It did not
appear consistent with policyholders expectations of participation in future
increases in interest rates, if such rate increases should arise over the lifetime
of their policy.
In 1953, the year following Redingtons paper, two senior life actuaries,
G.V.Bayley and Wilfred Perks, presented a paper95 to the institute that advo-
cated applying immunisation only to the paid-up policy values (i.e. the sum
assured that would be provided in the event that the policyholder chose not
to pay any further regular premiums). This would mean that future regular
premiums, by being invested at the prevailing interest rate when they are
received, could participate in any intervening increase in the interest rate.
However, this failed to address the fundamental interest rate solvency risk:
under this strategy, falls in interest rates below that assumed in the premium
basis could result in future premiums being inadequate to fund the contrac-
tual liabilities.
93
Haynes and Kirton (1952), p.155.
94
95
Bayley and Perks (1953).
Haynes and Kirtons paper advocated what is perhaps the most natural
strategy: match the accrued guarantees with bonds, and invest the remain-
der of the funds in risky assets. Whilst sounding straightforward, even this
strategy was not without complications in the context of regular premium
with-profit business. For example, in the early years of a with-profit policy,
a realistic valuation of future regular premiums could exceed the present
value of the guaranteed sum assured, implying that all of the accrued assets
could be invested in risky asset classes such as equities. This was generally
viewed amongst life actuaries as being uncomfortably imprudent. However,
if a with-profit fund was considered in aggregate, as was arguably reasonable
for a pooled investment vehicle, then this strategy could result in with-profit
equity/bond allocations that were sensible and sustainable.
Finally, it is interesting to consider from a historical perspective if
Redingtons concept of matching asset and liability value sensitivities can be
viewed as an antecedent to the dynamic replication concept that underlies
derivative pricing and hedging. Dynamic hedging assumes the continuous
matching of the instantaneous sensitivities of asset values to changes in under-
lying fundamental variables. Redington noted that immunisation was not a
static solution but rather the equations define the position at a moment in
time. Their solutions change continuously [emphasis added]96
Redingtons paper presented some charts that showed how example asset
and liability values changed as a function of interest rates, showing that there
are levels of interest rate at which they are solvent (i.e. assets exceed liabili-
ties) and levels of interest rate at which they are not. The liability valuation
curves included heuristic allowances for the effects of cash maturity guar-
antees and annuity options. But he did not explicitly pursue the possibility
that the asset value curve could be altered by dynamically updating the asset
duration to match the liabilities new duration following an interest rate shift.
In the discussion that followed the Institutes meeting at which the paper was
presented, A.T.Haynes (of Haynes and Kirton) commented:
Where the asset and liability curves might cross owing to a change in the rate of
interest the position could be converted immediately by altering the assets in
such a way as to immunise the liabilities. The asset curve would then follow the
same form as the liability curve 97

96
Haynes in Discussion, Redington (1952), p.331.

97
This is an explicit description of the concept of the dynamic hedging of a

non-linear liability, over 20 years before Black and Scholes used such a con-
cept to price derivatives! However, the embryonic dynamic replication think-
ing of Redington and Haynes here is entirely in the context of asset-liability
management: it did not occur to them that such a replication concept could
be used to derive arbitrage-free prices for the options that were increasingly
being embedded in life assurance contracts. As we shall discuss later, this topic
would confound the profession for many decades to come.
Actuaries andEquities (19121948)
We saw above how both British and American life offices shifted substantially
away from illiquid mortgages and loans and moved further into liquid gov-
ernment and corporate bonds between 1890 and 1920 (the US shift start-
ing some fifteen years earlier). Over the following 30 years, another radical
change in actuarial investment thinking occurred: equities emerged as a major
asset class for British life offices. In aggregate, British life offices increased their
equity allocations from under 2 % in 1920 to 21 % by 1952.98 This time,
British and American life office investment practices diverged. The British
offices had more surplus capital and a more flexible with-profit bonus distri-
bution system than their US counterparts. These features allowed the British
offices more freedom to tolerate and manage the greater volatility associ-
ated with equity investment. The American offices were also under a more
prescriptive and restrictive regulatory system that varied somewhat from state
to state but which generally curtailed equity investment (Fig.3.8).
Nineteenth-century actuarial orthodoxy had been unambiguous: equities
were an unsuitable asset class for life assurance business. Bailey thought it
self-evident that equities were too risky for such liabilities. Other nineteenth-
century actuarial papers followed this doctrine. In 1891, Mackenzie wrote
that ordinary shares should be rigidly excluded.99
Actuaries first intellectual embrace of equity investment occurred in the
1920s. This was in part a reaction to the poor performance of the long-term
government bonds that life assurers had invested so heavily in during the
First World War. The high-inflation post-war environment had resulted in
substantial falls in long-term bond prices and consequent losses for life offices.
As noted by The Economist: From the beginning of 1916 to the end of 1920,
98
Dodds (1979), Table A.3.
99
50 25%
Equities and Property (LH
45 Axis)
21%
UK Life's Equity & Property Allocation (%)
UK CPI (RH Axis)

40 17%
35 13%
30 9%
UK CPI
25 5%
20 1%
15 -3%
10 -7%
5 -11%
0 -15%
1870 1890 1910 1930 1950 1970
Year
Fig. 3.8 UK life offices equity and property allocations and UK inflation
(18701970) (Dodds (1979))
[bond] depreciation has been enormous and unprecedented and has

wiped out nearly whole of the surpluses which would otherwise have been
available for bonuses to policyholders.
The extraordinary burst of inflation that occurred during and after the First
World War was an important factor in motivating actuaries reappraisal of
equities as a potentially useful asset class for with-profit business. The six years
from the start of 1915 to the end of 1920 produced annual CPI increases
of 13 %, 18 %, 25 %, 22 %, 10 % and 15 % respectively.100 Whilst with-
profit guarantees were written in nominal money terms, there was a convic-
tion that bonuses should be used to ensure the policy delivered a satisfactory
real return (especially for endowment business where the pay-out would be
used to secure a pension income). The impact of unexpected inflation shocks
on the performance of long-term nominal bonds in the early 1920s helped
to support the argument amongst actuaries that equities, as a real asset class,
could offer some form of inflation risk protection. As the chairman of the
Pearl put it in 1928:101
Office of National Statistics Composite Consumer Price Index.

100
Scott (2002), p.80.

101
The effects of the War in matters of finance have taught us that it may be
safer to have a proportion of our investments in first-class ordinary shares,
rather than entirely on a fixed monetary payment such as is given by gilt-
edged investments.
The increase in long-term bond yields that occurred in the early 1920s
generated disturbing falls in the market values of life office asset portfolios.
It probably did not have an adverse impact on life offices overall economic
balance sheets, however, as the assets were unlikely to be substantially, if at all,
longer than life office liabilities. In this era, actuaries were reluctant to increase
the discount rate used in liability valuation unless absolutely necessary, and
so the stated surplus positions in the early 1920s were likely to have been
a prudent interpretation of the economic reality. Later in the 1920s, bond
yields started to fall again. In the midst of economic depression and a policy
of cheap money, consol yields fell from over 4.5 % in 1925 to 3.5 % by the
mid-1930s. Whilst this resulted in a recovery in bond prices, it resulted in a
new problem: such low rates made it increasingly difficult to for bond invest-
ments to sustain historical bonus rates, although the competitive pressure to
do so was intense. John Maynard Keyness speech at the 1927 annual general
meeting of National Mutual captured this sentiment:
Policyholders have become accustomed to bonuses which presume, in most

cases a net rate of interest of not much less than 4.5% after payment of
income tax. If an attempt is made to put British Government securities on a
basis appreciably below 4 per cent net, it is certain that a powerful stimulus
will be brought into play to find more profitable outlets for invested funds [emphasis
added].102
The British with-profit policy was by this time unambiguously positioned

as a long-term investment contract. As such it increasingly competed with
other investment vehicles such as investment trusts. Investment yields and
bonus rates had become increasingly important marketing statistics for with-
profit funds both in competing with each other and with other investment
offerings. A post-tax consol yield below 3 % put in serious peril not only the
ability to make bonus distributions commensurate with a 4.5 % net return,
but also the ability to meet the guaranteed liabilities (which had typically
been set using a 3 % premium rate). In this context, the 45 % dividend yield
offered by equities was very alluring.
102
Scott (2002), quoting from Keynes (1983).
The First World War had created an unintended concentration of British

life offices assets in long-term government bonds. The post-war economic
environment and its consequent impact on both realised and prospective life
assurance investment performance created a context where life assurers and
actuaries were ready for another significant change in asset allocation profile.
A stream of research, from both within the actuarial profession and outside it,
emerged in the 1920s that provided an intellectual basis for the new openness
to equity investing. But the seeds of this thinking can be found earlier.
George May, an actuary at the Prudential in the UK, presented a notable
paper103 to the Institute of Actuaries in 1912. May had a highly distinguished
career. He was secretary of the Prudential from 1915 to 1931; he also held
significant advisory roles with the British government between 1915 and
1941. He was made a Baron in 1935. In his 1912 paper, he did not explicitly
advocate investing in equities, but his was the first actuarial paper to present
a technical justification for life offices taking on more investment risk. He
paved the way for the change in actuarial attitudes to investment that would
emerge more explicitly over the following ten to 20 years.
His rationale for increasing investment risk was twofold. First, he argued
historical analysis suggested that by careful selection, relatively high yielding
investments can be obtained which will show a margin considerably greater
than is required to cover the risk run.104 He didnt, however, provide any sta-
tistical analysis or historical statistics, claiming that I do not feel it desirable
to introduce examples into this paper; further, I am sure that it is only by
personal investigation that one is in a position to form a reliable judgment.105
The second pillar of his argument for increasing investment in riskier assets
was that risk should be measured at the total portfolio level rather than at the
individual security level. He pointed out that individually risky assets may
diversify so that, at the margin, they do not materially increase the overall port-
folio risk. He suggested that Baileys already historical investment principles
should be extended with a further law: That in order to minimise the result of
temporary fluctuations and to secure the safety of the capital in the best way,
it is desirable to spread these investments over as large an area as possible.106
Actuaries were naturally familiar with the diversification concept in mortality
risk. Indeed, it was the fundamental principle of insurance business. But this
was the first time the diversification concept had been used to argue that life
103
May (1912).
104
May (1912), p.135.
105
May (1912), p.143.
106
May (1912), p.153.
offices should be more willing to invest in high-yielding securities that are sig-
nificantly riskier than those that had been typically held by life assurers.
Mays advocacy of the adoption of greater investment risk by life offices
anticipated the post-First World War zeitgeist. This experience highlighted
the benefit of diversification (or the risk of a lack of it). Life offices were ready
to invest in something else. During the decade of the 1920s, three men in
particular had an especially great influence on British actuaries thinking on
life office equity investment: John Maynard Keynes, the famous British econ-
omist; Edgar Lawrence Smith, an American economist and investment man-
ager; and H.E.Raynes, the actuary of the Legal & General Assurance Society.
John Maynard Keynes joined the board of National Mutual, a medium-
sized British life office, in 1919 and became its chairman in 1921, holding
that position until 1938. He also had a less substantial role on the board
of another life office, Provincial Insurance, during this period. In the early
days of his involvement with National Mutual, Keynes was not yet the global
statesman he would become in his elder years. His revolutionary General
Theory of Employment, Interest and Money was still some fifteen years away
from publication. He had, however, played a prominent role as an economic
advisor at the Treasury during the First World War, and his 1920 book, The
Economic Consequences of the Peace, was a bestseller that raised his profile and
marked him out as an influential commentator on economics and politics.
Although he only ever played a part-time role there, he had a major influence
at National Mutual, especially on investment policy, where the office pursued
a distinctive strategy. In the actuarial journals of the period, National Mutual
was often referred to as Keynes office.
Throughout his adult life, Keynes was an active investor. As well as his role
at life offices, he managed money for himself and his friends, was involved
in various investment trusts, and he managed the endowment fund of Kings
College, Cambridge, where he was a Fellow. Keynes was a man of strong opin-
ions, often flavoured by a strain of iconoclasm, and he always had the courage
to back his convictions. This is reflected in his approach to investment. As
an investor, Keynes has been described as a scientific gambler.107 His invest-
ment philosophy was active and aggressive. As might be expected of a man of
Keyness breadth, his investment views transcended any specific type of asset
and he regularly took short-term positions in commodities, currencies, equi-
ties and bonds as dictated by his macroeconomic and geopolitical outlook.
He had a noticeable preference for investing in liquid securities: illiquid assets
could not support the short-term active speculation that was at the heart of
107
Skidelsky (2003), p.520.
Keyness investment style. When applied to life assurance, this was the polar
opposite of Baileys conception of life offices reaping illiquidity premia from
assets that would never be sold.
National Mutual was not one of the largest offices, but this appears to have
suited Keynes, providing him with an institution which had a certain critical
mass but which could be made more malleable to his vision than the larger
offices. His vision was of mutual life assurance as the leading savings vehicle
for the middle classes. He strongly advocated the promotion of endowment
assurance as a more savings-orientated alternative to whole-of-life assurance.
He argued that life offices should pursue an active investment policy and he
piloted such a policy at National Mutual. He viewed investment as the critical
function of a life office. In January 1922, in the first of his chairmans speeches
at the annual meetings of National Mutual, he said:
I venture to say that at the present time life assurance societies must stand or fall
mainly with the success or failure of their investment policy. The labours of the
great actuaries of the 19th century have carried actuarial science to a point
where great improvements or striking innovations are no longer likely But
investment, on the other hand, provides both more pitfalls and also more
opportunities than formerly.108
Most significantly of all, Keynes held a strong view that the investment
policy of a life office should feature material allocations to equities. In his
speech at National Mutuals annual meeting of 1928, he explained:
The arguments in favour of holding a certain proportion of ordinary shares are

to be found in the advantage of spreading the fund between assets such as
bonds, expressed in money values, and assets representing real values the
second reason is the fact they are undoubtedly under-valued relatively to bonds
after making all due allowance for risk.109
National Mutual was a pioneer of life office equity investment. Prior to

Keyness involvement, National Mutuals equity allocation stood at 3 %. By
1927, it had increased to 18 %. The other British life offices had not yet fol-
lowed suit. In 1926, the average equity allocation across the industry stood at
only 4 %. But they noticed, and in time, they followed. His heavy emphasis
on investment strategy and appetite for investment risk never sat completely
comfortably with this generation of the actuarial profession. As late as 1942,
F.C.Scott, the managing director of Provincial Insurance wrote to Keynes:
Keynes (1983), p.118.

108
Keynes (1983), pp.156, 158.

109
My conception of our responsibility is that our first primary duty is to so invest

our funds, which are the security we offer our policyholders for the fulfilment of
our contracts, that the public confidence will never be impaired. I am some-
times tempted to think that you would pervert my morals and encourage me in
my secret vice to regard the investment of our funds, not as incidental to our
main business insurance but as an equally important contributor to the
profits of the Company.110
Keynes would indeed pervert his morals, and most of the rest of the actu-
arial profession. But he was not the only important actor in this scene. Edgar
Lawrence Smith was an economist at Columbia University and a Wall Street
investment manager. His pioneering short book, Common Stocks as Long Term
Investments, was published at the end of 1924. It, perhaps inevitably, reso-
nated with Keynes, who wrote a positive review of the book in the Nation and
Athenaeum. The book also registered with the British actuarial profession, and
was referred to in the papers and discussions of the profession in the second
half of the 1920s and beyond.
Smiths book was the first significant quantitative comparative analysis of
the long-term empirical returns of equities and bonds. Smith considered the
returns that had been produced from diversified portfolios of equities and
high-quality bonds over various 20-year holding periods between 1866 and
1922. He demonstrated that the long-term returns of equities had consis-
tently exceeded those of high-quality bonds. He also showed that the income
received from equities had also tended to exceed that generated by bonds over
long-term investment horizons. Neither of these observations would neces-
sarily have been especially surprising to financial practitioners. But a more
provocative conclusion of his analysis was that equities would even tend to
outperform bonds in periods of low inflation or deflation as well as in infla-
tionary periods. It was generally accepted wisdom that equities were a real
asset class whilst bonds were money assets, and as such equities would outper-
form in times of unexpectedly high inflation and vice versa. Smith argued that
the evidence suggested that over a horizon of 20 years equities would outper-
form bonds under any economic conditions with a high degree of likelihood.
Smith produced a novel argument for why equities performed so robustly.
He argued that the corporate practice of only paying out a small propor-
tion of annual earnings as dividends and reinvesting the remainder facilitated
a compound interest effect in future earnings which allowed dividends to
be smoothed over time, and hence protected equity values from economic
110
Keynes (1983), p.80.
ownswings and deflation. Such an argument had much resonance with

d
Keynes and the British actuarial profession of the time, but the idea that
dividend policy could play such a pivotal and positive role in equity value
creation would be rejected by the financial economists that would emerge a
few decades later.
Smiths data analysis did not include any statistical calculations such as
confidence intervals or significance tests. His paper did not overly focus on
equity tail risk or the possibility that the 20-year outperformance could only
be obtained by enduring an uncomfortably volatile path. However, he did
include a short section that considered how long a period of equity underper-
formance an investor may have to suffer. He considered investing in equities
at the start of 86 different years from 1837 to 1922 and found that only four
of the 86 investment projections resulted in equities producing a negative
return over the first four years. He concluded that there was a 78 % prob-
ability of equities delivering a positive return over one year, an 87 % chance
over two years, and a 94 % chance over four years. To modern eyes, this may
appear to be a stretching use of overlapping data periods and a significant
understatement of equity risk (the assumption of a normally distributed log
annual return with a mean of 7.5 %111 and volatility of 18 % implies a prob-
ability of around 20 % that equities would produce a negative return over a
four-year period, assuming annual returns are independent over time).
Finally, it is important to note that Smiths advocacy of equities as a long-
term investment was primarily targeted at the savings strategies of individual
investors. It was expressly not intended for insurance companies where their
investments are all made to protect dollar obligations. Their sole concern is so
to conduct their affairs that they will always have available more dollars than
they will be called upon, under their commitments, to pay out, at the same
time deriving a profit above operating expenses. It is natural and right that
(they) should concentrate their attention upon promissory obligations pay-
able in a fixed number of dollars.112
But by the 1920s, for better or worse, British with-profits business had
moved substantially beyond the provision of fixed monetary amounts in the
event of life contingencies. Endowment assurance and large bonus loadings
in with-profit premium rates had furthered the evolution of life assurance into
a long-term investment vehicle where mortality protection was a secondary
feature. Perhaps perversely, the high degree of actuarial caution in the setting
of with-profits premium rates had created a vacuum that risk-taking filled.
111
Assuming a cash rate of 5 %, an equity risk premium of 4 % implies an (arithmetic) expected return
of 9 % and hence a geometric expected return of c. 7.5 %.
112
Smith (1924), p.12.
Keynes and Edgar Lawrence Smith were both outsiders to the British actu-
arial profession. H.E.Raynes, however, as the actuary of the Legal & General
Assurance Society, was in its inner sanctum. He presented two influential
papers to the Institute of Actuaries in 1928 and 1937 that advocated a greater
role for ordinary shares in life office asset allocations.113 Raynes was doubt-
less influenced by the practices of Keynes and the writings of Smith (both
are referred to either directly in his papers or in the formal discussion of the
papers that followed their presentation to the profession). But he was not an
immediate convert. As late as mid-1927, Raynes contended that consider-
ations of security excluded ordinary shares as investments for life offices.114
He underwent his Damascene conversion when he attended an
International Congress of Actuaries in Canada in 1927. He had discus-
sions with T.B.Macaulay, the actuary of Sun Life of Canada (and father of
Frederick Macaulay of bond duration fame), in which Macaulay strongly
advocated investment of a proportion of funds in common stocks as a mea-
sure to combat the trouble of a depreciation of currency [i.e. inflation].115
This inspired Raynes to develop a piece of empirical analysis of UK equity
data that was very similar in spirit to Edgar Lawrence Smiths US equity work,
if less comprehensive. The paper, The Place of Ordinary Stocks and Shares
(as distinct from Fixed Interest bearing Securities) in the Investment of Life
Assurance Funds, was presented to the Institute of Actuaries in 1928. A fol-
low-up paper, Equities and Fixed Interest Stocks during Twenty-Five Years,
provided further statistical analysis to the first paper, but no new conclusions,
and was published in 1937.
Raynes analysed the investment performances of a diversified portfolio of
UK equities and a portfolio of debentures over the fifteen-year period from
1912 to 1927. Like Smith, Raynes found that equities strongly outperformed
the bond portfolio, both in terms of capital appreciation and income. He
found this was true during periods of rising interest rates and falling inter-
est rates. Raynes concluded that equities had a significant role to play in life
office asset allocation for the two distinct reasons that Keynes forwarded in his
chairmans speech of 1928: inflation protection as per T.B.Macaulays argu-
ment; and for the more fundamental reason that they appeared good value
to the long-term investor who could ride out the higher short-term volatility
that comes with equity investment.
Given Rayness role as a leading life actuary, his paper is surprisingly light
on the risk implications of equity investment, and how equity asset allocation
113
Raynes (1928), Raynes (1937).
114
G.H.Recknell in Discussion, Raynes (1928), p.35.
115
Raynes (1928), p.22.
might be considered in the context of the fixed liabilities of life assurance

contracts. Interestingly, it was a visitor to the actuarial meeting, the eminent
economist Ralph Hawtrey, who was prepared to look beyond the quantitative
results of a relatively small data sample, suggesting that if [insurance compa-
nies] were going to take the risk of investment in ordinary shares, then they
ought to have a correspondingly larger capital.116
Perhaps with the benefit of hindsight, in reviewing Rayness papers and
their accompanying actuarial meeting discussions, a creeping sense of hubris
and complacency can be detected emerging in actuarial thinking and atti-
tudes. Early in the 1928 paper Raynes writes:
Whether greater security to the capital of a fund is given by the investment of a

proportion in ordinary stocks and shares might be considered by deductive rea-
soning from the principles of economics, but while no doubt the result of such
a process would lead to the same result, a statistical investigation will, I think,
prove more convincing to actuaries.117
This sentence creates the sense that actuaries thought themselves above the
theories and deductions of others such as economists, and instead would use
their uniquely endowed capacities of observation to judge what was right.
This sense of hubris and complacency can be increasingly identified when
considering some of the doctrines of the actuarial profession in the second
half of the twentieth century. Arguably, it started when actuaries were seduced
by the cult of equities in the 1920s.
Thus, the intellectual groundwork for greater life office equity allocations
was laid in the decade of the 1920s by the triumvirate of Keynes, Smith and
Raynes. Lewis G.Whyte, a senior Scots actuary, wrote in 1947 that by the
time Mr Raynes paper was submitted to the Institute of Actuaries in 1927,
the movement [to invest life office funds in equities] had received a large
measure of professional approval.118 But equity asset allocations in aggregate
across the British life offices remained fairly modest through the 1920s. A few
life offices such as National Mutual, Scottish Widows and Equity & Law led
the way with larger equity allocations than the rest of the industry. Equity
allocations gradually climbed during the 1930s in an environment of endur-
ing low government bond yields. The post-Second World War economic
environment then provided equity allocations with a significant further fillip.
This was driven by near necessity. The cheap money era of Hugh Dalton,
the Labour chancellor of the exchequer, drove consol yields down to 3.1 %
116
R.G.Hawtrey, in Discussion, Raynes (1928), p.42.
117
Raynes (1928), p.22.
118
Whyte (1947), p.221.
in 1945, presenting a real threat to life offices ability to meet the 3 % (after-
tax) return assumption almost universally used in with-profit premium rates
at the time. As in the 1850s, low interest rates had prompted life offices to
take more investment risk. J.B.H.Pegler, the investment secretary of Clerical
Medical presented a paper119 to the Institute of Actuaries in 1948 that argued
that this environment demanded a less risk-averse attitude to investment: if
the assets are invested, no matter how safely, to provide a yield less than that
assumed in the calculation of premium rates, the soundness of the invest-
ments is somewhat illusory.120
Pegler proposed investment principles that turned Baileys risk-minimising
ethos on its head. He suggested: It should be the aim of life office investment
policy to invest its funds to earn the maximum expected yield thereon whilst,
in a nod to George May, Investments should be spread over the widest pos-
sible range in order to secure the advantages of favourable, and minimize the
disadvantages, of unfavourable, political and economic trends. These prin-
ciples were perfectly reasonable long-term investment objectives of an invest-
ment strategy, but the disregard for how such a strategy should ensure assets
could meet guaranteed liabilities is striking. With the benefit of hindsight, it
is hard to resist the thought that actuaries were starting to surrender rigour
and prudence under the significant competitive pressures of their marketing
departments.
Valuation andSurplus, 18001952

The need for a valuation of life assurance liabilities arose with the introduc-
tion of the regular premium whole-of-life assurance policy. Level premiums
for an increasing risk implied that a portion of the premiums received in the
policys early years must be reserved to fund the larger frequency of claims
that was expected to occur in later years. A liability valuation determined how
much needed to be reserved. This was well-understood by James Dodson in
his original designs of whole-of-life assurance in the mid-1750s. What he did
not anticipate was that by the end of the century a test of whether adequate
reserves were being held would be necessary for two distinct purposes: to
determine if the fund was solvent (i.e. if it had sufficient assets to meet its
contractual liabilities); and to determine how much surplus could be distrib-
uted to the with-profit policyholders (and how the distribution should be
equitably allocated across different policyholders).
119
Pegler (1948).
120
Pegler (1948), p.180.
Through most of the near-150-year period between the Napoleonic Wars

and end of the Second World War, the prudent with-profit premium rates
of the established British life offices had led to an embarrassment of riches.
The demonstration of solvency with respect to life offices contractual liabilities
could be done with relative ease and was not a particular focus of actuarial
minds. The other side of this coin was that significant surpluses were being gen-
erated that had to be distributed to with-profit policyholders, at least in part.
With-profit funds were a form of pooled investment vehicle. There were
no set rules on how the earned surplus should be distributed between cur-
rent policyholders and the estate of the office; or, once the policyholders
allocated surplus was determined, how it should be distributed across the dif-
ferent generations of current policyholders. Questions such as these featured
prominently in the thinking of the British actuarial profession from the 1850s
onwards, and triggered some of its most impassioned debates. Fundamentally,
two distinct tasks arose in determining with-profit bonuses: determining how
much surplus had arisen over a period by measuring the changes in the values
of assets and liabilities; given the determination of the surplus, decide how
much to distribute to existing policyholders and how to distribute it amongst
them. We will return to methods for distributing the surplus at the end of this
section. First we consider how actuarial thinking developed on how the size of
surplus was determined through the valuations of assets and liabilities.
Liability Valuation (18001952)
The actuarial valuation of liabilities was, and arguably still is, the quintessen-
tial actuarial task of life actuaries. It has been a topic of much debate through-
out most of the history of actuarial thought in both life assurance and defined
benefit pensions. The second half of the nineteenth century was arguably the
period of its greatest technical development in life assurance. Throughout
this period, the liability valuation was calculated prospectively as the pres-
ent value of a set of projected cashflows. Technical debates centred around a
handful of methodology choices for the projection and discounting of these
cashflows. The major (and interrelated) valuation methodology topics were
(and, to some extent, still are!):
The treatment of future premiums in the liability valuation. All liability

valuations could be described as the difference between the present value of
the assurance claim plus related expenses; and the present value of the
future premiums due over the remaining term of the policy. But there was
significant debate over the treatment of future premiums, and, in particu-

lar, whether it was appropriate to recognise the profit margins in future
premiums in todays liability valuation.
Related to the above treatment of profit margins was the treatment of
expenses. In particular, should future expenses be explicitly modelled in the
valuation, or be implicitly allowed for in the treatment of future premiums.
Should the mortality and interest rate assumptions used in the valuation be
best estimates of future experience and market conditions; or should they
include margins for prudence; or be tied to the basis that was currently in
use for pricing the product; or that was used in pricing the product at the
time of the valuation.
Should the with-profit liability valuation include an explicit allowance for
future bonuses or only the accrued contractual obligation.
As discussed in Chap. 2, the first life office liability valuations were per-
formed by William Morgan in the 1780s. The solvency of the Equitable was
never doubted during this period, and the object of his valuations was pri-
marily to limit the pressure for faster distribution of surplus. At this time,
Morgans liability valuation basis was the same as the Equitables premium
basis (a 3 % interest rate and the Northampton mortality table). At the time
of the liability valuation, the premium basis was arguably already recognised
as being significantly more prudent than a best estimate of future experience.
The liability valuations included the bonuses already accrued, but made no
provision for future bonuses. The value of the future (gross) premiums was
deducted from the reserve. He did not explicitly reserve for future expenses.
He likely believed that the margins in the liability valuation (premium) basis
adequately allowed for them.
Over the course of the first half of the nineteenth century, a number of other
prototypical actuaries, both within the emerging group of life offices (Joshua
Milne, T.R.Edmonds and others such as David Jones, Charles Babbage and
Griffith Davies), and outside it (such as John Finlaison and Augustus De
Morgan) were actively considering all aspects of nascent life office actuarial
practice. The concept of a net premium valuation emerged as an alternative
conceptual approach to Morgans liability valuation approach by the middle
of the century (though as we discuss below, Morgans approach could be cast
as a form of net premium valuation). It is difficult to precisely date the first
conception of the net premium valuation, especially as there were no profes-
sional actuarial journals in the first half of the nineteenth century. However,
we know it was established in actuarial thinking by 1853 as it was discussed at
length in the testimonies of leading actuaries such as Finlaison and Edmonds

to the 1853 Parliamentary Select Committee on Assurance Associations.121
To understand the workings of the net premium valuation and its ratio-
nale, it is perhaps easiest to start from a consideration of the more intuitive,
gross premium valuation. We can loosely define the gross premium valuation
as a liability present value that is assessed using best estimate assumptions
for all future experience, including an explicit projection of future expenses,
and which includes the contractual premiums in the valuation. Consider a
new without-profit regular premium whole-of-life policy. The office sets the
regular premium for the contract such that it expects to generate a profit. If
the actuary performed a gross premium valuation of the liability at the point
of inception of the policy, it would therefore have a negative value: the pres-
ent value of the future regular premiums would exceed the present value of
the sum assured. Negative valuations created difficulties. Most obviously, the
contract could be surrendered and the surrender value must be non-negative.
More generally, it was argued that it was inappropriate to immediately recog-
nise the profit that was expected to be generated over the lifetime of the policy
in the current valuation of the liability.
The net premium valuation method aimed to only recognise the profit
loading in the premiums at the point at which each premium is received. The
liability valuation would be made on the assumption that the future regular
premiums would be net premiumsdefined by Bailey in 1878 as the pre-
mium which, according to the best judgement that can be formed, will suffice
for the risk but leave no margin beyond.122 The net premium would therefore
be calculated on best estimate assumptions for the future behaviour of the
risks underlying the contract, but would make no allowance for expensesthe
assumption being that expenses are adequately covered by the profit margin
in the gross premiums that is being discarded in the net p remium valuation.
By construction, a net premium valuation would produce a liability value of
zero at the inception of the policy.
There was much heated debate in the mid-nineteenth century over the
relative merits of using Morgans method of valuing liabilities using gross pre-
miums and the premium basis or valuing using net premiums and a realistic
valuation basis. This was further confused by the observation that Morgans
method could be cast as a net premium method, in the sense that the gross
premiums were the net premiums implied by the premium basis (but not by
a best estimate basis). Like any net premium valuation, Morgans approach
Parliamentary Report (1853).

121
Bailey (1878), p.118.

122
would generate a liability value of zero at inception. But it did so by put-

ting the wrong parameters into the net premium valuation formula, and this
made the progression of the reserves through the life of the policy less natural
or transparent than a pure net premium valuation (where the reserve pro-
gresses by reserving the net premium portion of the gross premiums as they
are received and the remainder is used for expenses and released to surplus).
The founding of the Institute of Actuaries in 1848 and Faculty of Actuaries
in 1856 provided an important professional infrastructure for the develop-
ment of British actuarial thought and practice. It is no coincidence that the
third quarter of the nineteenth century is one of the most fertile periods of
development in actuarial thought. In particular, their professional journals,
the Journal of the Institute of Actuaries and the Transactions of the Faculty of
Actuaries, and the practice of presenting these papers at formal professional
sessional meetings were powerful vehicles in disseminating and debating actu-
arial theory and practice. A series of Institute papers by Charles Jellicoe in
the early 1850s,123 followed by a paper by Thomas Sprague in 1863, helped
to establish the net premium valuation as the fundamental piece of actuarial
machinery for liability valuation.
The net premium valuation quickly became an article of faith for the actu-
arial profession, but, to the professions credit, its limitations were highlighted
and debated as early as 1870. The discussions of these limitations produced
unusually ill-tempered exchanges amongst actuaries. Sprague, by then a pre-
eminent actuary and vice-president of the Institute, wrote in 1870 of the
intolerant manner in which some of the advocates of the net premium method
of valuation insist upon its being the one method that should be employed in
all cases.124 Clearly, there has always been something about liability reserving
methods that ignites actuarial passions like few other things can!
The net premium valuation, in its original form, was parsimonious and
elegant. When its assumptions for interest rates and mortality rates held true,
surplus emerged from the contract in a most natural way: when a regular
premium was received, the liability value would increase by the net premium,
whilst the asset value (before the deduction of expenses) would increase by
the actual premium. Thus the profit loading in the premium would be rec-
ognised with the receipt of each premium. If the net premium portion of the
actual premium was reserved when it was received, and mortality and inter-
est rate experience was as assumed in the net premium valuation, then the
reserve would be exactly sufficient to meet claims as they fell due. But the net
123
Jellicoe (1851), Jellicoe (1852), Sprague (1863).
124
Sprague (1870), p.411.
remium valuation had some counterintuitive features. As we have seen, it

p
did not explicitly allow for future expensesthe difference between the actual
and net premium was expected to take account of those. But there was noth-
ing fundamental in the method that could capture the impact of unexpected
increases in future expectations for expenses (which would be especially prob-
lematic if the new expectation for expenses exceeded the margin of the actual
premium over the net premium).
Indeed, it might be argued that the elegance of the net premium valua-
tion could quickly unravel if anything unexpected happened. Suppose interest
rates fell after the inception of a contract. If the net premium valuation used
the old, higher interest rate, this would be imprudent, as assets could not be
expected to earn that rate. But if a new lower interest rate was used in the net
premium valuation, this resulted in a higher net premium being used in the
valuation, even though the terms of the contract and the actual size of the pre-
mium payable had not changed (a feature succinctly described by Redington
as a technical idiosyncrasy which has no counterpart in the facts125). Again,
there was a risk that this could result in using net premiums in the valuation
that exceeded the actual premiums net of expected future expenses.
A number of papers were presented to the Institute around 1870 that raised
objections to the net premium valuation concept. Some of these merely high-
lighted that there were some actuarial tasks where the net premium valu-
ation was clearly inappropriate. Sprague highlighted that in a valuation of
liabilities for the purposes of determining a transfer value from business to
another, it was manifestly unjust126 to ignore the actual premiums that were
a contractual reality of the business. He also highlighted that there were US
regulations that required net premium valuations to be made on a prescribed
basis that produced higher net premiums than the actual premiums that were
being charged in the marketplace, which he described as a manifest absur-
dity. Makeham presented a paper that went further than Sprague, arguing
that the net premium valuation approach was not even appropriate for its core
application of determining surplus distribution.127 Sprague also noted that
the superintendent of the New York Insurance Department had conceded
that valuations for Government purposes will ultimately have to be made
upon actual premiums received by the different companies.128 A firmer regu-
latory context was similarly emerging in the UK following the failure of the
125
126
Sprague (1870), p.414.
127
Makeham (1870).
128
Sprague (1870), p.417.
Albert Insurance Company in 1869the 1870 Life Assurance Companies

Act created, for the first time, a regulatory requirement for British life offices
to assess prospective reserves on life policies. The methodology, however, was
not prescribed to the same degree as in the USA.British actuaries retained
more freedom to choose how to determine their reserves, but were compelled
to disclose their methods. Thus the British life industrys freedom with pub-
licity regulatory regime was born.
The net premium valuation retained a loyal following, despite the above
criticisms. In 1878, the influential Arthur Bailey (the pioneer of actuarial
thinking on investments) presented a staunch defence of the net premium
valuation. As might be expected of Bailey, his position was sophisticated and
nuanced. He argued that the net premium valuation should be made only
with best estimates of future experience, and that the premium basis should
play no direct role. But he also argued that the implementation of the net
premium valuation method should include a comparison of the actual pre-
miums and net premiums of the contract in order to ascertain what portion
[of the actual premium] represents the real prime cost of the risk [i.e. the
net premium].129 His advocacy of the net premium valuation was explicitly
when the object is to determine the amount of divisible surplus.130
The debate rumbled on, and over the following decades a number of
practical complexities emerged that further challenged the implementation
of an elegant, parsimonious net premium valuation. The net premium val-
uation implicitly assumed that the expenses associated with the policy are
incurred uniformly over the products duration (or, more accurately, over the
premium-paying term). But practices in the acquisition of new business were
changing such that much of the expense associated with a policy was incurred
at its inception in sales commission payments. This meant that a standard
net premium valuation could result in a loss being recognised at inception,
even though the policy was profitable. A fundamental attraction of the net
premium method was that it deferred the recognition of profit, but it was
arguably now overstating this deferral by initially stating a loss.
The practice of paying significant up-front commissions for new business
sales had first emerged in Germany, and it was the President of the German
Life Institute, Dr Zillmer, who first proposed an actuarial adjustment to the
net premium valuation that would allow for the change in the expense pat-
tern.131 His proposal was first made in 1863 and it essentially allowed for an
129
Bailey (1878), p.119.
130
Bailey (1878), pp.12526.
131
Zillmer (1863).
asset to be created that recognised that an initial expense had been incurred
that would otherwise be due later in the contract. Zillmers work did not have
an immediate impact in Britain as the tendency at the time was for sales com-
missions to be paid with each regular premium received rather than at the
policy inception. But by 1870 British commercial practices had increasingly
moved towards the up-front commission model, and net premium valuations
started to generate losses on new business. Spragues 1870 paper introduced
Zillmers approach to the Institute and commended it as a very able work.132
It was broadly well-received as sound actuarial practice in Britain and the
USA, though there were reservations that it could be imprudently misused or
abused. US regulators were particularly sceptical and commuted commisions
were not admissible as assets in the US regulatory returns.
Zillmers logic could be shoe-horned into the net premium valuation
method. It detracted from one of its fundamental advantagesparsimony
but it did not deliver a fatal blow. A more serious and fundamental challenge
to the logic of the net premium valuation arose in its application to with-
profit business and its bonus system. The logic of the net premium valuation
as the natural method for recognising the emergence of surplus demanded
the use of best estimates of future experience for mortality and interest rates
in the valuation. This ensured that the surplus that emerged would simply be
the difference between the actual and net premiums (providing the experi-
ence was consistent with the best estimate used in the valuation basis). By
this time, with-profit premium rates typically included bonus loadings that
set them at significantly more expensive levels than without-profit premium
rates. The premium rates were set by using interest rate and mortality assump-
tions that included substantial margins relative to the experience that was
expected. This, in itself, was not a fundamental issue for the use of net pre-
mium valuations: if the net premium valuation was performed using best
estimate interest rate and mortality assumptions, the net premium would be
significantly lower than the actual premium and a surplus would gradually
emerge with the receipt of each regular premium. However, assuming the
experience assumptions were borne out in practice, such an approach would
result in the same cash amount of surplus emerging with each premium. This
was not consistent with the way that surplus was distributed to with-profit
policiesthe approach of adding a reversionary bonus to the sum assured
meant that an increasing cash value of surplus would be added as policy dura-
tion increased. The cost of bonus was not perfectly aligned with the premium
pattern. So, for growing with-profit businesses writing high volumes of new
Sprague (1870), p.420.

132
business, this implementation of the net premium valuation could result in

a larger surplus being recognised than was desirable. In order to defer the
recognition of this surplus, it became standard practice in the final quarter
of the nineteenth century to add loadings to the assumptions used in the net
premium valuationthe assumed interest rate would be artificially lowered
by as much as 1.5 % below the interest rate that was expected to be earned.133
Valuation practice had come full circle! This was very similar to Morgans
approach of 100 years earlier of using the premium basis to produce a bas-
tardised form of net premium valuation. The approach worked in the sense
that it produced a slower emergence of surplus that was commensurate with
the reversionary bonus system. But it worked by putting the wrong (inter-
est rate) parameters into the wrong (net premium) formula. It was now very
hard to explain what the net premium valuation was intended to mean. There
was an increasing sense amongst many in the actuarial profession that some-
thing more logical and scientific ought to be done. This movement arguably
began with Spragues 1870 paper, but it was not until 1907 that an alternative
reserving system was clearly proposed when C.R.V.Coutts, the assistant actu-
ary of the National Mutual, presented his paper Bonus Reserve Valuations.
Coutts proposed replacing the net premium valuation machinery with a
more direct and transparent valuation method. He argued that the actual
gross premiums charged by the contract should be used in its valuation,
together with explicit estimates of future expenses and best estimates of inter-
est rates and mortality. Of course, for with-profit business the presence of
bonus loadings in the premium basis would result in negative liability values,
especially at early policy durations. He proposed to address this by a contro-
versial means: he argued that the future reversionary bonuses that the office
intended to pay should be explicitly reserved for as a part of the current liabil-
ity: in the valuation balance sheet, the present value of the bonus at the rate
to be declared at a given valuation should be treated as a liability for the future
existence of the policies.134 To the layperson, this may seem a very natural
approach. But it created a great deal of tension and controversy within the
actuarial profession. There were two main concernsone technical, and one
relating to policyholder communication.
The technical concern was that the bonus reserve valuation required several
accurate estimates of the various dimensions of future experience, and the
valuation result would be highly sensitive to those assumptions. The mar-
gins in with-profit premium rates and net premium valuations were large
133
Coutts (1908), p.162.
134
Coutts (1908), p.161.
and approximate, and the notion of, for example, having to estimate interest
rates at a granularity of 1/16 of 1 % in valuations created some consternation
amongst actuaries.
The second concern related to the management of policyholders expecta-
tions: if actuaries started to explicitly reserve for future bonus as though it was
a promise, did that turn it into a promise? In the discussion at Staple Inn fol-
lowing the presentation of Couttss paper, S.J.H.W.Allin commented, they
should be very careful about doing anything which tended to give policyhold-
ers the idea that the whole of the funds of the company belonged to them
to do as they liked with It was even possible that policyholders, if their
bonuses were reduced might insist that the contract between them and the
company had been broken.135 Of course, the application of a net premium
valuation with bonus loadings in the assumptions also reserved for future
bonus, but this was an implicit approach buried in the actuarial machinery
it was believed that this opaqueness reduced the risk of encouraging inappro-
priately demanding expectations from policyholder.
The net premium and gross premium methods went on to live in a some-
what uncomfortable coexistence. The stability of the net premium valuation
was attractive to actuaries in measuring the emergence of distributable surplus
from one valuation period to the next; gross premium valuations (without
future bonuses) would be the preferred approach for testing the solvency or
absolute adequacy of the reserves to meet liabilities. Redingtons seminal 1952
paper that introduced his immunisation theory was actually a paper primarily
about reviewing these with-profit liability valuation methods. In this review,
he emphasised that the distinct questions that actuaries attempt to answer
could not be answered by the same valuation:
A valuation has two main purposes, and the fundamental difficulty is that these
two purposes are in conflict. The first and primary purpose is to ensure that the
office is solvent. The second is to allow the surplus to emerge in an equitable way
suited to the bonus system. The solvency criterion leads to a changing valuation
basis On the other hand, the pursuit of equity of emergence of surplus tends
to lead to stable valuation bases, influenced mainly by retrospective
considerations.136
Later in the paper, he elaborates further:
Surplus can be estimated only by a passive valuation policy which leaves the
valuation basis both of assets and liabilities unchanged Yet when we turn to
Coutts (1908), p.170.

135

136
the question of solvency there is no word to be said in favour of a passive policy;

it is only an active policy, paying full regard to existing (and estimated future)
experience, which has any significance.137
This provided an intellectual basis for performing distinct actuarial valua-

tions in distinct ways for distinct purposes. It was broadly representative of
the British actuarial practices that had emerged over the first half of the twen-
tieth century, although there was not universal agreement within the profes-
sion on the respective roles of the different valuation methods. The Staple Inn
discussion of Redingtons paper makes clear that there remained many life
actuaries who had decided that the net premium valuation should be regarded
as a piece of actuarial antiquity. But time has proven it to be a resilient idea. At
the start of the twenty-first century, it remained enshrined in the regulatory
solvency regimes of many countries around the world.
Asset Valuation (18291952)
Nineteenth-century actuarial thinking on valuation was resolutely focused on

the liability side of the balance sheet. Actuaries generally took asset values as
they were given to them by the life offices investment department. This started
to change in the mid-twentieth century, most notably in the late 1940s and
early 1950s, when a number of sophisticated actuarial papers were published
discussing how assets should be valued for the purposes of measuring solvency
and surplus. The increased holdings in risky asset classes such as equities and
property, together with the substantial movements in interest rates that were
experienced in the 1930s and 1940s, were both significant in motivating this
new actuarial focus on the asset side of the balance sheet.
Early nineteenth-century actuaries were aware that asset valuation could
play an important role in the assessment of surplus. William Morgan, in his
address to the Equitable general court in 1829, stated:
That the Society may probably have the right to estimate its Stock in the Public
Funds at the price it bears at the time of determining the surplus, I do not pre-
tend to dispute The great fluctuation in the value of the funded property
[long-term government bonds] will always render the surplus very precarious. It
might perhaps be desirable that the Stock in the Funds should be invariably
fixed at some mean price, rather than that the amount of the surplus should be
suffered to depend on property perpetually changing its real value.138
137
138
Ogborn (1962), p.202.
This is an interesting early example of actuaries disdain for market prices.

But it is important to understand the context of this statement. Morgan
was specifically considering the determination of surplus that had emerged
between two valuation dates for the purpose of setting the bonuses that would
be distributed to with-profit policyholders. His comment is essentially a
reflection on the amount of bonus smoothing that he believed was desirable
to have in with-profit distributions over time and between generations. This
need not have any direct implication for his view of whether market values
were a sound indicator of economic worth or whether or not they should be
used when considering the absolute solvency of life offices.
The actuarial preference for the asset side of the balance sheet to use some-
thing more stable than market values was well-established by the early twen-
tieth century. The standard practice was to hold the lower of an assets book
value and market value, a practice summarised as always write down, never
write up. A rationale for this approach can be found in R.K. Lochheads
1930s textbook for actuarial students:
The established practice of British Offices is to write down security values when
prudence dictates such a course and only in very exceptional circumstances to
take credit for capital appreciation apart from realized profits on maturity or
sale. The justification for this procedure lies in the necessity for creating a reserve
against adverse movements in the value of assets, so endeavouring to secure that
the life assurance fund shall be represented by assets of equal value.139
Lochheads justification is notably focused on solvency rather than the

equitable distribution of surplus. It is also interesting to observe that his jus-
tification can be construed as a prudential solvency approach where assets are
valued at market value less a reserve for risk. More generally, a number of pos-
sible justifications were forwarded in support of the always write down, never
write up approach, relating to both measuring surplus and ensuring future
solvency: it produced a prudential valuation (any excess of market value over
book value was a hidden reserve); it could achieve more consistency with
the net premium valuation of liabilities (which tended to use sticky interest
rate assumptions that would only be revised when absolutely necessary); it
smoothed the release of surplus to policyholders over time and produced a less
volatile public balance sheet.
As mentioned in the liability valuation discussion above, a gross premium
valuation became the standard actuarial practice for the internal assessment
Lochhead (1932), p.66.

139
of the solvency position of with-profit funds. In that context, and in the con-
text of surplus assessment using a bonus reserve valuation, the use of market
values for assets was more typical, and was increasingly viewed as the natural
valuation approach to take in such cases, as was argued by T.R.Suttie in his
important Institute paper of 1944:140
There does not seem to be any object in making a bonus reserve valuation unless
it is based upon the true facts, i.e. the actual premiums to be received, the actual
renewal expenses, the experience rate of interest, and the market value of assets.
Hidden reserves seem entirely contrary to the principles of a bonus reserve valu-
ation The position is quite different under a net premium valuation. The
valuation basis being more or less fixed, hidden reserves are necessary, not only
to avoid violent fluctuations in the rate of bonus if appreciation or depreciation
occurs, but also to enable the results of the published net premium valuation to
correspond with those of the unpublished bonus reserve valuation.141
The net premium valuation was still used widely, both in measurement of
surplus for bonus determination, and in the published regulatory returns of
life offices. The always write down, never write up asset valuation method was
invariably used with net premium valuations in the first half of the twentieth
century. As the asset and investment thinking of the profession developed in
the 1940s, the justifications for its use came under inevitable scrutiny. It made
the relative treatment of assets and liabilities somewhat opaque, particularly in
the context of volatile long-term interest rates. In the event of a general fall in
the level of interest rates, assets book values could substantially understate the
market value of long bonds, creating a substantial hidden reserve. Meanwhile,
the actuary would be under pressure to reduce the valuation interest rate
used for the liabilities to reflect the reality of the new market prices. But
the assumed valuation interest rate was at the discretion of the actuary and
need not be directly related to the market yield. The net impact of the rate
fall would therefore be confused or even arbitrary. Similarly, in the event of
an interest rate rise, the assets would be written down to their new market
values, and the actuary would have some choice over whether this was offset
by a reduction in the size of the estate or by increasing the liability valuation
interest rate. It was the role of actuaries to use their judgement in these choices
so as to meet their twin objectives of ensuring the solvency of the fund and
delivering an equitable distribution of surplus to different generations of poli-
cyholders (and holders of different types of with-profit policy). But it is easy
140
Suttie (1944).
141
Suttie, in Discussion, Suttie (1944), p.226.
to see how a combination of a liability valuation that assumes premiums that

have no direct relationship to the actual premiums that will be received and
that uses an interest rate that is not directly related to that which is expected to
be earned, together with an asset valuation that bears no direct relationship to
current asset market values, could make non-actuaries somewhat suspicious
of these actuarial dark arts.
There was a more specific issue with the logic of the always write down,
never write up asset valuation approach: if a reserve against adverse asset
movements was desirable, as per Lochheads textbook, why was the difference
between market values and book values the right size for this reserve? If such
a reserve was important, should a more risk-based and scientific approach not
be used to determine its necessary size? This line of argument was developed
in a couple of papers presented in 19471948Peglers Institute paper that
arose in the discussion of equity investment above; and a Faculty paper writ-
ten by the investment actuary Lewis Whyte.142
Peglers 1948 paper that advocated the maximising of expected investment
returns also advocated what was, for the time, a novel approach to dealing
with equity valuation and the distribution of surplus arising from equity
investment. He argued that equities should always be valued at market value,
dismissing book values as illogical and based on fallacies.143 The recogni-
tion of the surplus created by an increase in market values could be partly
deferred by creating an explicit reserve for depreciation. This reserve would
be set based on an assumption about how far equity market values could
fall. Pegler s uggested an assumed fall of 50 % in equity market values might
be a reasonable basis, though he conceded the choice would be subjective.
He argued for a similar approach to be applied to fixed interest securities,
and suggested a long-term interest rate rise of 0.5 % would be a reasonable
assumption for determining the depreciation reserve. The sizes of these two
stresses are interesting: both were plucked out the air rather than being the
result of any exhaustive statistical study, but both are reasonably consistent
with the range of stress sizes applied to equities and long-term interest rates in
twenty-first-century insurance solvency assessments.
In his paper, Whyte argued for an asset valuation approach that had a simi-
lar effect to Peglers market value plus depreciation reserve approach. Whyte
proposed that assets should be valued at a notional value, which was defined
to reflect its fair value less an appropriate margin for adverse price fluc-
tuations.144 In the context of setting a notional price for equity holdings he
142
Whyte (1947).
143
Pegler (1948), pp.19293.
144
Whyte (1947), p.235.
wrote it would be helpful to have in mind a price below which the security in
question would be considered as unlikely to fall,145 though he did not venture
any quantitative suggestions. For long-term government bond yields, he, like
Pegler, suggested a margin of 0.5 % (Peglers paper came after Whytes and
Peglers assumption was influenced by Whytes).
Both these approaches are notable for advocating a prospective risk-based
asset reserve rather than an arbitrary hidden reserve based on historical
transaction prices. This concept was fairly revolutionary, and inevitably was
resisted by the more traditional thinkers. The depreciation reserve approach
also required difficult assumptions to be made about how much depreciation
would be reasonable to reserve for. A parallel to the net premium valuation
versus bonus reserve valuation debate can be observed hereactuaries were
reticent to adopt new methods that required them to make new, difficult and
explicit assumptions relative to the more opaque requirements of their estab-
lished methods.
The application of the depreciation reserve concept to bonds was especially
confusing. Actuaries understood that from an asset-liability perspective they
were often more exposed to falls rather than rises in interest rates. So it did
not make sense to hold a reserve based on how much bond values would fall
if rates went up. These were teething issues that would be ironed out in the
coming decades. Pegler and Whytes contributions should be recognised as
important groundwork for the later development of solvency capital systems
that were based on the assessed riskiness of assets market values.
The Equitable Distribution ofSurplus (18001944)
Above, we identified that, historically, a central purpose of the actuarial valu-

ations of assets and liabilities has been to determine the amount of surplus
that emerged between two valuation dates, and hence what was available to be
distributed to with-profit policyholders as bonus. The measure of the emer-
gence of surplus was not necessarily intended to most accurately capture the
economic profit that had emerged over the period. There was a (sometimes
implicit) objective to distribute more or less than the amount of profit that
had emergedto attempt to smooth volatile returns over time, or simply
because there was a view that some surplus should be retained as a contribu-
tion to the estate. At the Equitable, William Morgan had established in 1800
that a maximum limit of two thirds of the surplus that had been assessed
to have emerged between valuations could be distributed to policyholders.
145
Whyte (1947), p.236.
But this was never a universally accepted standard and a number of actuar-
ies felt that the Equitable had inequitably under-distributed bonus. Charles
Jellicoe, the first editor of the Journal of the Institute of Actuaries, argued in
1851 that the whole of the earned surplus should be distributed to the poli-
cyholders, reasoning The practice of reserving a part of it seems unnecessary,
for there is not only the excess included in the future premiums, but very
generally a subscribed capital to fall back upon, beside the fund reserved for
the liability.146
How much of the surplus should be distributed to the current generation
of policyholders, and how it should be distributed amongst those policyhold-
ers, were decisions primarily driven by a consideration of what was equitable.
The concept of policyholder equity has long been amorphous and ambigu-
ous, especially for British life offices. What exactly was meant by equity in
the context of a pooled system of investment such as with-profits? When
unexpected events happened, how much cross-subsidy between the lucky and
unlucky policyholders was equitable? As R.H.Storr-Best put it in 1962: The
problems [with equitable surplus distribution] that have arisen result from
the conflict of these two ideason the one hand, the averaging of experience
and on the other the return to the policyholder of what is his due. The inter-
pretation of equity depends on the emphasis placed on each idea.147 Thus
policyholder equity was a matter of interpretation, precedent and, ultimately,
actuarial judgement.
As well as considering how much bonus should be distributed to each poli-
cyholder, the form the bonus took was another important element in s urplus
distribution. Chapter 2 discussed how William Morgan, in accordance with
Richard Prices advice, pioneered the distribution of surplus through an
irreversible addition to the sum assured (instead of by cash dividend). This
practice was adopted by the significant majority of British with-profit offices
during the nineteenth century, but not universally. Of 59 British life offices
writing with-profit business in 1870, 47 were found to be distributing surplus
by reversionary bonus, with the remaining twelve doing so by cash dividend
or by abatement of future premiums. Around half of the offices distributing
by reversionary bonus did so by a uniform rate applied to all with-profit poli-
cyholders sums assured. This was clearly the simplest mode of bonus, but it
constrained the amount of variation in the relative size of bonus that could
be distributed to different policyholders. The remaining half varied the rever-
sionary bonus according to age and/or the duration of the contract.148
146
Jellicoe (1851), p.26.
147
R.H.Storr-Best in Discussion, Cox and Storr-Best (1962a), p.40.
148
Cox and Storr-Best (1962a), p.47.
By the third quarter of the nineteenth century, the equitable distribution

of surplus had become one of the most active areas of actuarial research. It
featured heavily in the early volumes of the Journal of the Institute of Actuaries,
and was a favoured topic of Charles Jellicoe, its first editor (Jellicoe went
on to become President of the Institute between 1860 and 1867). Jellicoe
held the view that bonuses should be distributed in close accordance with
the profit that had been contributed by the specific policy: The surplus then,
being satisfactorily ascertained, it appears only consistent with justice and
proprietary that it or its value, as exactly as possible, should be returned to the
contributors.149
This view was concurrently becoming an established principle in the
USA.The US with-profits system was still nascent at this time. Considerable
freedom therefore existed to devise a surplus distribution system without
the encumbrance of precedence or pre-existing policyholder expectations.
Sheppard Homans, the chief actuary of Mutual Life Insurance Company of
NewYork, one of the largest US life offices, played a leading role in the devel-
opment of the theory and practice of equitable surplus distribution in the
USA.Homans would go on to have a distinguished actuarial career, becom-
ing the first President of the Actuarial Society of America in 1889, and, as
we saw earlier in this chapter, he published the first mortality table based on
US life office experience in 1868. He was invited to present a paper to the
Institute of Actuaries on his thinking on surplus distribution. His paper, On
the Equitable Distribution of Surplus, was published in the Journal in 1863.
Homanss paper starts by noting that the experience of the previous five years
at the Mutual Life of NewYork had been very profitablemortality experi-
ence had been around 70 % of the Carlisle table (and the table approximated
to the Mutual Lifes premium mortality basis), whilst investment returns had
been around 6.5 % compared to a premium basis of 4 %. He noted that this
size of surplus was unexpected and could not be expected to regularly recur.
Homanss approach to determining how surplus should be allocated amongst
policyholders started with the principle that each participant should be ben-
efited in proportion to the excess of his payments over and above the actual cost
of insurance. Homans did not dwell on this objective, writing that it was seem-
ingly recognised by all actuaries.150 Whilst that was something of a simplifica-
tion, his principle was consistent with the views published by Charles Jellicoe
over the preceding decade.
149
Jellicoe (1851), p.26.
150
Homans (1863), p.122.
Homanss principle still left plenty of latitude. Any insurance relies on pool-
ing and cross-subsidy, so the calculation of an actual cost of insurance expe-
rienced by an individual depends on a definition of what risks are pooled by
whom over what period. The bonus allocation system advocated by Homans
involved splitting the life fund into separate sub-funds with each sub-fund
including a cohort of policies of lives of the same age of policyholder and same
policy duration. The cost of mortality insurance was assessed within each of
these sub-funds based on the mortality experience of the cohort. A surplus
calculation was then made for each based on this assessed cost of insurance,
together with the funds earned rate of interest and the change in the cohorts
liability reserve over the assessment period. He did not suggest any smooth-
ing of the mortality insurance cost over time or any cross-subsidy between
cohorts: if the experienced mortality rate in a given year of the cohort of 50
year-old policyholders with a policy duration of two years was half of that
anticipated in the premium basis, and the experienced mortality rate of 61
year-old mortality rate with policy duration of one year was the same as that
anticipated in the premium basis, the first cohort would receive a contribution
to dividend from this particular source whereas the second cohort would not.
Homanss contribution method was first implemented by himself at Mutual
Life in the 1860s, and was then widely adopted by the US life offices. It did
not have a direct impact on bonus practices at the British offices at the time.
The reversionary bonus system had never aimed to deliver the granularity of
cohort-specific surplus distribution that Homanss system achieved. Over the
subsequent decades, the British method of bonus distribution moved further
towards the uniform reversionary bonus system, and hence further from a
method that could deliver the level of granularity in surplus distribution that
was implied by the contribution method (and that was present in the USA).
The British actuarial profession, however, continued to wrestle with its own
interpretation of equitable distribution of surplus over the following decades.
The contribution concept constantly resonated in the professions collective
conscience as an important reference point during these deliberations.
Surplus distribution was an ongoing topic of actuarial thought through
the final quarter of the nineteenth century. H.W.Andras, the actuary of the
University Life Assurance Society, presented a notable paper to the Institute in
1896 that shone further light on the performance of British bonus methods in
the context of the contribution concept.151 He analysed the uniform reversion-
ary bonus system and its practice in Britain at the time, considering how well
it could achieve equitable outcomes for policyholders. Andras concluded that:
Andras (1896).
151
This system of bonus distribution although having the advantage of simplic-

ity, is of too rigid a nature to be suitable to the varying circumstances of life
offices, if we regard as important the equitable distribution of profit amongst those
who have contributed to it [emphasis added].152
Andrass conclusion was partly based on the difficulty of using a uniform

bonus rate to distribute profit equitably when the in-force policies had been
written at different premium rates. This was one of the reasons why British
with-profit rates were subsequently so sticky through the first half of the
twentieth century, despite the substantial changes in interest rate environ-
ment over that period. However, Andras balanced his conclusion by noting
that in offices experiencing average results it [the uniform compound
bonus rate] would give bonuses approximately the same as the contribution
method, so that, in such cases, the merit of simplicity may be combined with
the principles of equity.153
As ever, the record of the Staple Inn discussion of the paper provides
an insight into the state of thinking amongst the leaders of the profession.
G.F.Hardy argued:
[The American contribution method] was a highly scientific plan, but on the
whole was not practicably workable at all events in this country the main
objection being that the whole of the mortality fluctuations at individual ages re-
appeared in the shape of fluctuations in the bonus In the present state of our
knowledge it was not known what the true rates of mortality were, and that was
a great argument in favour of adopting a simple method like the compound rever-
sionary bonus method, instead of the much more complex contribution plan.154
The above comments highlighted the fundamental tensions: the large

amount of calculation and complexity the administration of the contribution
method entailed was unappealing to British actuaries, but they were natu-
rally anxious that their uniform bonus rate approach would not be so blunt
a tool as to be self-evidently inadequate for delivering equitable treatment
amongst policyholders. Furthermore, the British system embraced cross-
subsidy between current policyholders and across policyholder generations
in a way that US with-profits had never donestability in bonus rates was a
fundamental feature of the British with-profit system.
152
Andras (1896), p.359.
153
Andras (1896), p.359.
154
In Discussion, Andras (1896), p.368.
The challenges of ensuring an equitable distribution of surplus further

increased for the British system of with-profits in the twentieth century, as
with-profits business was increasingly run on steroids. By the second quarter
of the twentieth century, bonus loadings in British with-profit premium rates
tended to be significantly larger than elsewhere in the world. The bonus load-
ings implied by the difference between with-profit and without-profit pre-
mium rates tripled between 1870 and 1940.155 And as was discussed in this
chapter, with-profit funds allocations to risky assets such as equities and real
estate increased significantly over the second quarter of the twentieth century.
Obtaining equity amongst policyholders was further complicated by changing
product mixes within with-profit funds (with endowment assurance becom-
ing increasingly popular relative to whole-of-life assurance). These features
ensured that the equitable distribution of surplus remained one of the major
topics of British actuarial thought during the first half of the twentieth century.
Alongside these changes in with-profit funds bonus loadings, investment
strategy and product mix, with-profit business was also exposed to exceptional
economic variability in the first half of the twentieth century, particularly
in the aftermath of the First World War. With-profit funds tried to main-
tain very stable with-profit premium rates through this period, even in the
presence of unprecedented gyrations in long-term interest rates. This meant
that the bonus-earning ability of policies written at different times could be
substantially different. The historical priority given to stable bonus rates over
contribution equity became an increasing concern, as was tactfully noted in
Lochheads actuarial student textbook:
The question has been raised whether the subordination of other considerations
to the stabilizing of bonuses is always perfectly fair as between policyholders of
different generations.156
In his presidential address to the Institute of Actuaries in October 1943,

H.E.Melville questioned whether current methods were generating equitable
distributions of surplus. This prompted a number of papers on the topic to
be presented to the Institute and Faculty in the subsequent few years, the
most important of which was T.R.Sutties The Treatment of Appreciation or
Depreciation in The Assets of a Life Assurance Fund.157
155
Cox and Storr-Best (1962a), p.66.
156
Lochhead (1932), p.75.
157
Suttie (1944). See also Anderson (1944).
Sutties objective was to discuss how appreciation or depreciation [in fixed

interest securities] should be treated in order to obtain the maximum practi-
cal degree of equity between the different generations of policyholders in an
office which distributes by means of a uniform reversionary bonus.158 His
analysis included a historical review of the twentieth-century performance of
the always write down, never write up asset valuation practice from the per-
spective of policyholder equity. He focused on two periods of volatility. First,
in the inflationary aftermath of the First World War:
During the war of 191418 and the years immediately following, there was a
very substantial rise in the level of interest rates and consequent depreciation of
investments. It would appear that in many cases life offices wrote down their
assets to their new market values and continued to make their valuations on the
same net premium basis as had been used in the past this procedure benefited
the more recent policies at the expense of the older. Further, if the level of pre-
miums charged for new policies remained unaltered, it benefited policies enter-
ing after the depreciation occurred, while if the premiums were reduced on
account of the increase in the rates of interest obtainable, new policies were even
more favourably treated.159
And second, the impact of the Great Depression:
In 1932 there was a reduction in the level of interest rates and consequent
appreciation in investments, but it would appear the majority of offices did not
write up the assets to their new market values so that the difference between
book values and the market values constituted a hidden reserve which was not
made available for distribution in bonuses. This again benefited the more recent
policies at the expense of the older policies, and greatly benefited the policies
entering after the appreciation unless the premiums for new policies were sub-
stantially increased.160
Suttie particularly focused on the high interest rate scenario. He considered

how current bonuses and assumed future bonus rates (in the context of a
bonus reserve valuation) should be impacted by a substantial rise in interest
rates, given the objective of ensuring equity between different generations of
policyholders. He used a model office to examine how different allocations to
current and assumed future bonus rates impacted on the different policyhold-
ers in his office. He concluded:
158
Suttie (1944), p.203.
159
Suttie (1944), p.214.
160
Suttie (1944), p.214.
The most equitable method of dealing with [bond asset] depreciation under a
uniform reversionary bonus system is to write down the assets to their new mar-
ket values and to increase the valuation rate of interest to that obtainable under
the new conditions, valuing as a liability a future rate of bonus at the rate which
the existing premium scale will support under the new conditions If these
arguments are accepted, depreciation will involve a decrease in the rate of bonus
declared in respect of the valuation period during which the change in value
occurs while, if conditions again become stable, future bonuses will be at a
higher rate than in the past.161
But he went on to note that Such abrupt changes in the level of bonus
rates are contrary to the usual practice of offices and it would undoubtedly be
difficult for one office to follow such a policy.162
Sutties analysis was essentially showing how a contribution approach to
bonus distribution would behave through time in an interest rate stress sce-
nario. Perhaps predictably, the discussion following his presentation of the
paper cleaved to the more conventional route of setting the current and
future bonus rates at equal levels that would result in a reduction in the
bonus reserve valuation that absorbed the asset depreciation. In essence, his
peers advocated spreading the recognition of the market value impact of the
interest rate increase over the full run-off of the existing business. Concerns
were voiced over the variability of bonus that would be implied by Sutties
proposed approach:
He did not know what would be said by the holder of a policy nearing maturity
if, just before maturity, there were a fall in prices and, as a consequence, a reduc-
tion of the sum payable at maturity. He felt that such a type of distribution
would be bound to cause dissatisfaction.163
Stability in bonus rates still trumped all else in British actuarial thinking on
surplus distribution. Unlike in the United States, cross-generational smooth-
ing had become a fundamental feature of with-profits. This cross-subsidy
could apply not only between different generations of existing policyholders,
but also between policyholders and the estate, as was expressed at the Staple
Inn meeting:
161
Suttie (1944), p.213.
162
Suttie (1944), p.213.
163
M.E.Ogborn, in Discussion, Suttie (1944), p.220.
With most offices, part of the profits of the past had been used to create contin-
gency reserves which should be drawn upon to deal with the depreciation of the
assets the aim should be to smooth out the effect of the depreciation by declaring
a current bonus at a reasonable level and by rebuilding the contingency reserves
gradually in the future [emphasis added].164
The British actuarial professions interpretation of equitable bonus policy

remained much broader, and its implementation more heuristic, than the con-
tribution method that dominated bonus methods in the USA.This was partly
a function of legacy: the British life offices, arguably and perhaps ironically
due to a lack of policyholder equity in the past, had accumulated significant
free reserves in their funds. These estates permitted the actuary a considerable
latitude in bonus policy. The inevitable arbitrariness entailed in deciding what
to do with the estate provided another reason not to be overly meticulous
about how to distribute the surplus generated by existing policyholders. The
British approach relied heavily on the judgement and expertise of the actu-
arial profession, and the publics trust in it. In that sense it was perhaps an
approach that reflected the time and place of Britain in the first half of the
twentieth centurya class-based society that placed unquestioning trust and
deference in its institutions and professions. Times would change.
164
L.Brown, in Discussion, Suttie (1944), p.218.
4
A Brief History ofFinancial Economics
forActuaries
This chapter gives an historical overview of the emergence of some of the key
ideas in financial economics (the branch of economics concerned with fields
such as securities pricing, portfolio theory, corporate financial and investment
theory and the behaviour of financial markets). Financial economics was not
developed by actuaries. You may well ask why it is included in a history of
actuarial thought. The answer is that, rather like probability and statistics,
once its foundations had been developed, its ideas had important practical
application in the world that actuaries occupied. As these ideas emerged, the
actuarial profession had to determine how to incorporate the new insights
provided by financial economics into its thinking and practices (and, indeed,
where theory did not translate into practice). This was not an easy process, as
these insights were often incongruent with the traditional actuarial perspec-
tive. It triggered great actuarial debate: was this incongruence merely a result
of the theoretical and unworldly nature of the economic studies, or did it
signal that some key areas of actuarial thought needed fundamental revision?
This process is arguably still underway. It is one of the more interesting and
fundamental aspects of the historical development of actuarial thought in the
second half of the twentieth century. To really appreciate it, we need to under-
stand something of the ideas of financial economics and how they originally
developed, and this is the object of this part of the book.
Our historical survey of financial economics focuses mainly on the remark-
able quarter of a century of activity bookended by the seminal papers of Harry
Markowitz in 1952 and Oldrich Vasicek in 1978. This period saw the devel-
opment of a handful of interrelated big ideas in financial economics that

DOI10.1007/978-3-319-33183-6_4
radically altered theoretical understanding and business practices in areas such

as investment portfolio management, corporate finance, asset pricing and risk
management generally and derivatives pricing and hedging in particular. We
also note the important empirical research of the 1980s that challenged the
classical financial economic theories of the earlier decades and pointed to a
richer, more complex economic reality.
The historical summary below is not intended to be an economics text-
book. Rather it is meant to provide an overview of how and when the essence
of the key ideas in financial economics emerged (particularly the ones that
were potentially of practical relevance to actuarial thought). The discussion of
how actuaries did and did not integrate these ideas into their thinking arises
more fully in the later parts of the book, which focus on post-war British
actuarial thought in life, pensions and general insurance.
Modigliani andMillers Irrelevancy Propositions

Much of early financial economics considered what possible causes could and
could not have an effect on the rational behaviours of individuals and firms in
well-functioning markets. A number of irrelevance and separation theories
followed where things that were intuitively believed to be important in mat-
ters such as corporate financial structure, corporate investment policy or indi-
viduals portfolio composition were shown to be theoretically irrelevant to all
or specific parts of a rational decision-making process. An early and important
example of such a theory was provided by the Yale economics professor Irving
Fisher in his seminal 1930 book The Theory of Interest.1 Fisher substantially
pre-dates the recognised classic period of financial economics that started in
the 1950sThe Theory of Interest arguably provides a bridge from macroeco-
nomics to the fledgling discipline of financial economics (that is, to the formal
and rigorous economic treatment of the fields of investment and finance).
For financial economists, Fishers book is best remembered for what has
become known as Fishers Separation Theorem. This theory states that a firms
value is maximised when it chooses to invest in projects that generate cash-
flow streams with positive expected present values when discounted using
market discount rates. This may appear as an unremarkable statementbut
the key point was that the choice of discount rate should be determined
only by the markets required return and should be independent of the par-
ticular risk and timing preferences of the firms specific set of shareholders.
These preferences are irrelevant to the shareholders own wealth because, in
Fisher (1930).
1
4 A Brief History ofFinancial Economics forActuaries 147
ell-functioning markets, shareholders would be able to take actions (buying

w
stock, selling stock, borrowing, lending) that would deliver to them whatever
risk and timing preferences they wanted, whilst preserving the value created
by the firms investment decisions. Hence, all shareholders wealth would be
maximised when the firm chose to invest to maximise the present value at
market discount rates, even if it resulted in a risk or income timing profile
that was not desired by some, or even all, of the firms actual shareholders. The
firms optimal choice of investments was therefore separate from the risk and
income-timing preferences of its shareholders.
Fisher was not the only US economist of the 1930s who was focused on
corporate investment theory and the valuation of the firm. In 1938, John
Burr Williams published his Harvard doctoral thesis, The Theory of Investment
Value.2 According to Mark Rubinstein, the eminent late twentieth century
financial economist and occasional historian, Williamss book contains what
is probably the first exposition of the ModiglianiMiller (1958) proposition
on the irrelevancy of capital structure.3 This principle is set out by Williams
in his Law of Conservation of Investment Value:
If the investment value of an enterprise as a whole is by definition the present

worth of all its future distributions to security holders, whether on interest or
dividend account, then this value in no [way] depends on what the companys
capitalization [financial structure] is.4
Thus Williamss Law of Conservation of Investment Value said that the

total value of the firm was determined by the cashflows its assets generate,
and this was not affected by how those cashflows were split up and packaged
into various forms of securities such as equities and debt. Williams never set
out formally to prove his law: his justification was based on rhetorical per-
suasion rather than a rigorous demonstration deduced from a set of stated
axioms. He reasoned that, Clearly if a single individual or single institutional
investor owned all the bonds, stocks and warrants issued by a corporation,
it would not matter to this investor what the companys capitalization was.5
Some 20 years later, ModiglianiMillers famous paper was published in
which Williamss argument was developed using a more formal economics.6
Indeed, the ModiglianiMiller paper was arguably as important to financial
2
Williams (1938).
3
Rubinstein (2006), p.79.
4
Williams (1938), p.72.
5
Williams (1938), p.72.
6
ModiglianiMiller (1958).
e conomics for how it influenced the way in which its theories would be devel-
oped as for its specific content. In particular, ModiglianiMiller pioneered
the formal use of no-arbitrage as a way of proving a financial theory. That is,
the proof was based on the demonstration that if the theory did not hold, an
arbitrage would exist such that some investors would be able to make limit-
less riskless profits at the expense of others: if it was an assumption of well-
functioning markets that such opportunities do not exist, then the theory
must hold in such a market. Virtually every notable theory of financial eco-
nomics published ever since has used a no-arbitrage argument in its formula-
tion. ModiglianiMillers version of Williamss Law was given in their famous
Proposition I:
The market value of any firm is independent of its capital structure the aver-
age cost of capital to any firm is completely independent of its capital structure
and is equal to the capitalization rate of a pure [unleveraged] equity stream of its
class.7
ModiglianiMillers no-arbitrage argument was built on the assumption

that a group of firms in a given class had perfectly correlated underlying
assets. From this assumption, it was possible to show that different segments
of the capital structure of the different firms in the class could be bought or
sold to create an arbitrage if Proposition I did not hold. This is, of course,
a somewhat artificial and theoretical arbitrage argumentin reality there
are not many groups of firms with perfectly correlated assetsbut it pro-
vided their argument with a new economic rigour. Like Williamss Law of
Conservation of Investment Value, Proposition I said that the total size of
the pie (the value of the firm) does not depend, in a well-functioning capital
market, on how you split up the pie (between shareholders and bondholders
and any other form of security holders). In essence, if we assume that the
firms asset cashflow is independent of its capital structure choice, then this
is really only saying that present values are additive. For example, in a simple
single-timestep model where the firms (uncertain) asset cashflow X, is paid at
the end of the period, and a debt of amount D is due to bond holders at this
time, the respective pay-outs are:
Shareholder pay- out = Max ( 0, X D )

Bondholder pay-out = Min ( X, D )

Total pay-out = M ax ( 0, X D ) + Min ( X, D ) = X

ModiglianiMiller (1958), pp.26869.

7
The sum of the pay-outs is always X and the present value of the sum of the
pay-outs is not a function of the choice of value for D. Proposition I also
said that the choice of capital structure had no impact on the firms total cost
of capital (and hence on corporate investment choices): the cost of capital
was solely a function of the riskiness of the investments that the firm made
(the asset side of the corporate balance sheet), and this was not altered
by the firms capital structure (the liability side of the corporate balance
sheet). In a similar vein to Fishers argument, ModiglianiMiller said
firms create value through the investments they make; these investments
should be discounted at market discount rates commensurate with the
risk of the investments; the firms capital structure and the individual
preferences of their capital owners were both irrelevant to these choices
and values. The behaviour of the firms cost of capital as a function of the
level of debt in the capital structure according to Proposition I can be
summarised by Fig.4.1 below.
The cost of capital is constant across the choice of capital structure, and is
therefore always equal to the cost of capital of a pure equity (unleveraged, zero
debt) firm. As debt is introduced into the capital structure, three things happen
that have a collectively offsetting impact on the firms weighted-average cost of
capital: the value of debt as a proportion of total assets increases; the expected
15%
12%
9%
6%
Weighted Average Cost of Capital

3% Expected return on equity
Expected return on debt
Risk-free rate
0%
Market Value (Debt) / Market Value (Total Assets)
Fig. 4.1 Cost of capital and capital structure example (ModiglianiMiller proposi-
tion I)
return on debt increases (starting from the risk-free rate); and the expected
return on equities increases (starting from the cost of capital of an unleveraged
firm). In the limit, very high debt levels make default increasingly more likely
and debt accordingly behaves more and more like (unleveraged) equity.
ModiglianiMiller did not say anything about how to value the respective
shareholder and bondholder claims, or how the expected return on equities
and debt specifically behaved. Their valuation statements were only relative:
the sum of their pay-outs was not a function of the choice of D and the firms
cost of capital was not a function of D. Absolute statements about the valu-
ation of equities and debt would require a specification for the stochastic
process driving the underlying assets of the firm (and a theory of valuation
of contingent claims that did not yet exist). ModiglianiMiller showed that
these relative properties must hold true for any stochastic process that drives
the asset values of the firm, and this had major ramifications for how manage-
ment can and cannot create value for the firms capital providers (in perfect
market conditions).
Whilst their theory could be derived from a relatively simple arbitrage argu-
ment and could be shown to ultimately be a simple expression of the additiv-
ity of present values, it still ran counter to perceived wisdomwhich was that
firms should prefer to use debt rather than equity until required bond returns
became prohibitive. ModiglianiMiller recognised that their assumption of
perfect markets was a drastic simplification, and that real-life complexities
may have an impact on optimal financial structure. Their paper prompted
decades of further academic research into how market imperfections such as
taxation,8 bankruptcy costs,9 management agency costs10 and information-
signalling11 could make capital structure relevant to firm value. But their
theory showed what did not matter: whilst expected equity returns would be
enhanced through debt financing, this only systematically increased the riski-
ness of the equity return. Raising debt merely geared up equity returns in a
way that individual shareholders could choose to do (or undo) with their own
borrowing and lending (rather like in Fishers Separation Theorem).
Modigliani and Miller published a second jointly-authored paper12 in
1961 that provided the equivalent irrelevancy theorem for corporate divi-
dend policy: in perfect capital markets, changes in dividend policy could not
8
Miller (1977).
9
Stiglitz (1972).
10
Jensen and Meckling (1976).
11
Ross (1977).
12
Miller and Modigliani (1961).
have an effect on the overall return earned on a firms equity, all other things
being equal. Again their argument was that all value was determined on the
asset side of the corporate balance sheetwhether firms chose to finance that
investment through retained earnings or new equity issuance or other forms
of capital raising was irrelevant to the economics of the value. Again they rec-
ognised that the theory was an abstractionfor example, in real life, dividend
decisions were likely to have important informational content for shareholders
but they once again highlighted that where these decisions did matter, it
was because of market imperfections or limitations rather than the fundamen-
tal economics. Miller-Modigliani were aware that what they were saying was
beguilingly simple:
Like many other propositions in economics, the irrelevance of dividend policy

is obvious once you think about it. It is, after all, merely one more instance of
the general principle that there are no financial illusions in a rational and
perfect economic environment. Values there are determined solely by real
considerations in this case the earning power of the firms assets and its invest-
ment policy and not by how the fruits of the earning power are packaged for
distribution.13
And yet, it is not difficult to find examples of twentieth-century economists

whose thinking on this topic was woolly at best. Recall Edgar Lawrence Smiths
argument that US corporations practice of retaining and reinvesting a signifi-
cant portion of their earnings was a vital factor in the ability of equities to
so strongly out-perform corporate bonds. Smith was not a substantial econo-
mist, but his views were endorsed by John Maynard Keynes, arguably the most
influential economist of the century. In reviewing Smiths book, Keynes wrote
of Smiths corporate-dividend-policy-as-explanation-of-equity-outperformance
argument: I have kept until last what is perhaps Mr Smiths most important,
and certainly his most novel, point. Well-managed industrial companies do not,
as a rule, distribute to the shareholders the whole of their earned profits.14
Miller and Modigliani brought a clarity to what mattersor rather to
what should not matterfor capital structure, dividend policy and corporate
investment policy. They did so by introducing a new formal rigour to the
embryonic discipline of financial economics, profoundly influencing its sub-
sequent practice. Franco Modigliani and Merton Miller received the Nobel
Prize in Economic Sciences in 1985 and 1990 respectively, both in part for
their two jointly written papers.
13
Miller and Modigliani (1961), p.414.
14
Keynes (1983), p.250.
odern Portfolio Theory andtheCapital Asset

M
Pricing Model
1952 was a vintage year for quantitative finance. Chapter 3 has already iden-
tified one seminal moment: the publication of Frank Redingtons paper on
immunisation theory. A seminal concept in financial economics was intro-
duced in this year by Harry Markowitz in his paper, Portfolio Selection.15 Like
Merton Miller, Markowitz spent much of his academic career at the University
of Chicago. This paper is the starting point for the branch of financial eco-
nomics known as modern portfolio theory. It also provided the foundation
for major developments in asset pricing theory. It was the first important
example of a quantitative, explicitly probabilistic analysis of the impact of
financial market risk on rational investor behaviour.
Markowitz considered how rational investors would construct portfolios
of risky investments. His analysis was built on two key assumptions: inves-
tors were risk-averse; and they cared only about the first two moments of
the probability distribution of future investment returns (or that all prob-
ability distributions of security returns could be described fully by their first
two moments). He also made the more basic assumption that investors were
non-satiated (they would always prefer more wealth to less). These assump-
tions implied that for a given level of expected return on their asset portfolio,
an investor would prefer a portfolio return distribution that minimised the
standard deviation (or variance) of the return; or equivalently, for a given
variance of return, they would prefer the portfolio return distribution that
maximised the expected return. Portfolios that produced the minimum vari-
ance for a given level of expected return were termed efficient portfolios by
Markowitz.
Markowitz used some elementary mathematical statistics to show that,
whilst the expected portfolio return was simply a value-weighted average of
the underlying expected returns of securities in the portfolio, the variance of
the portfolio return was a more complicated expression that was dominated
by the covariances between the securities. Markowitz showed that, given these
expressions for portfolio expected return and portfolio variance of return,
portfolio weights in the universe of available securities could be found that
would generate efficient portfolios. Moreover, an efficient frontier of portfolios
could be determined where each point on the frontier represented the portfo-
lio with the lowest variance for a given expected return.
Markowitz (1952).
15
The identification of these portfolio weights was a complex computational

problem when the investment universe consisted of a large number of securi-
ties. But Markowitz was able to generate illustrative explicit solutions for the
three-security case, and to make more general observations on the behaviour
of the possible portfolio combinations. In particular, under the assumptions
that there would be no short-selling of securities or gearing in the portfolio,
and that every securitys return was risky (i.e. had non-zero variance), he was
able to develop the diagram in Fig.4.2 below.
Figure 4.2 highlighted the power of the diversification of riskfar better
risk/return combinations could be obtained by constructing efficient port-
folios than holding individual securities. But Markowitz also showed that
diversification was about more than merely holding many different securities
it was driven by the covariances amongst securities, not merely by the num-
ber of securities in the portfolio. And whilst imperfect correlation between
securities was the mathematical key to diversification, the positive correlations
that tended to be empirically observed meant that there was a limit to how
much diversification could be generated:
This presumption that the law of large numbers applies to a portfolio of securi-
ties cannot be accepted. The returns from securities are too intercorrelated.
Diversification cannot eliminate all variance.16
Efficient Portfolio
E / V combinations
. .
Expected Portfolio Return
Security i Security j
.
Possible Portfolio
E / V combinations
Security k
Variance of Portfolio Return
Fig. 4.2 Portfolio risk and return (Based on Markowitz (1952) Fig.1))
16
Markowitz (1952), p.79.
Markowitzs approach echoed the arguments made by the British actuary

George May 40 years earlier. In 1912, May urged actuaries and investment
managers to give greater consideration to the benefits of diversification and
hence to spread these investments over as large an area as possible.17 He
emphasised that the riskiness of a security should not be measured by the
safety of the capital in each investment by itself .18 He never used the words,
but his argument that the stand-alone risk of the security was not the relevant
risk measure suggests the securitys risk should be considered by its marginal
contribution to the portfolios riskiness, which is exactly what Markowitzs
analysis implied.
Of course, May did not develop the mathematical analytical framework of
Markowitz that would prove so durable as an investor tool, and so fundamen-
tal as a building block of asset pricing theory. For this contribution Markowitz
received the 1990 Nobel Prize in Economic Sciences.
The next step in the development of modern portfolio theory was made by
James Tobin, a leading US macroeconomist (this time from Yale University).
In a 1958 paper he introduced what became known as the Tobin Separation
Theorem.19 Tobin extended Markowitzs analysis by considering how the pres-
ence of a risk-free (zero volatility) security impacted on the efficient frontier
of portfolio choices. The risk-free security has zero variance and therefore
naturally has zero covariance with all other securities. This seemingly trivial
observation has profound implications for the efficient frontier: it results in
a new, improved, linear efficient frontier and it implies that all efficient port-
folios are a combination of the risk-free security and the risky efficient port-
folio that is found at the tangent between the risky efficient frontier and the
risk-free return. This linear efficient frontier can be extrapolated to the right
of the risky asset portfolio by assuming negative positions in the risk-free
security can be obtained (that is, the investor can borrow at the risk-free rate).
Figure4.3 below updates the previous diagram to reflect the impact of the
presence of the risk-free asset.
The striking implication of Tobins analysis was that all investors should
hold the same sub-portfolio of risky assets, irrespective of their risk preferences
and wealth. An investor would obtain his preferred level of portfolio risk by
choosing the mix of this sub-portfolio of risky assets (labelled P in Fig.4.3)
and the risk-free security that produced their preferred level of portfolio
volatility. Thus the choice of risky asset sub-portfolio is separated from the
investors risk preferences. This is the Tobin Separation Theorem. In 1981,
Tobin also collected the Nobel Prize in Economics.
17
May (1912), p.153.
18
May (1912), p.136.
19
Tobin (1958).
Efficient Portfolio
E / V combinations
(with risk-freeasset)
. .
Exp ected Portfolio Return
Security i
Risky Portfolio,P
Possible Portfolio
.
E / V combinations
(risky assets)
Security k
Variance of Portfolio Return
Fig. 4.3 Portfolio risk and return (with risk-free security)
So far, the portfolio theory story has been focused on how rational investors
should behave in constructing investment portfolios. The story then moved
onto a distinct new phase: financial economists next considered what this
behaviour would imply for the pricing of risky assets. If investors followed the
behaviour described by Markowitz and Tobin, what type of risk would impact
on the price of an asset? In what way? At this time, there was no economic
theory of how risk impacted on equilibrium asset prices. A great deal of eco-
nomic theory explained how the risk-free interest rate was determined. And
there was a general recognition that risky assets should offer a risk premium.
But no one at this time (circa 1960) had an economic theory for how the risk
premium on a particular security should be determined.
Several financial economists in the USA (in particular, Sharpe, Lintner,
Mossin and Treynor) took the portfolio risk work of Markowitz and Tobin as
a platform and built, independently and simultaneously, theories of risky asset
pricing. Somewhat remarkably, they each produced more or less the same
theory at more or less the same time, and this theory became known as the
Capital Asset Pricing Model (CAPM). Rubinsteins History summarises this
state of affairs:
All four economists adopted nearly the same set of assumptions (mean-variance
preferences, perfect and competitive markets, existence of a riskless security, and
homogeneous expectations) and reached nearly the same two key conclusions:
(1) all investors, irrespective of differences in preferences and wealth, divide

their wealth between the same two portfolios: cash and the market portfolio,
and (2) equivalent versions of the CAPM pricing equation.20
A first glimpse of what was to come was offered by William Sharpe in

his 1963 paper, A Simplified Model for Portfolio Analysis.21 This paper was
still focused solely on portfolio theory rather than asset pricing theory and it
addressed the practical application of Markowitzs portfolio construction opti-
misationas Markowitz himself had noted, this would be a computationally
intensive process when the universe of risky securities was large in number.
Sharpe proposed restricting the covariance relationships between risky securi-
ties such that they could be summarised by a single common statistical factor
plus a single independent source of risk per security. This Diagonal Model
was much easier to work with in terms of computing the composition of
efficient portfolios. Sharpe also undertook some empirical analysis of 20 secu-
rities traded in the NewYork Stock Exchange between 1940 and 1951 and
concluded that the diagonal model may be able to represent the relationship
among [the realised returns of ] securities rather well.22
Sharpes next paper23 moved decisively into the territory of asset pricing
theory: rather than taking asset prices as exogenous inputs as per portfolio
theory, he considered how asset prices would behave if all investors rationally
pursued the Markowitz/Tobin portfolio construction methods. Lintners
paper24 was published the following year, and had already been written when
Sharpes paper was published, but as the first to publication, Sharpe generally
receives the greatest credit for the CAPMs derivation (Sharpe received the
Nobel Prize in Economics in 1990 for CAPM). At a high level, the funda-
mental insight that drove their asset pricing work was that if all investors held
the risky asset portfolio P as implied by Markowitz and Tobin, then P must
be the market value-weighted portfolio of all risky assets (generally referred to
as the market portfolio). Thus risky asset prices must be determined by a risk
pricing process that resulted in this occurring. This observation, together with
the observation that the market portfolio is mean-variance efficient and is on
the tangent with the linear efficient frontier, facilitated the geometric deriva-
tion of the famous CAPM pricing equation:
20
21
Sharpe (1963).
22
Sharpe (1963), p.292.
23
Sharpe (1964).
24
Lintner (1965).
ri = rf + i ( rm rf )

where:
ri is the expected return on security i;

rf is the risk-free rate of return;
rm is the expected return on the market value-weighted portfolio of risky
assets;
and i = ( i / m ) im
and where I is the volatility of security i; m is the volatility of the market

portfolio return; and im is the correlation between the returns of security i
and the market portfolio.
CAPM delivered a rigorous economic theory of how financial market
prices would be impacted by risk. It was also a highly intuitive result. The
CAPM stated that the expected return on risky assets would only be impacted
by the systematic, non-diversifiable component of its volatility (as measured
by its beta). Diversifiable risk would not be rewarded. This was an antici-
pated result: it was well-understood by this time that diversifiable risk was,
by definition, risk whose impact could be easily mitigated by investors, and
hence should not command a risk premium. Further, CAPM stated that the
expected return would increase linearly in proportion to the securitys risk (as
measured by its beta). The linear risk premium was also tentatively anticipated
by economists, but prior to CAPM they were unable to show why, or indeed
in what component of risk the risk premium would be linearly dependent.
The CAPM implies that the realised return on a given security can be de-
composed into two statistically independent components: an expected com-
ponent of the return (which is conditional on the realised market portfolio
return) and an unexpected component. Sharpes Diagonal Model of joint
realised security returns is consistent with this structure. Recall that in this
model all stocks have an exposure to a common factor and an independent
idiosyncratic risk (a correlation structure that was used to facilitate compu-
tationally efficient portfolio optimisations). However, the Diagonal Model is
not the only correlation structure for realised returns that is consistent with
CAPM. Empirically, significant correlations between the unexpected com-
ponents of the realised returns of different securities tend to be observed in
returns data (for example, across securities in the same industrial sector). This
is not inconsistent with CAPM.CAPM does not restrict the realised return
correlation structure such that the unexpected return components (relative
to the return component explained by the market portfolio) are uncorrelated

across all securities. It simply says that such correlations do not matter to the
pricing of the securities. In CAPM, the only correlation that matters to the
asset price is the correlation of the security with the market portfolio (which
is reflected in the securitys beta). That does not imply it need be the only
significant correlation that can be found between realised security returns.
More general asset pricing models were developed following CAPM, such
as Stephen Rosss Arbitrage Pricing Theory (APT).25 This model allows for
the possibility of a broader, indeed arbitrary set of statistical factors to drive
realised security returns and to be potentially relevant to the pricing of assets.
However, it does not provide a specification of how to determine what risk
premiums should be commanded by such factors in equilibrium asset prices
it only states what the relative behaviour of different asset prices must be to
avoid arbitrage for a given set of risk premium assumptions for the factors.
CAPM can be considered as a special one-factor case of the APT, and one in
which the risk premium structure is fully specified.
These key ideas of portfolio theory and asset pricing theory are now long-
established cornerstones of investment education and practice: they are found
in the syllabi of business schools and finance professions (including actuarial
ones) around the world; an assets beta is a universally recognised measure of
risk (and required return) amongst investment professionals.
Option Pricing Theory

The publication of the option pricing work of Black, Scholes and Merton
in 1973 ranks amongst the most famous and influential developments of
financial economics. Their work threaded together three concepts that had
individually emerged earlier in the twentieth century: Brownian motion as
a description of the behaviour of security prices; valuation by the law of no-
arbitrage; and the use of risk-neutral probabilities in valuation.
The earliest known example of the modelling of financial assets as a type of
Brownian motion was produced by the French mathematician Louis Bachelier
in his doctoral thesis, which was accepted at the Sorbonne in 1900.26 In fact,
the stochastic modelling of Bacheliers thesis is now generally recognised as
the first mathematical modelling of Brownian motion in any context, beating
25
Ross (1976).
26
Bachelier (1900). For a usefully annotated English translation and accompanying historical discussion,
see Davis and Etheridge (2006).
Albert Einsteins famous particle motion paper by a handful of years (Einstein

was unaware of Bacheliers work).
Bacheliers motivation for developing a stochastic model of security prices
was not merely to obtain probability distributions of future prices of funda-
mental assets like equities or bonds or their path behaviour. Such develop-
ments would in themselves be revolutionary for the time, but to Bachelier the
stochastic description of assets was merely a means to an end: his thesis was not
on how to model equities or bonds, but on how to value options (specifically
options on French government perpetuities). He had the conceptual insight
that the valuation of such options would require a continuous-time stochastic
process to be specified for the path of the underlying asset. This is an astute
insight given that no such form of mathematical model existed at the time.
Bachelier wasnt merely identifying new applications for existing mathematical
frameworks, he was identifying the need for a mathematical apparatus that did
not exist. So Bachelier proceeded to develop it and attempted to apply it to the
problem of option pricing. As well as introducing the mathematical modelling
of Brownian motion, he also developed the Chapman-Kolmogorov equation
and made other important developments that showed how the mathematics
of physics could be transferred to the emerging field of stochastic analysis. His
work was not mathematically rigorous, but it was recognised by the leading
stochastic theoreticians of the following half-century such as Kolmogorov as
they developed the formal mathematics of probability theory.
Despite being well-known to probability theorists (and being cited in the
most important stochastic analysis papers of the century), Bacheliers work,
and in particular, his attempts at option valuation, went completely unno-
ticed by the economics profession for over 50 years. This changed sometime
in the mid to late 1950s, when a leading American statistician of the time,
Jim Savage, sent a note to a few leading economists asking if they had read
Bachelier. Paul Samuelson, the MIT professor who can be viewed as the leader
of the new breed of quantitative financial economists that emerged in the third
quarter of the century, was one of the recipients of the note. Samuelson had
not heard of Bachelier or his work, but he browsed the MIT library and found
a copy of his 1900 thesis. Inspired by what he read, he circulated Bacheliers
work amongst his peers and students. Bachelier had not solved option pric-
ing, but he was able to make some accurate statements about option pricing
behaviourfor example, showing that an at-the-money option price would
increase in proportion to the square root of its term to maturity (this is not
strictly true under more realistic stochastic processes for the underlying asset
than the arithmetic Brownian motion assumed by Bachelier, but is nonetheless
usually a good approximation). He was also able to derive many probability
statements for the behaviour of the underlying asset such as the probability
that it would reach a particular price over a given time period (again under the
assumption that it followed an arithmetic Brownian motion).
Bachelier had done enough to show Samuelson that his application of
continuous-time stochastic processes to derivative pricing opened up many
possibilities for further research. And the work that had been done by stochas-
tic mathematicians in the intervening fifty years to develop Bacheliers model-
ling intuitions into rigorous mathematics provided a ready-made modelling
toolbox for a new generation of financial economists.
The next notable contribution to follow Bachelier on the application of
Brownian motion as a description of the stochastic path behaviour of asset
prices took a full 59 years to appear. M.F.M.Osborne, a researcher at the US
Naval Research Laboratory in Washington DC, published a paper, Brownian
Motion in the Stock Market,27 which broke new ground in several ways.
Osborne proposed that stock prices followed a geometric Brownian motion
that is, it is the natural logarithm of the asset price that is normally distrib-
uted, rather than the asset price itself. Osborne does not appear to have been
aware of Bachelier, but the use of a geometric rather than arithmetic process
was fairly intuitive: the possibility of negative prices was removed; and the
process could be viewed as a multi-period generalisation of the single-period
mean/variance description of security returns introduced by Markowitz a few
years earlier. From this point on, geometric Brownian motion would be the
standard modelling assumption for the stochastic process of an asset price.
Secondly, Osborne performed some statistical tests of how well empirical asset
price behaviour (US stock prices from various sources going as far back as
1831) conformed to the geometric Brownian motion model. Osborne tested
for this in two ways: by considering the cross-sectional dispersion of many
stock price changes on a given date; and by analysing how the variance of
stock prices increased with the length of the measurement interval (up to ten
years). Rubenstein has noted that Osborne was the first researcher to perform
this second type of test on asset price data.28 Osborne concluded that the
empirical data accorded reasonably well with the geometric Brownian motion
model (though he noted that the empirical data produced tails that were fatter
than those predicted by the model).
In 1965, Paul Samuelson, now armed with Bacheliers option pricing thesis
and Osbornes geometric Brownian motion proposal and supporting empiri-
cal stock price analysis, synthesised these ideas in a paper that examined option
Osborne (1959).
27

28
pricing under the assumption that the underlying asset follows a geometric
Brownian motion.29 Samuelson, with the assistance of the world-leading sto-
chastic processes professor Henry McKean, who was also at MIT, was able
to derive a call option pricing formula under the assumption that the option
had some constant required return and the underlying stock had a constant
expected return of (and followed a geometric Brownian motion with vola-
tility ). The theory of stochastic analysis was now sufficiently developed to
have permitted a closed-form solution to be found for the discounted present
value of the option pay-off. But, critically, Samuelson had no theory for how
to determine the option discount rate . He recognised that the call option
was a leveraged exposure to the underlying asset in the sense that a given
change in the stock price would have a bigger proportional impact on the call
price than the stock price. In the CAPM terminology, the call options beta
was bigger than that of the underlying asset. So he argued it must be the case
that was greater than . But the size of the options leverage (and hence beta)
also evidently varied with the underlying asset priceit was a stochastic pro-
cess itself. Samuelson had no strategy for allowing for that. He was close. He
had perfected the application of stochastic processes to determine the options
expected pay-off, but he had no way of transforming it into an economically
justified present value.
1973 saw the publication of two papersthe BlackScholes paper30 and a
paper by Merton31that used a crucial economic insight to determine how
Samuelsons formula could be parameterised to deliver an arbitrage-free option
pricing formula for the first time. Collectively, the publication of these papers
represented a watershed moment: the era of modern option pricing theory
starts here. Scholes and Merton were awarded the Nobel Prize in Economics
in 1997, primarily for the contributions in these papers (Fisher Black died in
1995 and so could not receive the award).
The BlackScholes paper arrived a few months before Mertons. Although
the papers were written contemporaneously and cover similar ground, the
BlackScholes one is the natural starting point for the reader as it is somewhat
more direct and less elaborate (at least for the non-mathematician). Merton
is more general and more formal. The critical breakthrough in Black and
Scholess treatment of the option pricing problem came with the recogni-
tion that changes in the option value were entirely driven by changes in the
stock priceover an instantaneous period, the option and the underlying
29
Samuelson (1965b).
30
Black and Scholes (1973).
31
Merton (1973).
asset were therefore perfectly correlated assets. This implied that a continu-
ously rebalanced portfolio of the option and the underlying asset could be
constructed that would generate a certain rate of return. That is, a portfolio
could be constructed that included one unit of the underlying asset and an
appropriate short position in the call option that provided an equal and offset-
ting exposure to movements in the underlying asset. The short position would
be determined by the rate of change of the call option value with respect to
the underlying asset. In a footnote in their paper, Black and Scholes attribute
this critical idea of the riskless dynamic portfolio to Robert Merton.
Mathematically, the portfolio would generate an instantaneous certain
return if it was of the form:
1
Hedge Portfolio = S C ( S ,t )
C
S
The certain return earned by the hedge portfolio must be the risk-free rate in
order to avoid arbitrageif it was greater (lower) than the risk-free rate, an
investor could borrow (invest) at the risk-free rate and create unlimited profits
by investing in (shorting) the hedge portfolio. Black and Scholes assumed the
stock price followed a geometric Brownian motion. As the call option price
is a function of the stock price, stochastic calculus (in particular, Itos lemma)
can be used to express a change in the call option price as a function of the
derivatives of its value with respect to the underlying asset and time:
C C 2 S 2 2C C
dC ( S,t ) = + S + 2
dt + s dZ ( t )
t S 2 S S

where and are the drift and variance terms of the underlying assets geo-
metric Brownian motion and dZ(t) is the underlying assets Brownian motion
or Weiner process.
From this and the assumption that the hedge portfolio must earn the risk-
free rate, together with the elementary boundary conditions of the option
pay-out at maturity, a partial differential equation can be derived which is
familiarly known as the heat-transfer equation of physics (or simply the heat
equation), and for which there was a well-established pre-existing solution.
Just like that, the arbitrage-free option pricing equation was obtained (under
various technical assumptions such as investors having the ability to continu-
ously rebalance asset positions without transaction costs). The observation that
dynamic hedging could replicate the options pay-offs (or equivalently, gener-
ate a risk-free portfolio) was pivotal and had transformed the option pricing
challenge into a strictly relative problem: it did not require a grand, unified
explicit model of how the economy priced all risky securities; instead, it took
the current value of the underlying asset as an exogenous input; that input,
together with a description of the underlying asset prices stochastic process
and the risk-free interest rate, was enough to define the arbitrage-free price
(at least for their case where the stochastic process was well-behaved and fol-
lowed a geometric Brownian motion). Moreover, not all characteristics of the
underlying assets stochastic process had to be specified in their option pricing
equation. The expected return of the underlying asset ( in Samuelsons nota-
tion) did not even feature in the BlackScholes pricing equation.
Black and Scholes were not satisfied with providing a single route to their
arbitrage-free pricing equation: they also provided a second, alternative
approach to its derivation that was based on the CAPM. The CAPM was
a single-period model. It had no direct answer to the problem of valuing a
security whose beta changed continuously and stochastically. However, Black
and Scholes showed that the CAPM could be used to determine the return
required on the option at any given instant. They did this by first showing that
the options beta can be related to the underlying assets beta by the equation:
C S
C = S
S C
This allowed the required return of the option to be written in terms of the
underlying asset beta, and by equating this with the expected return implied
by the Ito expansion of the change in the value of the call option as a function
of the underlying asset and time, they were again able to obtain the heat equa-
tion that they obtained in the dynamic hedging derivation (and the various
terms cancel out). In the historical context of the development of financial
economics of the previous 20 years, this CAPM approach to the option pric-
ing formula would perhaps have been the most natural to a contemporary
economist, but it was the dynamic hedging argument that proved to be most
revolutionary in its broader applicability to derivative pricing and hedging.
Black and Scholes did not use the concept of risk-neutral valuation in
either of their two derivations of the option pricing formula. They noted,
almost in passing, that the option price does not depend on the expected
return on the underlying asset (though the expected return on the option does
depend on the expected return on the underlying asset through the above beta
relationship). They did not note that their option pricing formula gave the
risk-neutral expected option cashflow discounted at the risk-free rate (i.e. the
Samuelson formula with = = r ).
It was Merton who generalised and extended the BlackScholes result in
his 1973 paper, and arguably he provided a more mathematically rigorous
treatment of the derivation of the pricing formula. Merton highlighted that
the BlackScholes formula was equal to the risk-neutral cashflow expecta-
tion discounted at the risk-free rate, though this was not a central focus of
his paperthe power of the generality of risk-neutral valuation was not yet
appreciated or anticipated. Merton also generalised the BlackScholes formula
by introducing a stochastic (rather than fixed) interest rate into the model. He
showed that in this case, the dynamic hedge portfolio needed to include a
third asset alongside the underlying asset and call option: a zero-coupon risk-
free bond that matures at the option maturity date. He also extended the for-
mula to allow for the impact of dividends paid on the underlying asset during
the term of the option. As a final flourish, he even included an option pricing
formula for a down-and-out exotic call option (where the option expires if
the underlying asset falls to a specified level at any time before maturity).
With these two papers, the permanent and rigorous foundations of
arbitrage-free option pricing theory had been laid. The vast research pro-
gramme on option pricing theory that their work inspired over the following
decade would focus on a few key themes: applying option pricing theory
to the theory of capital structure; understanding the fundamental valuation
ideas that could be identified in the arbitrage-free option pricing formula (in
particular, risk-neutral valuation); fully developing the mathematical rigour
of the arbitrage-free valuation so as to apply it to a broader range of problems
(such as yield curve modelling). We discuss each of these threads of develop-
ment briefly below (yield curve modelling is discussed in greater detail in the
next section).
The financial economics academic profession around this time was almost
obsessed with option pricing. Exchange-traded option contracts were,
especially in the 1970s, relatively unimportant, niche financial instruments
(though trading volumes increased significantly in the years following Black
ScholesMerton). The academic interest in option pricing theory was primar-
ily driven by the recognition that option-type pay-offs arose in many forms
of more widely relevant contingent claims. Perhaps most importantly, the
ModiglianiMiller capital structure proposition could be viewed as a form of
put-call parity, where the equity of the firm was a call option on the value of
the firms assets, and the debt of the firm was a risk-free bond less a put option
on the value of the firms assets. For Black and Scholes, this was the ultimate
area of interest, as is reflected in the title of their paper, which refers to the
pricing of corporate liabilities. Modigliani and Miller showed that, in well-
functioning markets, the sum of the claims on the firm always added up to
the same total. But they did not say anything about how to value each of the
different claims. Option pricing theory opened up the possibility of answer-
ing this question.
The application of the BlackScholes option pricing formula to the valua-
tion of corporate liabilities was further developed by Robert Merton in a 1974
paper.32 The paper did not introduce significant new, fundamental theories.
Rather, it formally clarified the conditions required for the use of the Black
Scholes formula in the valuation of corporate liabilities, and it tabulated and
discussed the corporate bond prices, yields and spreads implied by the for-
mula for various assumptions for firm asset volatilities, and the term and level
of debt (all assuming the value of the underlying assets of the firm follows a
geometric Brownian motion). This paper has resulted in the terminology of
the Merton model being used to refer to corporate bond pricing models that
use the insight that a corporate bond is a risk-free bond less a put option on
the value of the firms assets. Such models have been widely used in modern
risk management practice in banking and credit risk modelling.
Curiously, the analogous insight for equitiesthat equities can be viewed
as a call option on the value of the firms assetshas not been so widely
used as the basis for the stochastic modelling of equity returns and the mea-
surement of equity risk. Such a modelling approach can provide a logical
economic explanation for the well-documented failings of modelling equities
directly as a geometric Brownian motion. Since the 1970s, academics and
practitioners have taken geometric Brownian motion as the starting point
for equity modelling, and have then extended it in various complex ways to
capture the fat tails and negative skew that are observable in empirical equity
return data and that are implied by equity option market prices. By starting
with the assumption that it is the value of the underlying assets of the firm
that follow the geometric Brownian motion, and then recognising that equity
is a call option on the value of the firm, similar dynamics can be produced
and arguably in a more economically coherent way. Under this approach, an
option on the equity of the firm becomes a compound optionan option
on an option. Robert Geske, a financial economics professor at UCLA, devel-
oped this logic and obtained the valuation formula for the compound option
when the underlying firm asset value follows geometric Brownian motion in a
32
Merton (1974).
paper published in 1979.33 In his paper he argues that this approach to valu-
ing equity options is a better theory that produces better empirical results:
The introduction of these leverage effects adds a new dimension to theoretical

option pricing. Any change in the stock price will cause a discrepancy between
the compound option value and the BlackScholes value [when applied directly
to the equity of the firm]. The qualitative discrepancies between these two for-
mulas correspond to what practitioners and empiricists observe in the market
namely, that the BlackScholes formula underprices deep-out-the-money options
and near-maturity options, and it overprices deep-in-the-money options.34
We now move onto the next thread of post-BlackScholesMerton

research: the general application of risk-neutral valuation as a method of valu-
ing options. Cox and Ross explored this topic in an important paper pub-
lished in 1976,35 where they argued:
The fact that we use a hedging argument to derive [the option pricing equation]
and the fact that P(S, t) [the price of the option] exists uniquely means that
given S and t the value of the option, P, does not depend directly on the struc-
ture of investors preferences. Investors preferences and demand conditions in
general enter the valuation problem only in so far as they determine the equilib-
rium parameter values. No matter what preferences are, as long as they deter-
mine the same relevant parameter values, they will also value the option
identically . A convenient choice of preferences for many problems (although
one can envision problems where another preference structure might be more
suitable) is risk neutrality. In such a world equilibrium requires that the expected
returns on both stock and option must equal the risk free rate.36
By this time, the concept of risk-neutrality in valuation was not new. As we

saw above, Merton noted in his 1973 paper that the BlackScholes formula
was equivalent to the risk-neutral expected cashflow of the option discounted
at the risk-free rate. Furthermore, Samuelson and Merton had derived an early
glimpse of the risk-neutral valuation equation in a paper published in 1969.37
Nonetheless, Cox and Ross highlighted the generality of its applicability as
33
Geske (1979).
34
Geske (1979), p.76.
35
Cox and Ross (1976).
36
Cox and Ross (1976), p.153.
37
Samuelson and Merton (1969). There are also other competing claims for the earliest derivation of risk-
neutral valuation, but the Samuelson and Merton work was most closely related to the option valuation
literature.
a computational technique for option valuation. They illustrated this point

by showing how it could be used to value options under a range of assumed
stochastic processes for the underlying asset that were alternatives to the now
ubiquitous geometric Brownian motion. Perhaps most importantly, they
established the link between risk-neutral valuation and arbitrage-free prices.
As Merton himself has written, They [Cox and Ross] were the first to recog-
nise that this relation [risk-neutral valuation] is a fundamental characteristic
of arbitrage-free price systems in continuous-trading environments.38
Cox and Ross collaborated with Mark Rubinstein to publish in 1979
another seminal paper in the history of option pricing theory: Option
Pricing: A Simplified Approach.39 The paper attempted to cut through the
increasingly sophisticated stochastic calculus of the option pricing theory of
the previous few years and clarify the underlying economics. They did this by
abandoning the use of continuous-time stochastic processes to describe the
underlying asset behaviour. Instead, they considered option pricing theory
in the presence of the simplest possible stochastic process for the underlying
asset: a single discrete time-step where the underlying asset can take two pos-
sible end-period values with specified (non-zero) probabilities (often referred
to as a binomial tree). They then extended this tree structure to multiple
time-steps and used a backward recursive process (like that used by Pascal in
1654in his solution of the Problem of Points discussed in Chap. 1) to find
option values at every node of the tree.
Cox, Ross and Rubinstein showed that the key option pricing theory
resultsrisk-neutral valuation produces the arbitrage-free price and the same
pricing function can be obtained by a dynamic hedge strategy; the option price
is not a function of the underlying assets expected return but does depend on
its volatilityapplied in this simpler setting. They then showed that the bino-
mial tree option pricing formula converged with the BlackScholes formula
as the number of time-steps in the binomial tree increased towards infinity.
This method was very useful for two quite different reasons: it provided a
highly intuitive approach to teaching option pricing theory that allowed the
fundamental economic logic to be accessed without resorting to advanced
mathematics; and the backward recursive tree process provided an advanced
analytical approach for valuing path-dependent option features.
A remarkable decade in the history of option pricing theory was rounded
off by two papers co-authored by Michael Harrison, a Stanford Professor in
38
Merton (1990), p.335.
39
Cox, Ross and Rubinstein (1979).
operations research.40 These papers took the risk-neutral valuation logic of

the 1976 Cox and Ross paper, and further developed it into a more gen-
eral and mathematically rigorous valuation theory. These papers showed that
an equivalent martingale measure could be used to define the arbitrage-free
prices of all forms of securities (under appropriate technical conditions). This
meant that, in an arbitrage-free market, there existed a probability measure
such that the discounted prices of all securities (stocks, bonds, options, other
derivatives) must follow a martingale: that is, the expected value of the ratio
of the security price to the risk-free account must be driftless over time. The
papers mathematical rigour came at the expense of accessibility. As noted
by Davis and Etheridge, This paper [Harrison and Pliska] has turned finan-
cial economics into mathematical finance.41 It marked the completion of an
extraordinary stream of economic research that found immediate and exten-
sive application in the financial sector.
In stark contrast to the binomial trees of Cox, Ross and Rubinstein, the
complete and generalised mathematical description of dynamic hedging,
replication, arbitrage, martingales and risk-neutral valuation was formidable.
A quantitative revolution in finance was first heralded by Redington and
Markowitz in 1952. Their work required a limited mathematical tool-kit to
understand itessentially little more than elementary differential calculus
and, in Markowitzs case, also basic mathematical statistics. By 1981, work
at the cutting-edge of financial theory required the post-doctoral quantitative
skills of the rocket scientist. Unfortunately, this inaccessibility could get in the
way of the big ideas that were highly intuitive and yet still with profound
implication for the management of financial risk. We will revisit this topic in
later discussions of how actuarial thinking evolved alongside these develop-
ments in the second half of the twentieth century.
Option pricing theory was, and is, just thata theory. The development
of the arbitrage-free pricing models of Black, Scholes, Merton, Cox, Ross and
the rest all rely on some assumptions that are quite evidently unrealistic. Of
these, perhaps the most fundamentally important is that the underlying asset
of the derivative follows a form of stochastic process that permits dynamic
hedging to deliver perfect replication of the options pay-off. If, for example,
the assumption of geometric Brownian motion was relaxed to also permit dis-
continuous jumps to occur in the underlying asset price, it would not be pos-
sible to perfectly replicate the option pay-off through continuous rebalancing
of a portfolio of the underlying asset and cash. The logic of arbitrage-free
Harrison and Kreps (1979); Harrison and Pliska (1981).

40
Davis and Etheridge (2006), p.114.

41
pricing would therefore no longer hold. Given the empirical tendency of mar-
kets to suffer short-term jumps in price that are difficult to reconcile with
a continuous form of stochastic process for the price, this has and should
prompt caution in applying the model in real-life applications (including
actuarial ones).
As with any model, it is crucially important that the model user under-
stands its assumptions and limitations. These limitations have generally been
well-understood by market practitioners, arguably since the 1970s and par-
ticularly since the equity market crash of 1987. Market option prices are not
fully explained by the BlackScholes modelas is evidenced by the smile in
BlackScholes implied volatilities that has long been observed. Nonetheless,
these implied volatilities are also the currency in which traders quote option
pricessuch is the ubiquity of the BlackScholes model. The academic devel-
opments in option pricing theory of the 1970s have had a most profound
impact on the practices of structuring, pricing and hedging complex financial
market securities; and also in the theoretical valuation of forms of non-linear
contingent claim that can occur in an immensely wide array of economics
and businessfrom executive compensation to the valuation of oil fields. As
we shall see later, eventually it played an important role in British actuarial
thought and practice, particularly in the life sector.
Yield Curve Modelling

Economic theories of the term structure of interest rates have been part of
economic thought since around the start of the twentieth century. Irving
Fisher, the American economist whose eponymous separation theorem we
discussed earlier, was one of the early theorists on what determines the shape
of the interest rate yield curve. In 1896 he developed what became known as
the Expectations Hypothesis. This says that long-term interest rates simply
reflect the markets expectations for the future path of the short-term interest
rate.42 Under this hypothesis, the expected return on risk-free bonds is the
same for all terms of bond.
The Keynesian revolution in macroeconomics that emerged in the 1930s
as a response to the Great Depression generated, amongst other things,
new ideas in the theory of interest rates. One such idea was the Liquidity
Preference Hypothesis. This was mainly developed by the British economists
John Maynard Keynes and John Hicks as part of broader economic models
42
Fisher (1896).
of the behaviour of saving, investment and money.43 The Liquidity Preference

Hypothesis said that the expected future path for the short-term interest rate
was not the only factor that would impact on the prices of long-term risk-
free bonds: a risk premium would need to be offered to investors to induce
them to bear the additional price volatility associated with holding longer-
term bonds. Under this hypothesis, the risk premium offered by longer-term
bonds would always be positive and would increase with the bond term.
A third economic theory of the determination of the risk-free yield curve was
developed in US academia in the 1950s and 1960s, most notably by Modigliani
and Sutch (yes, that Modigliani).44 This theory, which became known as the
Market Segmentation Hypothesis or Preferred Habitat Hypothesis, could be
viewed as a generalisation of the Liquidity Preference Hypothesis. It argued
that different investors had different investment horizons and this horizon
determined investors bond maturity preferences. Investors would require a
risk premium to induce them to hold bonds with terms different to their
maturity preference. The Liquidity Preference Hypothesis can be viewed as
a special case of this theory where all investors have an instantaneously short
investment horizon. Under the Market Segmentation Hypothesis, the term
premiums embedded in risk-free bond prices could be positive or negative
at different points on the term structure, depending on investors investment
horizons and the supply of bonds.
In a world of deterministic interest rates, a simple arbitrage argument
shows that the Expectations Hypothesis must hold true. So theories of how
yield curves are determined are really about how interest rate uncertainty
impacts on bond prices. This is a complex and quantitative problem to tackle
rigorouslyto make meaningful progress, a mathematical description of the
nature of interest rate uncertainty is inevitably required; and, intuitively, any
theory must recognise that risk-free bonds of similar durations are close sub-
stitutes for each other and must be priced in a consistent way that does not
create any trivial free lunches (i.e. the bond prices must be arbitrage-free).
These two ingredientsa stochastic process for an economic variable and
its use in the determination of arbitrage-free asset priceswere the keys to
BlackScholesMertons option pricing breakthrough. So it is unsurprising to
find that the breakthrough in how interest rate uncertainty impacts on bond
prices was directly inspired by the BlackScholes dynamic replication and no-
arbitrage argument.
Keynes (1936); Hicks (1939).

43
Modigliani and Sutch (1966).

44
This breakthrough was achieved in a seminal paper by Oldrich Vasicek

that was published in 1977.45 Vasicek was a Czech probability theorist who
migrated to the USA in the late 1960s and began a career in investment bank-
ing in California. In his role as a researcher at a bank, he attended seminars by
Fisher Black and Myron Scholes in the early 1970s where they presented their
early ideas on option pricing. Vasicek saw that a risk-free bond of a given term
could be viewed as a form of derivative security where the short-term interest
rate was the underlying asset. This opened up the possibility of applying the
dynamic hedging ideas of BlackScholes to find a formula for the arbitrage-
free behaviour of the entire yield curve.
Vasicek first specified a general mathematical structure for the uncertainty
in future interest rate behaviour by defining a stochastic modelling framework
for the short-term interest rate. He assumed that the instantaneous short-term
interest rate (which we will refer to as the short rate) followed a continuous-
time stochastic process that had a few key properties: it was continuous (the
short rate could not jump); it was Markovian (its future value depended only
on its current value and not on previous values); and it had a single source of
randomness (what would now be called a single-factor model). He then sup-
posed that the yield on a risk-free bond of any term could be expressed as the
expectation for the short rate over the term of the bond plus a risk premium
component. Of course, if there are no restrictions on the how the risk pre-
mium component behaves as a function of the term of the bonds, this state-
ment is just a tautologythe risk premium at any point in the term structure
is merely the difference between the market bond yield and the expected short
rate path over the bond term. One of Vasiceks key results was to show that
in order for the yield curve to be arbitrage-free, there were well-specified con-
straints on what form the risk premium component could take across the
term structure of risk-free bonds.
The assumption that the short rate process had a single source of random-
ness (and that the risk premium component of bond prices did not have any
additional source of randomness) implied that all variation in bond prices of
all terms must be driven by the same single source of randomness. This, in
turn, implied that all risk-free bond returns would be instantaneously per-
fectly correlated (though bond expected returns and volatilities would vary
with term). This perfect correlation implied that some combination of any
two bonds of different maturities could be held that would (instantaneously)
produce a certain return (this would involve being long one bond of a given
term, and short another bond of another given term). In other words, the
45
Vasicek (1977).
short-term interest account could be replicated by a dynamic combination

of any two risk-free bonds. Like BlackScholes, Vasicek argued that as this
dynamic portfolio was instantaneously riskless, its rate of return must be the
short rate in order to avoid arbitrage.
This argument was very generalfor any two bonds of two different terms,
portfolio weights could be found such that the hedge portfolio would be
instantaneously riskless. The relative portfolio weights would be a function
of the relative sizes of the bond volatilities at the two different bond terms,
and these in turn would be determined by the short rates stochastic pro-
cess. Vasicek had not yet considered a specific form of stochastic processthe
dynamic replication argument held for any short rate process that satisfied his
general modelling assumptions. From this construction, and in very similar
style to BlackScholes, Vasicek used the dynamic hedge portfolio argument to
derive a partial differential equation for the bond price of any term t, which
was a function of the current short rate, its drift and variance processes and
the risk premium assumed to apply to the bond of term t.
This analysis uncovered the key arbitrage-free property of risk-free bond
prices in Vasiceks framework. It showed that every risk-free bonds instanta-
neous expected return in excess of the short rate must vary in linear propor-
tion to the bond prices instantaneous volatility. So at any given moment, a
ratio of the excess expected return to volatility existed that was constant across
all terms of risk-free bond. The risk premium component of all bond yields
was determined by this ratio, which Vasicek called the market price of risk.
Mathematically, this market price of risk could be written as:
( t ,T ) r ( t )
q (t ) =
( t ,T )

where (t,T) is the instantaneous drift rate at time t of the price of a bond
maturing at time T; r(t) is the short rate at time t, and (t,T) is the instanta-
neous volatility at time t of the price of a bond maturing at time T.In Vasiceks
framework, no-arbitrage implied that q(t), the market price of risk at time t,
did not vary as a function of the bond term T.
In this setting, the risk premium component could be positive or negative,
and it could vary in size over time, but at any given instant, it could only
take one sign across all points on the yield curve and it varied across the yield
curve in this prescribed linear way. If a bond had a risk premium different to
that implied by the market price of risk and the bonds volatility, it would be
possible to generate arbitrage profits by constructing a riskless portfolio that
generated returns in excess of the short rate.
This restriction on the term structure of the risk premium had sig-
nificant implications for the economic theories of the term structure. The
Expectations Hypothesis (market price of risk is zero) and the Liquidity
Premium Hypothesis (market price of risk is always positive) could both be
accommodated by Vasiceks no-arbitrage relationship, but it placed significant
limits on the broader Preferred Habitat Hypothesis. In essence, Vasicek had
shown that investors ability to reconstruct bond maturities through dynamic
portfolios of bonds of other maturities meant that there was a limit to the
way in which maturity preferences could impact on the relative differences
in arbitrage-free bond prices. For example, in Vasiceks model, if the ten-year
bond offered a negative risk premium and the 20-year bond offered a positive
risk premium, investors could perfectly replicate the ten-year bond pay-off
through a dynamic combination of the 20-year bond and cash that would
cost less than the ten-year bond price.
To complete his analysis of this general arbitrage-free bond pricing frame-
work, Vasicek then noted that the bond pricing equation that he derived
through the dynamic hedging process was equal to the bond price implied
by the expected short rate when the market price of risk is assumed to be
zero (though this was not a central object of his analysis, it was essentially an
incidental by-product of his real-world modelling presentation). In the same
way that BlackScholesMerton showed that the risk-neutral stock process
determined the value of any option on the stock, Vasicek had shown that the
risk-neutral short rate process determined the arbitrage-free price of risk-free
bonds of all terms.
Having established the above framework and its general properties, Vasicek
then proceeded to illustrate it with the use of a simple example: he assumed
that the short rate followed a mean-reverting process with a normally dis-
tributed volatility process (sometimes referred to as an Orstein-Uhlenbeck
process):
dr = ( r ( t ) ) dt + dZ ( t )

He also assumed the market price of risk was some constant q (i.e. it did not
vary over time). These model dynamics were never meant to be taken as a rec-
ommendation of a realistic description of interest rates, but rather to provide a
simple and tractable illustration of Vasiceks pricing framework. Nonetheless,
this short rate model is universally referred to as the Vasicek model. With
this modelling specification, the arbitrage-free bond pricing equation could
be solved to provide an analytical formula for the bond price as a function of
term. And the expected return and volatility of every risk-free bond were well-
defined functions of the parameters of the short rate process and the market
price of risk (as well as the bonds term).
Vasiceks demonstration that the BlackScholesMerton idea of arbitrage-
free pricing by dynamic replication could be used to price the risk-free term
structure was the permanent foundation for a new branch of quantitative
finance. A vast amount of yield curve modelling research followed in the next
two decades that further developed Vasiceks big idea. One of the most signifi-
cant developments to arise in the years following the Vasicek paper came from
a pair of papers published by Cox, Ingersoll and Ross (CIR) in 1985.46 These
papers were far less accessible than Vasiceks paper. Vasiceks paper had been
very focused in its ambitions: he showed that, under certain assumptions, a
given stochastic process for the short rate together with a market price of risk
could determine the arbitrage-free pricing of risk-free bonds of all terms. Cox,
Ingersoll and Ross broadened the setting. Rather than specifying a framework
where the stochastic process for the short rate is specified exogenously, they
developed a broader stochastic model of the wider economy (production pos-
sibilities, wealth, individuals utility) and all the assets that traded in it, and
then derived the short rate stochastic process that was implied by the specified
economy system.
Before specifically considering interest rate behaviour, their first paper
derived the by now ubiquitous risk-neutral valuation property for their eco-
nomic system. It is interesting to note that Vasicek did not focus the presen-
tation of his bond pricing mathematics in a risk-neutral setting. He used a
real-world probability measure, and showed that the bond pricing equation
could use these real-world probabilities together with the market price of risk
assumption to obtain the bond price (and this could be presented as a risk-
neutral valuation where the market price of risk transforms the real-world
probability measure into a risk-neutral one). After Vasicek, the biggest appli-
cation of the arbitrage-free yield curve modelling that he pioneered was
increasingly in interest rate derivatives pricing and hedging rather than in
macroeconomic descriptions of how interest rate uncertainty impacted on
yield curve behaviour. These applications were most efficiently addressed
by working in risk-neutral probability measures, and the use of real-world
probabilities increasingly disappeared from the yield curve modelling litera-
ture. Cox, Ingersoll and Rosss grand economic framework seamlessly moved
between real-world probabilities and risk-neutral probabilities.
Cox, Ingersoll and Ross (1985), 1 and 2.

46
The CIR paper, A Theory of the Term Structure of Interest Rates, broke
new ground beyond Vasicek in a number of important ways. Like Vasicek, they
derived the arbitrage-free bond pricing equation via a stated stochastic process
for the short rate. Their short rate model (below) assumed that the volatility of
the short rate moved in proportion to the square root of the short rate:
dr = ( r ( t ) ) dt + r ( t ) dZ ( t )

This was a slightly more sophisticated and realistic model of interest rate
behaviour than that used by Vasicek. Most notably, it precluded the possibil-
ity of negative interest rates (this has traditionally been seen as a desirable
feature of interest rate models, though that view has been subject to some
revision in the 2010s!). Like Vasicek, they derived an analytical solution for
the risk-free bond pricing equation implied by the short rate process. Cox,
Ingersoll and Ross also went a step further and considered how the same valu-
ation framework could be used to value bond derivatives and, as an illustra-
tive example, derived the arbitrage-free pricing equation for a call option on
a risk-free bond. This was an early illustration of the potential applicability
of stochastic yield curve models to pricing and hedging interest rate deriva-
tives. A huge expansion in the types and volumes of such securities occurred
in the final two decades of the twentieth century. This was facilitated by the
advanced mathematics of these models and motivated further advancements
in the theory and its implementation.
Finally, Cox, Ingersoll and Ross also highlighted the limitations of the
single-factor models that had been exclusively considered by Vasicek and
themselves up to this point. The single-factor model structure had severe
limitationsmost obviously, it implied that the entire term structure was
perfectly correlated. From a derivatives pricing perspective, it would prove
difficult to recover the market prices of derivatives that depended on both
bond price volatilities and correlations with such models. From a risk man-
agement perspective, these models implied that a ten-year liability could be
perfectly hedged by dynamically rebalancing long and short positions in a
three-year and two-year risk-free bonds. These limitations had been recog-
nised by Brennan and Schwartz in the immediate aftermath of the Vasicek
paper, and they published a specific two-factor arbitrage-free interest rate
model in 1979.47 Cox, Ingersoll and Ross developed a more general multi-
factor modelling setting within the arbitrage-free pricing framework first
47
Brennan and Schwartz (1979a).
pioneered by Vasicek. You may recall that the perfect correlation of risk-free
bond returns was an important step in Vasiceks bond pricing derivation. Cox,
Ingersoll and Ross showed that Vasiceks dynamic replication argument could
still work in this multi-factor, decorrelated yield curve worldmore factors
just meant more risk-free bond holdings of different maturities were required
to construct the dynamically riskless portfolio.
These three themes beyond Vasicek that Cox, Ingersoll and Ross devel-
opedmore realistic volatility behaviour for the short rate; the application
of the model to interest rate derivative pricing; and multi-factor short rate
modellingwere themes that were further developed by academics and prac-
titioners over the following few years. Hull and White,48 Black, Derman and
Toy49 and BlackKarasinski50 were important contributions that pursued
these themes in an extremely active research programme that was at least
partly motivated by the huge growth in the trading of interest rate derivatives
in securities markets.
A notable departure from these threads of development arose in a paper by
Heath, Jarrow and Morton (HJM) published in 1992.51 The HJM approach
reconsidered how to describe the stochastic evolution of the yield curve. As
we have seen, short rate models explicitly specified a stochastic process for
the short-term interest rate, and then deduced the arbitrage-free pricing of
all risk-free bonds from that process. So one explicit equation (the short rate
process) determined the behaviour of every point on the yield curve. This
was very parsimonious, but it also restricted modelling flexibility and meant
that there were limited degrees of freedom to fit to calibration targets such as
volatilities of different points of the yield curve (or to simultaneously fit to the
prices of many different types of interest rate derivatives).
Instead of only explicitly specifying a stochastic process for the short rate,
HJM suggested explicitly specifying a stochastic process for every point on
the yield curve. This general framework provided almost unlimited freedom
in specifying the volatility and correlation structure of the different points on
the yield curve. This flexibility was very powerful in the context of interest
rate derivative pricing (which was the intended purpose of their framework).
HJM then found the no-arbitrage conditions that specified the risk-neutral
drifts that must be generated by each forward rate of the yield curve in order
to avoid arbitrage.
48
Hull and White (1990).
49
Black, Derman and Toy (1990).
50
Black and Karasinski (1991).
51
Heath, Jarrow and Morton (1992).
Their paper was closely related to the quantitative work of the Harrison
option pricing papers discussed above. HJM showed that their interest rate
framework could be embedded as the interest rate modelling piece of the
broader mathematical economic framework of Harrison and Pliska. They
were therefore able to make use of the equivalent martingale measure con-
cept of Harrison and Pliska to determine the drift processes for any forward
rate on the yield curve. Where Vasiceks yield curve modelling was inspired by
BlackScholes, the modelling framework developed by HJM was inspired by
Harrison and Pliska: with this paper, yield curve modelling had fully caught
up with and become an integral part of option pricing theory.
Market Efficiency
Our discussion of the emergence of key ideas in financial economics has
focused mainly on the development of economic theories. That is, a range of
theoretical results have been discussed (for example, the Capital Asset Pricing
Model) that have been developed deductively from a set of starting axioms
(investor risk aversion and non-satiation, and so forth). In all cases, these
results were subject to various forms of empirical testing, and such testing
has consistently formed a substantial part of financial economics research
output. But this final section is somewhat different in that it is related to a
stream of work that is intrinsically empirical: it is focused first and foremost
on how well real-life financial markets worknot in theory, but in practice.
In particular, this stream of financial economics considers the informational
efficiency of financial markets prices. Pricing efficiency in this context refers
to how well market prices reflect relevant information and how quickly prices
react to new information. Its empirical nature and its implications for the
possible lack of usefulness of large swathes of financial services practitioners
have made it one of the most contentious areas of financial economics. This
was true many decades ago and it remains true today, particularly as later
research has painted a more complex and nuanced picture of real-life market
behaviour than that implied by market efficiencys major research results of
the 1960s and early 1970s.
We noted some detailed empirical studies of stock price behaviour in the dis-
cussion of option pricing theoryin particular, Osbornes 1959 research that
provided an empirical basis for the use of geometric Brownian motion as a rea-
sonable model of stock price behaviour. There are also some earlier examples of
empirical research that date back to the first half of the twentieth century. But
improvements in the collation of security price data and growing computing
power stimulated a new wave of empirical analysis of security prices in the 1950s.
Besides Osborne, another important example of this empirical work was

provided by Maurice Kendall, who was director of research techniques at the
London School of Economics. Kendall presented a detailed empirical study
of the time series behaviour of financial market prices to the Royal Statistical
Society in 1952.52 The study provided the most detailed statistical analysis
to date of the time series behaviour of stock prices. Kendall considered UK
equity market behaviour over the ten-year period between 1928 and 1938,
and wheat prices on the Chicago Board of Trade between 1883 and 1934. In
both cases he could find little evidence of statistically significant serial correla-
tions at any time-lag. As a trained economist with a faith in rational market
responses to the business cycle, this lack of trend or apparent signal in the
price process confused and alarmed Kendall:
At first sight, the implications are disturbing it seems that the change in price
from one week to the next is practically independent of the change from that
week to the week after. This alone is enough to show that it is impossible to
predict the price from week to week from the series itself The series looks like
a wandering one, almost as if once a week the Demon of Chance drew a ran-
dom number from a symmetrical population of fixed dispersion and added it to
the current price to determine the next weeks price.53
Professor R.G.D.Allen, in his vote of thanks, shared Kendalls despondency,

noting that the papers results were a very depressing kind of conclusion to
the economist.54 However, some speakers from the floor had a different eco-
nomic interpretation for the lack of trend or predictability in the price data.
A Professor Champernowne commented that the low serial c orrelation coef-
ficients found in this particular series may reflect the success with which the
professionals are doing their job,55 whilst Professor Paish elaborated: It seems
inevitable that where prices are based on expectations markets are as likely to
go down as up. If the markets thought they were more likely to go up they
would have gone up already.56
A decade after Kendalls empirical study, Paul Samuelson published a theo-
retical paper, Proof that Properly Anticipated Prices Fluctuate Randomly.57
52
Kendall (1953).
53
Kendall (1953), p.13.
54
Kendall (1953), p.26.
55
Kendall (1953), p.27.
56
Kendall (1953), p.30.
57
Samuelson (1965).
This paper provided some mathematical formality to the intuitions of

Professors Champernowne and Paish: in well-functioning markets, the
no-serial-correlation results of Kendall were exactly what should happen. But
Samuelson was not the first to argue that well-behaved prices should exhibit
random fluctuations. Louis Bacheliers arithmetic Brownian motion process
was almost certainly the first no-serial-correlation model of the stochastic
paths of financial market prices. In his (translated) words: the mathematical
expectations of the buyer and the seller are zero.58
In 1965, Eugene Fama, then a young assistant professor at the University
of Chicago Business School, published a research paper, The Behaviour of
Stock Market Prices.59 As an empirical study of the statistical properties of
stock market prices, this covered some similar ground to Kendall, but was
more comprehensive and wide-ranging. Like Kendall, Fama found that serial
correlations in daily stock returns were generally very low (Fama tested the
returns of US stocks in the Dow Jones Industrial Average over the period
19571962). He also performed some other forms of statistical tests such as
runs tests to provide further evidence of statistical independence of returns
through time.
Famas paper covered a couple of other important topics that would
become increasingly relevant in future years. First, Fama analysed the shape
of the distribution of returns. He concluded that there was strong statis-
tical evidence that daily stock returns had fatter tails than those implied
by a normal distribution. Second, as well as considering the behaviour
of individual stocks, he also analysed empirical data on the returns of
mutual funds (he considered 39 mutual funds over the period from 1950
to 1960). His analysis led him to two conclusions: mutual funds, as a
whole asset class, did not beat the equity market over the period; and no
mutual fund consistently outperformed the others year-on-year through
the ten-year period.
In his 1965 paper, Fama argued that the observations of statistical inde-
pendence of returns through time and the inability of any mutual funds to
consistently outperform the market or each other were both forms of evidence
in support of what had become known as the Random Walk Hypothesis
the idea that market prices varied randomly and unpredictably from one
period to the next. In a further paper, published in 1970, he developed these
58
David and Etheridge (2006), p.28.
59
Fama (1965).
ideas further. This paper, Efficient Capital Markets: A Review of Theory and
Empirical Work,60 is one of the most famous and influential financial eco-
nomics papers ever published. As its name suggests, it was a review of the
by-then abundant empirical analysis of stock market behaviour that had accu-
mulated over the previous 15 years. But it was more than a review. Fama took
those various threads of analysis and wove them into a clear body of evidence
in support of the notion of efficient markets, which he defined as where secu-
rity prices at any time fully reflect all available information.61
The theory of market efficiency was concerned with how prices responded
to information. In Famas crystallisation of efficient markets, he proposed
three levels of market efficiency that corresponded to three different informa-
tion sets: weak-form efficiency, where efficiency meant prices fully reflected all
information in historical price movements; semi-strong efficiency, where effi-
ciency meant prices fully reflected all publically available information (a semi-
strong-form efficient market therefore must also be weak-form efficient as
historical prices were public information); and strong-form efficiency, where
efficiency meant prices fully reflected all information, both public and private
(so a strongform efficient market was also semi-strong and weak-form effi-
cient). His paper reviewed the empirical evidence that had been published in
relation to each form of informational efficiency.
The evidence for weak-form efficiency was naturally found in the statistical
testing of historical price data. These tests took two broad forms: testing for
statistical independence of returns through time (mainly by serial correlation
testing such as that done by Kendall and Fama); and testing the profitability
of mechanical trading rules (the idea being that any excess profitability of
such rules would not be consistent with efficient markets). Famas 1965 paper
included some analysis of these mechanical trading tests, and he published
a paper in 1966 with Marshall Blume with further testing of such rules.62
Famas review of the evidence relating to weak-form efficiency allowed him to
conclude that the results are strongly in support.63
The empirical evidence for semi-strong market efficiency was largely based
on analyses of how market stock prices reacted to major public announce-
ments of relevant information such as earnings statements and stock splits.
The basic idea was that if the market was efficient the price impact of these
announcements would be immediate, and subsequent expected returns would
60
Fama (1970).
61
Fama (1970), p.383.
62
Fama and Blume (1966).
63
Fama (1970), p.414.
therefore be unaffected by the announcement. Fama again concluded that

these studies invariably provided support for the semi-strong form of the effi-
cient market hypothesis. The implications of the semi-strong hypothesis are
the most provocative to investment professionals as it implies that active fund
management cannot be expected to outperform the market except by luck.
This is consistent with Famas survey of mutual fund performance in his 1965
paper. Such studies also have a longer historyin 1933, the US economist
Alfred Cowles published a paper which showed that buying-and-holding
would tend to outperform the recommendations of stock market forecasters.64
Evidence in support of the semi-strong hypothesis can provide an intellec-
tual basis for market indexing or passive investment managementa form of
investing that has rapidly grown in popularity since the 1980s.
In considering the strong form of market efficiency, Fama conceded that
there was evidence that corporate insiders have monopolistic access to infor-
mation that is not in the share price. But even in this case he argued that the
investment community is unable to access and use such information to out-
perform the market. Overall, Famas conclusion was emphatic: In short, the
evidence in support of the efficient markets model is extensive and (somewhat
uniquely in economics) contradictory evidence is sparse.65
Inevitably perhaps, however, empirical evidence contradicting the effi-
cient markets hypothesis quietly started to accumulate in the decade follow-
ing Famas emphatic declaration of efficient markets victory. The watershed
moment arrived in 1978 when Professor Michael Jensen, a leading financial
economist of the period, edited a special edition of the Journal of Financial
Economics that was dedicated to reviewing this stream of research.66 Amongst
other studies, this research included several analyses of the returns on a diverse
range of mechanical trading strategies (including securities such as investment
trusts and exchange-traded stock options). If strategies were identified that
earned statistically significant excess risk-adjusted returns (after trading costs),
this would be regarded as evidence inconsistent with weak-form market effi-
ciency. Unlike in Famas mechanical trading tests of the 1960s, the edition ten-
tatively concluded that several such strategies could deliver excess risk-adjusted
returns. But this type of study raised an interesting question: did excess risk-
adjusted returns look good for these strategies because markets were mispricing
assets or because the theoretical models for assessing the risk-adjusted required
returns were wrong (or were missing some features that are important to these
64
Cowles (1933).
65
Fama (1970), p.416.
66
Jensen (1978).
complex strategies)? This left some ambiguity in the conclusions which helped
to shape the future direction of financial economics research.
Jensens special edition helped to create an environment within the finan-
cial economics profession where challenge to the accepted wisdom of perfectly
functioning financial markets was an accepted part of academic orthodoxy. It
ushered in a new era of empirical research in financial economics where the
identification of potentially irrational market behaviour was suddenly highly
in vogue.
In 1981, Robert J.Shiller published a provocative paper where he argued that the
volatility of stock market returns was much, much higher than could be explained
by changes in rational expectations for levels of future dividend pay-outs.67 Using a
dividend discount model for equity market valuation, he showed how, under some
assumptions about the stochastic properties of the dividend pay-out process, a rela-
tionship between the year-on-year volatility of dividend pay-outs and year-on-year
volatility of equity price changes could be established. Shillers long-term empirical
analysis of dividend pay-outs and stock market volatility in the USA implied that
market volatility was five to thirteen times too high to be attributed to new informa-
tion about future real dividends.68 In deriving this analysis, the dividend discount
model he used assumed a constant real required return. He inverted the analysis
and considered how volatile the real discount rate would need to be to generate the
observed level of market volatility. He found it would need to have an annual stan-
dard deviation of 47 %, which he dismissed as economically unfeasible.
Shillers work generated considerable academic controversy and prompted a
notable response from Robert Merton, one of the financial economics profes-
sions established leaders of the period. In a paper with Terry Marsh published
in 1986,69 the authors argued that Shillers conclusions were highly d ependent
on his assumed form of stochastic process for dividends. Marsh and Mertons
key point was that firms managers liked to smooth dividend pay-outs as
much as possible. But, as was shown by Modigliani and Miller decades ear-
lier, the rational or intrinsic value of the firm should be determined by the
performance of the firms underlying assets, and not by its dividend policy. If
investors understood that managers preferred to smooth dividends over time,
then they would be more sensitive to changes in dividend pay-outs (if the firm
still had to reduce dividends even though management prefer to pay stable
dividends, this signalled things must be pretty bad). Their general point was
that inferring rational levels of return volatility from observed dividend policy
67
Shiller (1981).
68
Shiller (1981), p.434.
69
Marsh and Merton (1986).
was very difficult because dividend policy did not necessarily have a direct
relationship with the true value of the firm. To prove this, they showed that
the opposite statistical conclusion could be reached from Shillers data when
they specified an alternative form of stochastic process for dividend pay-outs
(which they argued fitted better to empirical dividend pay-out behaviour).
Despite Mertons protestations, the genie was out of the bottle. Other lead-
ing financial economists followed Shillers lead and produced further analysis
to support the argument that volatility in stock market returns was inexpli-
cably high. Richard Roll, in his presidential address to the American Finance
Association in 1987,70 presented an empirical analysis that argued that, even
with the benefit of hindsight, 60 % of US equity stock market daily price
volatility was inexplicable (in the sense that the price variation in a firms
stock could not be explained by observable new information relating to the
firm, its industry or general economic and market impacts). This was not the
order of magnitude of excess volatility that Shiller had reported, but it was
perhaps all the more plausible for that. Rolls address opened the door to pos-
sible behavioural explanations: Several authors have suggested that volatility
of asset prices can be better explained by psychological factors, fads, etc., than
by information. The results above are actually consistent with such a view.71
If Shiller and/or Roll were right that short-term equity volatility was inex-
plicably higher than could be justified by changes in fundamentals, what did
that imply about long-term equity behaviour? If extra volatility was continu-
ously feeding into stock returns without any form of self-correction, equity
prices would become infinitely dislocated from underlying economic reality.
Eugene Fama, the economist more associated with efficient markets than any
other, worked with another Chicago economist, Kenneth French, to provide
some further insights into the empirical behaviour of longer-term equity
returns. Fama and French published two significant papers on this subject
in 1988.72 The first paper, Permanent and Temporary Components of Stock
Prices, identified statistically significant mean-reversion (negative serial cor-
relation) in historical (19261985) US stock market returns over three- to
five-year horizons. Previous tests of serial correlation in stock market returns
such as Kendalls had used equity data series of a more limited size (Kendall
used a total equity data horizon of ten years). Fama and Frenchs more com-
prehensive data analysis suggested there was a noteworthy cumulative effect
which was highly significant over longer holding periods.
70
Roll (1988).
71
Roll (1988), p.565.
72
Fama and French (1988a); Fama and French (1988b).
Seen alongside the work of Shiller, Ross and others, Fama and Frenchs
research suggested that short-term equity market volatility was excessively high,
and that some of this excess or temporary volatility was removed over time
by a form of correction mechanism in equity market prices (which manifested
itself statistically as a material mean-reverting component in the price process).
Their second paper of 1988, Dividend Yields and Expected Stock Returns,
took this analysis further: if mean-reversion was an important element of long-
term equity market behaviour, was it possible to observe at any given point in
time whether this mean-reverting component of returns was above or below its
mean level? Fama and French suggested this was possible and indeed trivially
easy: dividend yields appeared to be meaningful predictors of long-term equity
performance. High dividend yields predicted strong returns over the following
two to five years, low dividend yields predicted the opposite.
From a market efficiency perspective, this was a profound challenge to even
the weak-form of the efficient market hypothesis. But there were some caveats.
First, whilst long-term expected returns did vary with the starting level of the
dividend yield, it was unclear whether this was mispricing resulting from fads,
bubbles or some other form of irrational behaviour, or whether this reflected
rational changes in required returns due to time-variation in the riskiness of
equities or in investor risk appetite. Furthermore, an inevitable consequence
of analysing longer-term empirical behaviour is that there is a smaller sample
size to observe. As the leading twenty-first-century financial economist John
Cochrane has pointed out, when dealing with such long-term trends, we may
really only have a few observable data points:
What we really know is that low [stock] prices relative to dividends and earnings
in the 1950s preceded the boom market of the early 1960s; that the high p rice/
dividend ratios of the mid-1960s preceded the poor returns of the 1970s; that
the low price ratios of the mid-1970s preceded the current boom.73
This limited volume of empirical data constrained the degree of consensus

reached within the financial economics profession on the topic of long-term
security price behaviour, and it continues to do so today. Robert Shiller and
Eugene Fama received the 2013 Nobel Prize in Economics for their work in
this field (along with Peter Hansen). Their acceptance speeches featured an
exchange of views that highlighted how much work remained to be done to
find a consensus explanation for their empirical findings.
Cochrane (2005).
73
Whilst it took until the 1980s for the notion of mean-reversion and time
diversification to gain academic credence, it has arguably been part of inves-
tor intuition for as long as equity markets have existed. As custodians of long-
term liabilities, mean-reversion in long-term returns doubtless played a role
in first attracting life offices and their actuaries to equities as an asset class in
the 1930s. For example, in a letter to F.C.Scott, the managing director of the
Provincial Insurance Company, in June 1938, John Maynard Keynes wrote:
A valuation at the bottom of the slump tends to bring out an unduly unfavour-
able result as against an investment policy which on the whole avoids equities;
since it allows nothing for the nest egg in hand arising out of the fact that such
a valuation is assuming in effect that one has purchased a large volume of equi-
ties at bottom prices Investment policy which is successful in averaging through
time will produce the same good results as insurance policy which is successful in
averaging through place [emphasis added].74
Time diversification can only arise if a component of the price change pro-
cess is temporary. As we shall see in Chap. 5, this idea was embedded in
how actuaries modelled and measured equity risk in the context of long-term
liability business in the late twentieth century. This was clearly inconsistent
with the financial economics of the 1960s and 1970s. It was not as inconsis-
tent with the financial economics of the 1980s and beyond as actuaries have
sometimes been led to believe.
74
Keynes (1983), p.67.
5
Life Offices After theSecond World War:
TheUnderwriting andManagement
ofFinancial Market Risk (19522004)
The history of actuarial thought in the British life assurance sector over the
second half of the twentieth century is tumultuous. This was increasingly the
case as the century wore on, reaching something approaching a crisis by the
centurys end, from which the profession started to reorientate itself in the
early 2000s. The root causes of the late twentieth-century challenges initially
emerged earlier in the century and have already been briefly noted. The most
important of these was the increasing trend, started in the 1920s and 1930s,
of abandoning the strictly risk-averse investment disciplines of nineteenth
century Bailey in order to pursue greater investment in equities and other
risky asset types.
This apparent increase in risk appetite could be viewed in the context of a
broader long-term trend in with-profits policy design towards it being unam-
biguously configured as a long-term savings vehicle as opposed to a more
balanced mix of savings and life assurance protection. As a long-term savings
product provider, with-profits funds were exposed to increasing competition
from other sectors of the financial services industry such as banks and asset
managers. The stress of competition doubtless placed commercial pressures
on actuaries, whose role was to determine the charging, reserving require-
ments and bonus policies of life products.
From an intellectual perspective, the radical developments in relevant
related disciplines that occurred over the second half of the century created
opportunities and threats for the actuarial profession. Financial econom-
ics, the application of advanced quantitative techniques to finance and new
modelling possibilities created by advances in computing technology each

DOI10.1007/978-3-319-33183-6_5
contributed to a revolution in financial and risk management practices in

other parts of the financial and corporate sectors. Would these developments
make traditional actuarial skills obsolete? Could the profession embrace these
new ideas and capabilities and add them to their professional toolkit? This
represented a formidable challenge for the British profession. As we have seen,
actuarial concepts and their applications evolved significantly over the 100
years between 1850 and 1950. The core technical skills required in actuarial
work, however, had not changed fundamentally over that period. The tried and
tested actuarial syllabus of the mid-twentieth century was not easily adapted
to the radical and technical developments in finance, quantitative techniques
and computer modelling that would emerge in the following decades.
The challenge was cultural as well as intellectual. The British actuary of this
era stood resplendent in his status as a purveyor of techniques too complex
and opaque to be understood by anyone else; his professional judgement was
beyond challenge; his track record was without blemish. He was not minded
to be told that his methods were antiquated or that his thinking was flawed.
An intellectual and cultural clash was afoot.
As ever, this is a story replete with remarkable actors, and it is perhaps all
too easy for hindsight to blind us to the real challenges and conflicts they
faced as they made the progress they did.
quities andWith-Profits: Smoothing

E
andGuarantees (19521976)
Chapter 3 noted how actuarial thinking on life office investment strategy
began a journey in the 1920s, away from the highly conservative nineteenth-
century tenets of Bailey. By the late 1940s, the Dalton cheap money era had
further prompted British life offices to move materially into equity and prop-
erty investments in an effort to maintain their investment yields in excess
of those that had been guaranteed in with-profit premium rates. This asset
allocation trend continued unabated into the 1950s and 1960s. By 1951,
2025 % of life office assets were invested in equities and property;1 by 1961
this allocation had grown to over 30 %.2 Life offices had a mixture of both
with-profit and without-profit liabilities. The without-profit business was
effectively backed by bonds, and so the with-profit businesss equity/property
exposures were significantly higher than the above figures. By 1981, Frank
Dodds (1979), p.176.
1
Dodds (1979), p.50.

2
5 Life Offices After theSecond World War... 189
Redington noted that for some British with-profit funds the e quity/property
allocation was 100 %.3 In the decades following the Second World War,
British with-profits business had transformed into something different, and
the actuarial thought of British life actuaries had to catch up.
There were a couple of key (and related) questions that required actuarial
answers. How did substantial equity/property investment impact on life office
solvency assessment? And how should with-profit bonuses be distributed in
the presence of significant equity/property investment? Most actuarial energy
was initially invested in the second of these questions. In the 20 years follow-
ing the end of the Second World War, equity markets performed remarkably
well (especially in nominal terms). This was driven by both a high rate of
dividend growth and a fall in equity market dividend yields. Between 1950
and 1960, dividend pay-outs more than doubled and market values increased
by a greater amount. Long-term interest rates steadily increased during this
era (from a low of 2.6 % in 1946 to 6.8 % by 1966). Life office solvency was
therefore generally not felt to be a major concern. Conversely, there was a
pressing need to deal with the embarrassment of riches that with-profit funds
had found themselves holding. This was made particularly pertinent by the
increasing competition from other forms of savings vehicle such as unit trusts.
The crux of the bonus distribution challenge was that the reversionary
bonus system did not recognise any equity/property capital gains until they
were crystallised in asset book values, and the writing-up of book values prior
to the sale of the asset was generally viewed as actuarially distasteful. In the
competitive environment, this was a fundamental issue. The only way equity
returns naturally fed into the assessment of distributable surplus was through
the dividends received. By 1960, the reverse yield gap had emerged: equities
generated a lower (initial) income than risk-free bonds. Distributable surplus
would therefore be reduced by increasing equity allocations, even following
a period when the equity holdings had materially outperformed bonds in
terms of market value growth. Moreover, during this time, market practice
was such that actuaries were under significant pressure not to reduce bonus
rates other than in the most exceptional circumstances. This further incentiv-
ised the actuary against releasing unsustainable rates of equity market value
increase into distributable surplus. As a result, with-profit pay-outs started to
significantly lag comparable unit trust pay-outs, and life office estates increas-
ingly bulged with undistributed and unrecognised equity market value gains.
This situation was unsustainablethe with-profit bonus system was
not fit for the purpose of distributing surplus from equity-dominated asset

3
a llocations. By the mid to late 1950s, British life actuaries started to recognise
that a more flexible bonus system would be necessary in order to distribute
equity capital gains equitably amongst the policyholders whose premiums
had generated them. It was crucial that this was done in a way that did not
create the expectation that such distributions could be predictably delivered
every year. This required two new devices: a special or terminal bonus that was
distinct from the reversionary bonus and that did not have the same policy-
holder expectations for stability attached to it; and an acceptable method for
writing up book values of equity/property assets to market values in order to
transfer some gains into distributable surplus. It is difficult to overstate how
counter to traditional British actuarial principles these developments were.
But several papers were published between 1959 and 1976 that progressively
established the new orthodoxy.
An Institute paper presented to Staple Inn in 1959 by the Equity and Law
actuary Norman Benz was an important early landmark in establishing these
radical ideas in mainstream actuarial thought.4 Those ideas were given further
support in another Institute paper written by P.E. Moody in 1964.5 These
papers reflected the growing recognition that intergenerational policyholder
equity and competition in the long-term savings market both demanded that
policyholders receive the capital gains generated by their investments. This
naturally led to an increasing embrace of the contribution principle of bonus
distribution that had been well-established in the USA for so many decades.
For example, in an influential 1968 paper,6 Skerman argued that the dis-
tributed profits should be allocated between policyholders in relation to the
contribution which their policies have made to them.
However, in the 1950s and 1960s (and beyond), there was still a prevalent
view amongst British actuaries that with-profit funds should deliver something
more stable than what was directly generated by the market returns of their
invested premiums. Both Moody and Benz advocated a smoothed distribution
of equity gains to policyholders. The concept of using the with-profit funds
estate to deliver a smoothed equity return to policyholders would remain
a permanent central tenet of British actuarial thought. The 1976 report of
the Institute of Actuaries Working Party on Bonus Distribution with High
Equity Backing stated that the general view of the working party was that
conventional with-profits business should smooth out the fl uctuations in
4
Benz (1960).
5
Moody (1964).
6
Skelman (1968).
investment return and that the estate should be used to level out fluctua-
tions in experience.7
Benz and Moody explored how to smooth the distribution of equity capital
gains to with-profit policyholders. In both papers there was a particular wari-
ness about distributing equity capital gains that arose through falls in yields
rather than through growth in dividend pay-out levels. Both Benz and Moody
advocated the idea of using an actuarial estimate of the future dividend growth
rate to determine a steady writing up of asset book values and hence release
to distributable surplus, irrespective of how the market yield was behaving. In
his paper, Skerman agreed that market value changes arising due to changes
in equity dividend yields should not enter into distributable surplus. In the
Staple Inn discussion of Skermans paper, A.C.Stalker advocated the use of
the historical average dividend yield to determine equity book value write-ups.
Whilst such approaches could be substantially dislocated from market value
movements, the implications of the Fama and French empirical equity studies
that would emerge some 25 years later arguably offer some degree of intellec-
tual support for these ideas. But such an approach was not without practical
difficulty: did with-profit funds have the capital to underwrite larger distribu-
tions than those implied by market value gains when equity market yields were
unusually high? Did policyholders understand that they may not get a mate-
rial portion of their equity capital gain when markets performed very well?
Whilst Benz and Moody advocated a quantitative method for developing a
smoothed distribution of the equity capital gains, Skerman did not welcome
such a mechanical approach to valuing assets. Instead, he argued that the
release of equity gains into distributable surplus should simply be a matter of
actuarial judgement as part of the actuarys role in the proper steering8 of the
with-profit fund. Such an approach placed a huge amount of discretion in the
hands of the actuary, even by British actuarial standards. In such a system, an
actuarial discipline that balanced policyholder security against intergenera-
tional equity and prudence against competitive marketing pressures would be
critical to the long-term success of with-profits as a popular and sustainable
long-term savings vehicle.
Life office practices evolved concurrently. Leading British with-profit funds
started to use terminal bonus as an additional degree of freedom to deliver
more equitable pay-outs to with-profit policyholders in the late 1950s. The
Prudential, with Frank Redington as its chief actuary, was a leader in this
innovation, introducing terminal bonuses in 1956. The terminal bonus was
Kennedy etal. (1976), p.12.

7
Skelman (1968), p.72.

8
an almost universally adopted mechanism amongst British with-profit offices

by the end of the 1960s. However, such was the success of equity markets
during this era that a real risk emerged of the terminal bonus becoming
established in the minds of policyholders and salesmen as just as stable and
predictable as the reversionary bonus. The reasonable expectations of policy-
holders were enshrined in UK legislation by the Insurance Companies Act
of 1974, and this further concentrated actuarial minds on bonus policy. But
the extreme equity market experience of 1973 and 1974 prompted several
offices to reduce terminal bonuses and one or two offices even reduced them
to zero.9 This suggests the actuary was able to use the terminal bonus mecha-
nism broadly as intended.
What of actuarial thought on the appropriate level of equity allocation in
with-profit funds and its implications for solvency? When equity investment
was first introduced to UK with-profit funds in the 1920s and 1930s, this
topic received little consideration. Equity allocations were small and solvency
buffers were usually large, and there was little discussion of what upper limit
should be placed on the equity allocation. Rayness influential papers of 1928
and 1937 said virtually nothing on this question. In the closing remarks in the
discussion of Rayness 1937 paper, something is finally mentioned in relation
to liabilities when George Recknell, the actuary at Keyness National Mutual,
suggested:
Concerning the percentage of ordinary shares which should be purchased there

was something to be said, as a very rough rule, for investing on more or less con-
ventional lines to the extent of the funds needed to support the contractual lia-
bilities, while leaving the surplus free to invest in ordinary shares and real estate.10
Haynes and Kirtons 1952 paper had similarly advocated Recknells form
of asset-liability matching: fixed liabilities (sums assured, including accrued
reversionary bonuses) should be matched by bonds, and the excess funds
could be invested in equities and property. A more ambitious discussion of
the equity asset allocation in a with-profit fund was developed in 1957in an
influential Institute paper written by J.L. Anderson and J.D. Binns,11 who
were, respectively, the actuary and investment secretary of Scottish Widows at
the time. Anderson and Binns suggested that with-profit funds should invest
more in equities than the maximum level implied by the above principle that
guaranteed benefits were matched by bonds:
9
Kennedy etal. (1976).
10
G.H.Recknell, in Discussion, Raynes (1937), p.505.
11
Anderson and Binns (1957).
Suppose that k is the maximum depreciation on present market values which is

envisaged. Estimate the value of a function we shall call the remainder R, i.e.
the excess of the total assets at market value over the liabilities on a gross pre-
mium basis It can then be argued that R/k can safely be invested in equities
provided the balance of the fund is reasonably matched.12
The full-matching approach advocated by Recknell, Haynes and Kirton was

equivalent to setting k=1, i.e. assuming equities could fall to zero. By set-
ting k to less than 1, the with-profit fund opened up the possibility that poor
equity market performance would result in the office failing to meet its guar-
anteed liabilities: of course, the probability of this possibility depended on the
assumed size for their k parameter (and on how quickly the fund could shift
out of equities if it wished to as market values fell). Their logic could be viewed
as a precursor to a modern portfolio insurance strategy, which is typically
couched in the logic of Black-Scholess dynamic hedging (though Anderson
and Binns gave no indication that it was their intention for with-profit funds
to reduce their probability of insolvency through very frequent rebalancing of
the equity allocation to the new level implied by market value changes).
Anderson and Binns noted that the Dow Jones Industrial Index had fallen
by over 80 % during 19291932, but dismissed this period as exceptional,
and suggested using a k parameter value of 0.6. In the discussion of the paper
that followed at the Institute meeting, their proposal excited little alarm,
except for one lone voice, S.H.Cooper, who noted:
If he had correctly understood the remainder, it followed that any deprecia-

tion in excess of the limit envisaged would absorb the whole of the free reserves
of the fund and would encroach upon the cover required for the basic liabili-
ties of the fund. In that context it seemed to him that they should be concerned
with possibilities rather than probabilities, and he had been rather surprised to
see the authors dismiss the American experience of 192932 as exceptional.13
Despite such occasional protestations, life office equity/property allocations

continued their inexorable rise in the following years. By 1976, allocations to
equity and property were around 50 % of life office assets.14 This excited little
actuarial concern. Conventional actuarial wisdom embraced equities as an
inflation-proof long-term asset which life offices were uniquely equipped to
hold on behalf of policyholders due to their large estates and the ability it gave
12
Anderson and Binns (1957), p.125.
13
S.H.Cooper in Discussion, Anderson and Binns (1957), p.143.
14
Dodds (1979), p.50.
them to protect policyholders from equity markets irrational short-term mar-

ket movements. In concluding his history of the first 200 years of Equitable
Life, Ogborn in 1962 mused:
What seems important at the time of writing is the greater freedom of invest-
ment which the life offices have felt both necessary and desirable, in the current
economic climate, for the purpose of giving their policyholders some share of
profits from the expansion of industry and some protection against the possible
effects of further inflation, should it recur.15
By the 1960s, British life actuaries had learned to feel quite sanguine about
the financial market risk exposures that their businesses were underwriting in
ever-increasing scale. This contrasted with the technical sophistication that
was developing in the pricing and management of financial market risk out-
side the actuarial world. An interesting next 40 years would ensue for the
profession.
quity-Linked Maturity Guarantees: Towards

E
Risk-Based Solvency (19711982)
Unit trusts and unit-linked business emerged as a major competitor to with-
profit policies in the UK long-term savings market in the 1960s. By the early
1970s, unit-linked business had become a significant business line for many
British life offices. These products were generally much simpler than with-
profit policies, and their financial management required less actuarial input.
An interesting complication arose however, when life offices attempted to dif-
ferentiate from the competition by providing maturity guarantees with their
unit-linked fundstypically these would provide a minimum guaranteed
pay-out of a return of the premium(s) paid at the specified maturity date. The
maturity date typically varied between ten and 25 years. These guarantees were
provided on underlying funds that were typically invested entirely in equities.
The pricing and reserving for these guarantees was not initially subject to
any actuarial sophistication. In some cases, the reserve was merely retrospec-
tive, i.e. an accumulation of the (small) additional premiums charged for the
provision of the guarantee. Life actuaries found themselves having to catch
up with their products, and by the second half of the 1970s they were further
prompted into action by the industry regulators stated intention to impose
Ogborn (1962), p.256.

15
some prescriptive measures on reserving for these guarantees. Somewhat per-

versely, the relative simplicity of the unit-linked product created an actuarial
risk management challenge. With-profit funds discretionary levers of invest-
ment policy, bonus policy and policyholder cross-subsidy were not available
to the actuary to mitigate the guarantee risk embedded in unit-linked busi-
ness. This meant that the office bore the full exposure of the risks that they
wrote in the policy; and the resultant transparency of those risks meant that
their measurement and modelling were a lot more straightforward than was
the case for with-profit business.
In 1968, at the Staple Inn discussion of Skermans with-profit bonus paper,
Jim Pegler (who at the time was only months from becoming the Institute
President) raised the general topic of how to balance security with surplus
distribution. Pegler raised the possibility of using a probabilistic basis for set-
ting reserves. This is notable as perhaps the earliest such reference in a British
actuarial paper, though Pegler recognised it would be challenging to imple-
ment and stopped short of advocating such an approach:
The author drew attention to the over-riding need to ensure security for policy-
holders they had to find a way of arriving at a reasonable degree of security,
to which absolute priority should be given He imagined that theoretically
they could approach the problem from the point of view of the Theory of Risk
and estimate a probability of ruin. But even supposing that a reasonably accu-
rate numerical figure could be obtained, he doubted whether it was helpful in
practice to consider whether a probability of, say, 1in 1,000 or 1in 10,000
should be aimed at; he was choosing figures more or less at random, as he had
little idea of the order of magnitude which was relevant.16
The application of the probability of ruin concept to a real-life with-profits

business and all its complexities was a daunting challenge in 1968. But for unit-
linked business, it was much more feasible. As actuarial thinking caught up
with unit-linked maturity guarantee practices, it was natural that a probability
of ruin approach to setting adequate maturity guarantee reserves should at least
be considered. The Prudential actuary Sidney Benjamin pioneered the develop-
ment of this train of thought with a paper that was first discussed at Staple Inn
in 1971. The meeting was described by one of the actuaries present as by far
the stormiest I have ever attended.17 Indeed, it was so stormy that Skerman, as
President, closed the meeting and classed the discussion as confidential so that
16
Pegler in Discussion, Skelman (1968), p.94.
17
P.Smith in Discussion, Corby (1977), p.274.
its content would never be published.18 The paper was never published by the
Institute. It was only formally presented and published five years later in 1976
at the twentieth International Actuarial Congress in Tokyo.
What was it about Benjamins work that provoked such a convulsive reac-
tion within the actuarial profession? In stark terms, the British actuarial pro-
fession had not been trained or educated to think about risk. Or, to put it
more kindly, they had been trained to assume that an insurance office could
diversify away risk. This worked well, up to a point, for mortality and other
insurance risks. But it did not work for financial market risk, which was inher-
ently non-diversifiableall policyholders had guarantees on the performance
of funds that would behave similarly. There was an elegant equivalence in the
retrospective and prospective reserves produced by charging for expected costs
and accumulating net premiums. But this was of no use when a contingency
reserve was required for a non-diversifiable risk that could not be fully funded
by the premiums received (at least not without a highly sophisticated dynamic
investment strategy which was far from the actuarial minds of the time). With
non-diversifiable risk, an adequate prudential reserve may be an order of mag-
nitude greater than the value of the premiums accumulated from charging
for the risk. Benjamins paper put the actuarial profession face-to-face with
the realities of non-diversifiable risk, taking it into uncharted waters. It was a
seminal moment in the history of the British profession. With the benefit of
hindsight, it marked the start of a 30-year struggle to modernise its thinking
on the measurement and management of financial market risk.
In the framework of Benjamins paper, the required maturity guarantee
reserve was calculated as the discounted present value of the maturity guar-
antee shortfall (i.e. the shortfall in the final underlying fund value relative to
the guaranteed maturity proceeds) assessed at a specified percentile level (the
accepted probability of ruin). This calculation required two key inputs: an
assumption for the acceptable probability of ruin; and a probabilistic model
of the behaviour of the underlying equity funds that would allow the prob-
ability distribution of the guarantee shortfall to be assessed. This reserve could
be funded by the discounted present value of the premiums that would be
charged for providing the guarantees. If the reserving requirement exceeded
this present value, this excess would have to be funded using the offices capital.
Benjamin developed his probabilistic equity model by considering the his-
tory of annual UK equity returns for the 51 years from 1919 to 1970. He con-
ducted various statistical tests on the time series and concluded that there was
not statistically significant evidence to reject the null hypothesis that the annual
S.Benjamin in Discussion, Corby (1977), p.277.

18
returns were independent. Given this result, he argued that a probability distri-
bution for the n-year equity index could be generated by randomly sampling
(with replacement) n points from the 51 historical annual return data points.
He then suggested that the probability of ruin be set at 2 %, and estimated
the reserve required by generating 50 simulation paths and using the one that
produced the largest reserve.
This was all undoubtedly methodologically quick and dirty, but it captured
the essence of the problem and it produced insightful results. Benjamins cal-
culations implied that the mean ten-year guarantee shortfall for annual pre-
mium business paying a premium of one and with a maturity guarantee of ten
was 0.11 (i.e. 1.1 % of the guarantee), and that the second percentile was 4.5
(i.e. 45 % of the guarantee). Discounting at a risk-free yield of 2.5 % implied a
starting reserve of 39 % of the maturity guarantee would be required. Some of
this reserve could be funded by accumulating the premiums charged for grant-
ing the guarantee, but, at typical charging rates, this still implied that the office
would need to fund an additional reserve of over 30 % of the maturity guaran-
tee. Benjamin himself wrote that this reserve was unexpectedly high and sug-
gested that it meant the contract is probably not a commercial proposition.19
There was something of the spirit of Bachelier in Benjamins research, at
least during this stage in his career. He had a talent for taking ideas such as
game theory or risk theory and envisioning how they could be newly applied
to actuarial problems. But his work needed further development and refine-
ment to be rigorously implemented and made accessible to the wider profes-
sion. A couple of papers quickly followed the Tokyo Congress that provided
this development: a substantial paper by W.F.Scott20 and a shorter note by
David Wilkie21 that were both published in 1977.
Scott followed Benjamins general probabilistic risk-based framework but
his implementation of the reserving assessment differed in some key ways.
Most significantly, Scotts analysis of historical equity returns led him to
reject the assumption that annual equity market returns were independent
over time. Scott used the same data as BenjaminUK equity market returns
from 1919 to 1970 as calculated from the de Zoete equity indexbut Scott
identified statistically significant serial correlation of 0.30 at a two-year lag.
Benjamin never explicitly tested serial correlation in his analysis of the inde-
pendence of historical returns but instead used runs tests and a few other
forms of statistical test of independence. Scott tested lags of one to four years,
19
Benjamin (1976a), p.25.
20
Scott (1977).
21
Wilkie (1977).
and whilst the serial correlation was negative for each of two, three and four
years, it was only the two-year lag correlation that was statistically significant
from zero. Nonetheless, Scott concluded this was sufficient evidence to reject
the independence hypothesis. This analysis is noteworthy for its early docu-
mentation of evidence of statistically significant negative serial correlation in
long-term empirical equity market returns. It anticipated Fama and Frenchs
work by more than a decade (and there is nothing to suggest Fama and French
were aware of this work). Scotts explanation of this negative serial correlation
also anticipated, at least in a heuristic way, Shiller and Rolls excess short-term
volatility arguments. Scott argued that the extraordinary UK equity market
volatility of 1973 to 1976 (during which time dividend yields moved from
3 % to 12 % to 5 %) could only be explained by concluding the market
panicked and then corrected itself 22 (though he was not the first actuary or
market participant to suggest financial markets could behave in such a way).
The stochastic modelling skills of the profession at this time were still
embryonic, and Scott had a preference for a simple model that would not
require Monte Carlo simulation methods in its implementation. So, rather
than developing a stochastic model of equities with negative serial correlation
in returns, he instead used a standard random walk model but with an annual
volatility that was lowered to reflect the long-term volatility that would arise
in the presence of the assumed negative serial correlation. Specifically, he used
an annual equity return volatility assumption of 10 % rather than the 19 %
that he found in the annual historical returns data. He did not provide any
statistical basis for the substantial size of this adjustment.
Scotts rejection of the independence of equity returns through time and
the consequent assumption that long-term returns would therefore be materi-
ally less volatile naturally resulted in significantly reduced reserving require-
ments for long-term maturity guarantees. Scotts modelling implied that the
ten-year annual premium return-of-premium maturity guarantee required
an initial reserve of 15 % of the guarantee rather than the 30 %-or-greater
estimate produced by Benjamin. And that was with a 0.5 % probability of
ruin instead of Benjamins less ambitious 2 % assumption.
Scotts paper was primarily focused on reserving, but he also considered how
to set the premium that should be charged for provision of the guarantee. He
argued the guarantee premium should be based on the expected guarantee
shortfall discounted at the risk-free yield, appealing to the equivalence prin-
ciple. This implied very low premiums for the guaranteefor example, the
expected cost of the ten-year guarantee was estimated at 0.5 % of the guarantee
Scott (1977), pp.37475.

22
(a Black-Scholes put option price would have implied 45 %). He also dis-
cussed how the reserving requirement could be mitigated and considered several
approaches. None of them, however, recognised that the guarantee was a put
option and that a body of theoretical work and trading practice had recently
been developed on how to price and hedge such options.
Wilkies paper further refined Scotts analysis. Again, the most significant
development was in the assumptions underlying the stochastic modelling of
the equity market return. Wilkie confidently grasped the nettle that Scott
had avoided: he developed an explicitly auto-regressive equity model that was
fitted to the serial correlations observed in the historical returns and that was
implemented using simulation. This modelling suggested Scotts downward
adjustment of annualised volatility from 19 % to 10 % was too great. Wilkie
also updated the historical dataset to include returns up to 1977, thereby
including the exceptional volatility period of 19731976. These resulted in
reserving estimates that were closer to Benjamins than Scotts.
Wilkie also criticised Scotts argument that the premium charged for the
guarantee should be calculated on the basis of expected cost, arguing that
shareholders would require a return on the capital that they needed to use to
support the business (Benjamins paper had made a similar argument). Wilkie
suggested that shareholders would require a return of 2 % in excess of the risk-
free rate on their capital on the grounds that the riskiness of their exposure
was comparable to reasonable quality corporate bonds. He argued this cost
should be added to the expected guarantee shortfall. Whilst such an approach
was more sophisticated than Scotts, it was still far from utilising the available
economic insights on option pricing that were published a few years earlier by
Black, Scholes and Merton.
Benjamin, Scott and Wilkie differed in their recommended modelling
assumptions, but they all shared a substantial common ground that was a radi-
cal departure from traditional actuarial thought: maturity guarantees on unit-
linked business ought to require a contingency reserve that should be assessed
by specifying a probability of ruin, , and then using a stochastic model of
the underlying unit funds to determine how much reserve is required to fund
guarantee shortfalls with probability (1-).
This rapid actuarial research output of 1976 and 1977 had one curious
institutional feature: none of it was published by the Institute of Actuaries.
The papers of Scott and Wilkie were both Faculty papers, whilst, as we
have seen, Benjamin had been published under the auspices of the twenti-
eth International Actuarial Congress. The Institute had not touched matu-
rity guarantees since the Benjamin Staple Inn debacle of 1971. This was to
be addressed by an Institute working party led by F.B.Corby, but it never
ublished its findings due to problems in resolution of fundamental points.23

p
So Corby went ahead and published his own paper with his personal views
in 1977.24
Corby rejected the probability-of-ruin-and-stochastic-model reserving
approach advocated by each of Benjamin, Scott and Wilkie. He argued that
it was unlikely that it [his working party] would be able to derive a model
of stock market behaviour which would be satisfactory for extrapolation into
the future and which would be generally acceptable as a basis for reserving
and that in the United Kingdom little attention has been paid to a ruin
probability approach.25
Corby explained his proposed alternative: The approach followed is to
assume a trend line for the performance of the relevant index together with a
spread about that line. To obtain the reserve at inception of a single contract,
it is assumed that all purchases are made at the top of the range and all sales
(i.e. maturities) at the bottom of the range.26
Thus the reserve for an annual premium product was calculated by assum-
ing that the guarantee shortfall at maturity was:
1 k n
(1 + r )
t
G
1 + k t =1
where r is the assumed long-term expected equity return and k is some mea-
sure of the extent to which the equity market can diverge from this trend.
This naturally led to the question of how to set the parameter values for k
and r. Corby did not have a clear answer for that. He tabulated results for a
range of values for k (0.2, 0.3, 0.4) and r (5 %, 7.5 %, 10 %) and noted that
values of k=0.4 and r=7.5 % produced results similar to Benjamins. So his
approach was simpler and more transparent, especially to actuaries who did
not have a familiarity with stochastic modelling, but it was somewhat arbi-
trary and Corbys only means of implementing it was to select parameters that
were consistent with the stochastic modelling analysis developed elsewhere.
His approach still demanded that a choice of stochastic model and calibra-
tion be made, even though he believed that it was not possible to make such
a choice in a way that would be satisfactory or acceptable.
23
Corby (1977), p.261.
24
Corby (1977).
25
Corby (1977), p.262.
26
Corby (1977), p.264.
The Corby paper was also the institutes first explicit engagement with
financial economics developments in option pricing theory. Corby noted the
existence not of the original Black-Scholes-Merton papers, but of a number
of papers that had been written more recently by academics at the University
of British Columbia on the application of option pricing theory specifically
to unit-linked maturity guarantees.27 These papers by Brennan, Schwartz and
the actuary Phelim Boyle are notable as the earliest work on applying option
pricing theory to the pricing of the financial guarantees found in life assur-
ance policies. They directly applied risk-neutral valuation to the guarantees
and showed that they could be hedged using portfolios with dynamically
rebalanced holdings of risk-free bonds and underlying units. Corby relegated
discussion of these concepts to an appendix and delegated the writing of the
discussion to another actuary, P.J.Nowell.
Nowell dismissed the option pricing work as a theoretical irrelevance, con-
cluding: Here practice and theory are irreconcilable the investment pro-
cedure [dynamic hedging] is a perfectly reasonable theoretical concept but
as a practical proposition it is one which contains risks greater than the risk
which it is designed to eliminate.28 Corby and Nowell threw the baby out
with the bath water. There was no analytical basis for their assertion that real-
life dynamic hedging would result in an overall increase in the life offices risk
position. And the fundamental economic insight of the guarantee replica-
tion argument and its implications for pricing appeared to be lost on them.
However, it is interesting to note that a couple of speakers at the Staple Inn
discussion of the paper raised the prospect of immunising the office from
guarantee losses by holding a negative exposure to the underlying units in the
contingency reserve.29 These actuaries did not refer to option pricing theory
or dynamic hedging. To them, delta hedging was an intuitive extension of
Redingtons immunisation theory.
Whilst Corby and Nowells discussion of option pricing theory and dynamic
hedging was rather perfunctory and dismissive, the topic and its application
to unit-linked maturity guarantees received further attention from the profes-
sion in the following years. In particular, T.P.Collins published an Institute
paper in 1982 which discussed in detail the concept of investing the guarantee
premiums in a dynamic hedging strategy.30 Collinss analysis was intelligent
and open-minded. He provided a thorough examination of the challenges of
27
Brennan and Schwartz (1976b), Boyle and Schwartz (1977). Also see Brennan and Schwartz (1979b; 2).
28
Corby (1977), p.273.
29
Fagan, p.282, and Seymour, p.285, in Discussion, Corby (1977).
30
Collins (1982).
a real-life implementation of a dynamic hedging strategy, such as transac-

tion costs and discrete-time rebalancing frequencies. He highlighted that the
hedging strategy is exposed to large short-term movements (up or down) in
the underlying asset value and that large switches in the hedge portfolio may
be required as the guarantee nears maturity when the underlying unit value is
close to the guaranteed amount.
Collins modelled what the historical performance of hedging and conven-
tional investment strategies would have been over the period of 19301978in
the UK. He concluded that the dynamic hedging strategy compares unfa-
vourably with the conventional strategy.31 However, he caveated this con-
clusion with the observation that The conclusion depends critically on
whether the [historical] price series used in the investigations is representa-
tive of the underlying distribution if the price had not risen sharply after
the end of 1974 the immunization [dynamic hedging] strategy would have
proved superior.32
Collins correctly identified the couple of key features of the historical
return series that drove his conclusion. One was the negative serial corre-
lation (mean-reversion) in the returns. This effect reduced the risk of the
conventional strategy, but did not reduce the cost of the hedging strategy
(which is driven by the short-term volatility experience and is not affected by
serial correlation). Secondly, the hedge strategy did not mitigate the impact
of exceptionally large short-term movements that could occur between hedge
rebalancing actions. This was partly a reflection of the limited sophistication
of his modelled hedge strategy implementation, but nonetheless reflected the
genuine gap risk that is present in a dynamic hedge program. The paper was
a perceptive contribution to thinking on the applicability of dynamic hedg-
ing of long-term guarantees, though its empirical emphasis missed the pricing
insights at the core of option pricing theory.
A working party report on unit-linked maturity guarantees was published
in 198033 (by which point Collinss paper had already been submitted but
not yet published). The working party tried to establish an actuarial ortho-
doxy on the vexed topic of how to reserve for these guarantees. Benjamin and
Wilkie were both members of the working party, whilst Corby was not. Its
conclusions were therefore unsurprisingly broadly consistent with the reserv-
ing approach advocated by Benjamin, Scott and Wilkie and did not consider
Corbys approach. The working party recommended that a probability of ruin
31
Collins (1982), p.280.
32
Collins (1982), p.281.
33
Ford etal. (1980).
approach to reserving should be used, and that a stochastic model of equity

returns be used to assess the reserve.
Wilkie developed a new version of his stochastic model of equity returns
for use by the working party in its reserving analysis. Its notable new feature
relative to the model he used in his 1977 paper was that the equity total return
index was not modelled directly, but instead as the ratio of dividend pay-outs
to dividend yield. Explicit stochastic processes were developed for dividends
(its logarithm followed a normally distributed random walk with upward
drift) and dividend yields (a mean-reverting lognormal process). The equity
price was the ratio of the two. This structure provided the flexibility to capture
the volatility patterns implied by Wilkies view of material mean-reversion
in long-term equity returnsit could be fitted to views for both short-term
and long-term equity return volatility that were consistent with significant
serial correlation in equity prices. It produced similar results to Wilkies 1977
model. That is, it produced a significant long-term mean-reversion effect
that resulted in lower long-term price and return volatility than a random
walk model with the same year-to-year volatility assumption. The reserves
estimated in the 1980 working party report were broadly similar to those
produced in Wilkies 1977 paper.
The working party report discussed dynamic hedging and, whilst noting it
was a subject which merited further investigation, they considered that there
is no basis for reducing maturity guarantees reserves because a company fol-
lows some form of immunization strategy and, in fact, a company that follows
such a strategy without fully appreciating the difficulties could well require
greater maturity guarantee reserves than would otherwise be the case.34
The working party report brought the British actuarial professions ten-
year saga with unit-linked maturity guarantees to a close. The quantum of
reserves recommended by the working party was not substantially different
to the rough-and-ready estimates produced by Benjamin a decade earlier. The
working partys two fundamental recommendationsthat appropriate risk-
based reserves should be held for unhedged market risk, and that no reduction
in reserves can be obtained by hedging the riskmade the maturity guarantee
feature of unit-linked policies commercially untenable and it was no longer
sold in volume by British life offices. In 1981, Duncan Ferguson, who would
go on to be Institute President in 1996, was able to state at a Staple Inn ses-
sional meeting: The considerable expertise which actuaries have now devel-
oped to handle linked policies has forced them to accept the logic that capital
guarantees are too expensive to be given and that in the end their offices
34
Ford etal. (1980), p.112.
can afford to do little more than provide policies under which the customer
takes nearly all the risks.35
This story reflects the stresses that arose when life offices underwrote, in a
very transparent way, significant volumes of non-diversifiable financial mar-
ket risk rather than (largely) diversifiable mortality risk. As we have seen, this
trend towards taking more market risk had been underway for the previous
50 years. But up until the unit-linked maturity guarantee product, this risk-
taking had been cloaked in the complexities of with-profits business. The
simplicity of the unit-linked maturity guarantee made this risk-taking highly
transparent, and it caught the profession unprepared.
Nonetheless, over the ten years following Benjamins stormy evening at
Staple Innthe sessional meeting that never wasthe profession made
real progress, not least because of the precocious young Scots actuary David
Wilkie. The profession had accepted the principle that life offices should hold
contingency reserves beyond the retrospective accumulation of premiums
to support financial market risk exposures. They had adopted a stochastic
modelling approach that could be used to provide reasonable estimates of
the required reserves. The working party and Collins had produced intel-
ligent and unprejudiced appraisals of the merits of dynamic hedging. Whilst
they had significant reservations, they had indicated a desire to constructively
engage with the new ideas of financial economics and option pricing theory
to further actuarial thought and practices. Significant and permanent devel-
opments in actuarial thought on the management of financial risk had been
achieved.
The Wilkie Model (19841995)

The Wilkie Model was the ubiquitous stochastic asset model for British
actuarial use from the mid-1980s to the end of the 1990sand, to a lesser
degree, beyond. Its genesis can be found in the stochastic equity modelling
that Wilkie developed and applied as a member of the Maturity Guarantees
Working Party. Wilkie further developed this work into a broader stochas-
tic asset modelwhich quickly became known as the Wilkie modelin a
paper presented to the Faculty of Actuaries in 1984 and published in the
Transactions of the Faculty of Actuaries in 1986.36
D.G.R.Ferguson, in Discussion, Redington (1981), p.389.

35
Wilkie (1986a).
36
The Wilkie model shared many stylistic and methodological similari-

ties with the equity modelling that Wilkie had developed for the Maturity
Guarantees Working Party. Equity prices were again modelled as the ratio of
dividend pay-outs to dividend yields using a time series model structure. The
model was expanded beyond the working partys asset modelling in two key
dimensions: stochastic models for the long-term government bond (consol)
yield and inflation were developed; and a cascade structure was introduced
such that the time series models of dividends, dividend yields and consol
yields included dependencies on past and present inflation as well as past val-
ues of themselves. The modelling approach was heavily based on time series
analysis and postulated some quite convoluted statistical relationships in the
joint behaviour of equities, consols and inflation. The model structure and
calibrations were driven directly by the relationships Wilkie found in the his-
torical data (based on UK series from 1919 to 1982). He was not concerned
with imposing many economic restrictions or null hypotheses on the pro-
jected asset behaviour. These characteristics were again evident when Wilkie
published an updated version of the model in 1995.37 This update extended
the asset coverage of the model, most notably by including models of short-
term government bond yields and property.
Wilkies preference for modelling historical behaviour as he found it, with-
out imposing strong economic prior beliefs, led to some striking implications
for the risk/reward profile of different asset classes and, especially, for what
could be generated by dynamic asset strategies. The structure and calibration
of the Wilkie model implied that most variation in equity dividend yield was
due to changes in the expected future return of equities (without any corre-
sponding change in risk) rather than being a signal of a significant change in
dividend growth expectations. As Wilkie himself noted in a 1986 paper,38 this
meant that dynamic asset strategies that switched between equities and bonds
using simple rules based on their relative yield levels could generate returns in
excess of those produced by simply buying and holding equities.
This begged an interesting quasi-philosophical modelling question: should
actuaries risk and capital assessments include credit for investment strate-
gies that take account of projected market inefficiencies? Should these inef-
ficiencies be assumed to prevail into the future? Are these strategies gaming
the asset model or are they reflective of a fundamental reality that ought
to be incorporated into the assessment of the risk profile of the business?
Such issues remain at large in twenty-first-century principle-based reserving.
37
Wilkie (1995).
38
Wilkie (1986; 2).
These questions generated very little discussion in the 1980s actuarial papers
and sessional meetings. They did, however, attract greater attention (and
discomfort) amongst actuaries in the 1990s, particularly from those in the
vanguard of modernising actuarial thought through the firmer embrace of
financial economic principles. Arguably, this is perhaps ironic as Fama and
Frenchs 1988 dividend yield and expected return empirical analysis that
emerged between the first publication of the Wilkie model and the profes-
sions eventual interest in financial economics could provide some intellectual
support from financial economics for Wilkies approach.
Risk and return anomalies also existed in the relative behaviour of other asset
classesin particular, between equities and property. Wilkie had a fairly limited
volume of available property data (28 years from 1967 to 1994) and it implied
average property returns significantly in excess of the average return generated
by the equity calibration, yet with a very similar risk profile. Such a feature
might possibly be rationally explained by illiquidity premiums or related trans-
action costs, but it was never Wilkies objective to try to provide an economic
explanation for market behaviour: he simply calibrated to what he found.
The other notable historical feature of the Wilkie model was its approach
to modelling the yield curve. In his original 1984/1986 paper, Wilkie only
attempted to model a single yield (the consol yield), so the issue of how to
consistently model all points on the yield curve never arose. But Wilkie chose
a model structure for the consol yield that was quite convoluted. The yield
was a function of its value in the previous three years (and hence was non-
Markovian). It was unnecessarily complicatedWilkie himself noted in his
original paper that a first-order autoregressive process could produce very sim-
ilar results and in the 1995 update of the model he implemented this change.
Writing later, Wilkie acknowledged that the original consol yield model may
have been an example of over-parameterisation.39
Wilkie never really grasped the nettle of producing a full stochastic yield
curve model. In his 1995 paper, he developed a short-term interest rate model
to go alongside the consol yield, but he left it to the user to decide how to join
the two dots. He referenced polynomial functions in the 1995 paper, and in
a later paper of 2003 he suggested a specific polynomial functional form to
interpolate between the two yields.40 He did not consider arbitrage constraints
or advocate the use of the arbitrage-free yield curve models that had been
developed since Vasiceks 1977 paper. (Interestingly, Brennan and Schwartz
derived the arbitrage-free yield curve pricing function for a t wo-factor model
Wilkie etal. (2003), p.319.

39
Wilkie etal. (2003), pp.35658.

40
where the first factor is the short rate and the second factor is the consol yield
in 198241). Indeed, Wilkie seemed uninterested in Vasiceks arbitrage-free
logic and its insights for constraints on the term premiums generated across
the yield curve. He was fairly dismissive of the entire stochastic yield curve
modelling stream of financial economic research, writing in 1995:
Many alternative yield curve models have been proposed in the academic litera-
ture Unfortunately, to my mind, they are usually based on an assumption
about how yield curves ought to behave rather than being based on how they
actually do behave.42
This comment highlights Wilkies deeply empirical modelling philoso-

phy. And there was some truth in his observations: the arbitrage-free yield
curve structures of the financial economics literature had features that were
deduced from the arbitrage-free condition rather than purely from empiri-
cal observation; moreover, most of these models were developed for the pur-
poses of derivative pricing rather than to describe the fullest description of
real-world yield curve dynamics. Equally, however, Wilkie offered no clear
rationale for why these approaches would not be the most appropriate way
of jointly modelling the behaviour of, say, a ten-year bond yield alongside his
cash and consol yield; and he did not raise or address the potential limitations
of using a yield curve model that could permit bond arbitrage and that pro-
vided no means of controlling the relative size of the term premiums available
across the term structure. Of course, the importance of this point depended
on the use to which the model was put. In the 1980s and 1990s, the model
was mainly used in the analysis of long-term asset allocation choices between
asset classes such as equities, property and long-term government bonds in
the context of funding long-term liability cashflows. For these purposes, the
absence of a robust yield curve model was relatively unimportant. But for later
actuarial applications such as the analysis of alternative derivative hedging
strategies for guaranteed annuity options, the model was wholly inadequate.
Of course, Wilkie never envisaged or advocated the use of the model in this
type of application when he developed it.
For context, it is worth noting, however, that some actuarial research was
published that did try to embrace the new yield curve modelling technology
of Vasicek etal. and showed how it could be made use of in actuarial prac-
tice. In particular, Phelim Boyle, who we met earlier as one of the pioneers
41
Brennan and Schwartz (1982).
42
Wilkie (1995), p.874.
of the application of option pricing theory to life assurance guarantees, pub-

lished a paper in the Journal of the Institute of Actuaries as early as 1978 that
showed how Redingtons duration calculations could be generalised to allow
for the volatility term structure implied by modern yield curve methods.43
Redingtons duration approach had assumed parallel yield curve movements,
which implied that long rates were as volatile as short rates. Empirical evi-
dence, and arbitrage-free yield curve models, suggested long rate volatility was
lower than short rate volatility. This meant that if matching the duration of a
set of liability cashflows with a long bond and a short bond, more would really
be required in the long bond than was implied by Redingtons duration calcu-
lation. This was a fundamentally important point for asset-liability duration
management and Boyle showed how the new financial economic modelling
approaches could cast fresh light on it.
The Wilkie models public domain and wide application meant that it was
subject to considerable peer scrutinymost notably, a working party was estab-
lished in 1989 to review the model and it published a paper with its findings
in 1992.44 But it focused mainly on the statistical quality of the time series fits,
especially for the inflation model, rather than on arguably the more fundamen-
tal question of whether such a statistically orientated and empirical approach
was desirable and what its economic limitations would be.
The Wilkie model was an important step forward for the British actuarial
profession in embracing the probabilistic analysis of the financial market risks
that increasingly dominated the balance sheets that actuaries oversaw. More
generally, it was also important in motivating the profession to develop its
quantitative modelling skills. But it was also a product of the profession as it
existed. In particular, Wilkies reticence to incorporate any of the insights that
could be delivered from the (then-recent) developments in financial econom-
ics into his model could be viewed in hindsight as a missed opportunity for the
further development of the professions financial risk management thinking.
ith-Profits: Reserving andFinancial Economics

W
(19872004)
Whilst an actuarial consensus on the uneconomical nature of capital guar-
antees as it pertained to unit-linked business had been established by 1981,
this view would typically not extend to the capital guarantees provided
43
Boyle (1978).
44
Geoghegan etal. (1992).
under with-profit policies. With-profits business and its guarantees had some
undoubtedly significant differences with unit-linked business and its maturity
guarantees. In the context of the risks and costs to the life office of providing
guarantees, linked and with-profit business had two differences of the utmost
importance. First, the office had significant control over the investment strat-
egy of the with-profit fund. If assets performed poorly, they could, in theory,
be switched into bonds that matched the guarantees before the surplus funds
were exhausted; second, in with-profits, the cost of the guarantees could be
borne by other generations of policyholders (rather than the office or estate)
through reductions in their future bonuses. If one generation of policyholders
did not have accumulated asset values sufficient to meet the minimum guar-
antee attaching to their policy, this could be funded by reducing the bonuses
that would be distributed to other policyholders. These broad powers of actu-
arial discretion provided risk management levers that could, in principle, sub-
stantially mitigate the financial market risks to the office that were created by
with-profit guarantees.
Actuaries, and indeed the UK life office regulator, were therefore significantly
more sanguine in the 1980s about with-profit guarantees than those found in
their unit-linked cousins. There was no body of opinion advocating the exten-
sion of the stochastic modelling and probability of ruin approach developed
for unit-linked maturity guarantee reservingthe considerable expertise that
Duncan Ferguson referred tointo the domain of with-profit reserving. This
was partly because it would be viewed as unnecessary as the office did not have
the same degree of risk exposure, and partly because it would be much harder
to model the complexities of with-profit business and actually demonstrate
that these risk management levers could be adequately operated by life office
actuaries to manage the guarantee risks. This scepticism around the benefits of
using stochastic modelling in the actuarial management of with-profit business
is well-reflected by Frank Redingtons comments in 1976:
He [Redington] expressed uneasiness at leaning upon computers to answer

these unanswerable questions. If there was a beam in the eye of the question,
why worry about a mote in the eye of the answer? It was easy to fail on the back
of an envelope as with a computer, but a great deal more instructive. This was a
countryside to explore on foot and not by fast car.45
Some British life offices did develop and internally use stochastic model
offices in the 1980s and 1990s, but the application of stochastic models
and probability of ruin approaches to British with-profit statutory reserving
45
Redington in Discussion, Kennedy etal. (1976), p.48.
was deferred for over 20 years until the beginning of the twenty-first cen-
tury. What about the use of option pricing theory in charging and reserv-
ing for with-profit guarantees? We saw above that the (unit-linked) Maturity
Guarantees Working Party did not actively make use of option pricing theory
in its work on the pricing, reserving and risk management of unit-linked
guarantees. However, in the following years the British profession, and in par-
ticular David Wilkie, started to explore the potential applicability of option
pricing in actuarial work. In 1987, Wilkie published a notable paper on the
use of option pricing in the analysis of with-profit guarantees and bonuses
with his Institute paper, An Option Pricing Approach to Bonus Policy.46
Wilkie showed that with-profit style pay-outs could be replicated by a
mixture of investments in unit-linked funds and put option contracts. The
recognition that with-profit guarantees could be considered as a form of put
option was not remarkable and, as we saw above, this was first considered by
Phelim Boyle some ten years earlier. Wilkie broke new ground, however, by
considering how different reversionary bonus strategies could impact on the
option costs associated with the guarantees. Wilkie also tentatively suggested
that the cost of the options that correspond to the reversionary bonus strategy
could be used as a reference point for a deduction from with-profit terminal
bonuses as a form of charge for the provision of the guarantee (when the fund
performance is good enough for a terminal bonus to be paid).
Wilkies approach resulted in relatively high estimates of the guarantee
cost as it did not allow for the cost-mitigating with-profit features of office-
controlled investment strategy or inter-generational cross-subsidy. A 20-year
single-premium contracts guarantee cost inclusive of reversionary bonuses
was estimated to require a deduction from the final proceeds of the policy of,
on average, around 15 % to 25 % across a range of stochastic scenarios.47 It
was, however, the natural first step in applying option pricing ideas to with-
profit guarantee charging and Wilkie, now established as the British actuarial
technical thought-leader of his generation, was the natural actuary to lead the
way. The paper also helped to more generally introduce option pricing ideas
to the actuarial profession. Collinss excellent work back in 1982 was never
presented at a Staple Inn sessional meeting, and before the presentation of
Wilkies paper in June 1987, option pricing theory had never been discussed
at an Institute sessional meeting. In opening the discussion of Wilkies paper,
J.M.Maud started with the telling statement:
Wilkie (1987).
46
Wilkie (1987), p.55.

47
Until a few months ago I was almost entirely ignorant about option pricing
theory. Having been introduced to the subject, I was astonished that a theory
with so many obvious uses in actuarial work is so little known to many of us.48
However, British actuarial thinking on the use of option pricing theory in

with-profits reserving and risk management lay largely dormant for the next
decade. At the start of the 1990s, more pressing matters were at hand. The
actuarial profession and its practices in the with-profits sector were under
scrutiny like never before. Similar to the experience with unit-linked matu-
rity products 20 years earlier, there was a sense that actuarial methods were
failing to keep up with a changing world. With-profit products were under
great competitive pressure from the alternative investment vehicles of asset
managers and banks. Life offices had evolved their with-profit offerings to
compete: conventional regular premium with-profit business was on the
wane, and was being superseded by unitised with-profits, which was a more
flexible, recurrent single premium product design. The net premium valua-
tion method was obviously less applicable to the solvency assessment of single
premium products, and, in any case, there was concern amongst actuaries at
the start of the 1990s that the net premium statutory valuation requirements
had become unreasonably stringent.
There was also increasing regulatory pressure to take explicit consideration
of Policyholders Reasonable Expectations (PRE) in the management of with-
profits, and for these considerations to be reflected in with-profit reserving
methods. This was not easy in the net premium valuation method, where no
explicit allowance was made for the future bonuses that policyholders could
reasonable expect to receive. Meanwhile, Canada and Australia had recently
enacted new solvency methods that dispensed with net premium valuations.
These changes were largely driven by pressure from accounting standard set-
ters to produce valuations that could conform to generally accepted account-
ing principles. Working parties were established by the Institute and Faculty
in 1993 to investigate a modernisation of with-profit reserving methods. This
started a process that meandered for a decade as actuarial leaders tried to
reconcile their traditional methods and training with the demands of a fast-
changing external environment.
One of the first deliverables from this process was the 1996 paper by a
working party, led by P.G.Scott, which had been established with the remit
to investigate alternatives to the net premium valuation for statutory solvency
48
J.M.Maud in Discussion, Wilkie (1987), p.78.
assessment.49 This paper proposed two fundamental changes to the statutory

solvency valuation method for with-profit business: the net premium valua-
tion should be replaced with a gross premium bonus reserve valuation (which
should include future reversionary bonuses but not terminal bonuses); and a
second value should also be reported, which was intended to be a measure of
the realistic value of the policy having reference to policyholders reasonable
expectations (and which was loosely interpreted as the policyholders asset
share, i.e. the accumulated value of their invested premiums).
The arguments presented in the paper and the professions discussion of
them rather resembled the actuarial discussions of the late nineteenth cen-
tury. Net premium liability valuations taken alongside asset market values
produced spurious variation in free assets; it did not have regard to the rea-
sonable (bonus) expectations of policyholders; but if a gross premium bonus
reserve valuation was used, what level of future bonus should be reserved for?
And why did the working party recommend reserving for future reversionary
bonuses but not terminal bonuses? (A certain logic for this position existed.
European Union legislation required the liabilities to be discounted based on
the yield of the assets backing the liabilities. The working party argued that
in the scenario that equities did not generate any capital growth, no terminal
bonuses would be paid, and hence they could be ignored in a statutory valu-
ation that only took account of equity dividend yields.)
Risk margins were incorporated into the valuation under both the net pre-
mium method then in place and in the gross premium approach proposed by
Scotts working party, but there was no rigorous basis for determining what
these margins should be. There was no underlying concept of what level of secu-
rity the solvency valuation should aim to deliverthere was no probability-
of-ruin objective. Furthermore, there was no consideration of option pricing
ideas or how financial market risk could be managed in with-profits and how
reserving methods could be aligned with risk mitigation (i.e. risk-based).
The two key recommendations of the Scott working partythat the net
premium valuation be replaced by a gross premium bonus reserve valuation
for conventional with-profits, and that a second valuation be submitted that
reflected the terminal bonus expectations of policyholderswere broadly
rejected by the actuarial profession. So another working party was established
in 1996 with terms of reference to determine whether the ideas put forward
by Scott could be developed into something better. This new working party,
led by P.W. Wright, published a paper reporting their findings in 1998.50
Scott etal. (1996).

49
Wright etal. (1998).

50
It rejected the key recommendations of the Scott working party and recom-
mended that the net premium method be retained for conventional with-
profits. However, part of their reasoning for this recommendation was not
that the Scott working partys findings were necessarily flawed, but rather
that conventional with-profits business was now being written in such low
volumes that a major change in its valuation method was hard to justify. As
Frankland put it at the Institute discussion:
At the start of this century prophetic actuaries spoke of the 20th century being
that in which a replacement would finally be found for the net premium valua-
tion method. Todays prophets appear to accept that, within a decade, the
replacement of the net premium valuation method for conventional with-profits
business will cease to be a matter of materiality in the valuation of a with-profits
life office.51
Like the Scott working party, the Wright working party recognised that
solvency and PRE must be taken into account in reserving for unitised with-
profits business. But unlike Scott, the Wright working party advocated achiev-
ing this through a single reserving figure which would be somewhat akin to the
greater of the two calculations suggested by Scott. The first would be a bonus
reserve test which would be the present value of guaranteed maturity values.
Like in the with-profit regulations of the time, this present value would be cal-
culated in a resilience scenario that featured a prescribed series of stresses to
assumptions. This scenario was not explicitly assessed using stochastic models
and probabilities of ruin as per the unit-linked maturity guarantee reserving
practices, but its intention was fundamentally similar (the effect was not so
severe because the tests were weaker than the tails produced by the stochas-
tic equity modelling of the Maturity Guarantees Working Party and because
asset allocations were less risky for with-profit business). The important point
that the Wright working party advanced was that it was not sufficient for this
reserve to be based only on the current level of the guaranteesome explicit
allowance must be made for future reversionary bonuses that would be con-
sistent with PRE.The allowance should be consistent with the reversionary
bonuses that would be paid in the circumstances assumed in the statutory
valuation basis (including those used in the resilience test). Like the Scott
working party, the Wright working party argued that no allowance should
be made for terminal bonus in this leg of the calculation (this was a principle
51
Frankland, in Discussion, Wright etal. (1998), p.1036.
that was much cherished by British life offices and its actuaries and that was
only permitted in EU legislation after a great deal of political horse-trading).
The second leg of the reserving calculation was the surrender value of the
policy that was consistent with PRE.This would generally be closely related
to the asset share of the policy (i.e. the accumulated value of the premiums
at experienced investment return, net of charges and expenses). A require-
ment for the reserve for a with-profit policy to have a floor that was equal to
or close to asset share implied that the policy could not provide substantial
capital support to the business, even when the level of guarantee was very low.
PRE implied that terminal bonuses had to be reserved for after alleven if it
was only the terminal bonus that had accrued within current surrender val-
ues. The Institute discussion of these proposals was very mixed. To some, the
principle of reserving for terminal bonus was anathema, and they argued that
PRE did not mean a guarantee of a surrender value that was closely related to
the policys asset share. To others, the proposals were merely a codification of
then-prevailing best practices. But whether they liked it or not, some 40 years
after the first terminal bonuses were paid on British with-profit policies, it
appeared inevitable that regulatory pressure on the explicit treatment of PRE
would result in their inclusion in statutory solvency assessment.
During this episode of statutory reserving navel-gazing and prevarication,
there was very little research published by the profession on the application
of option pricing concepts to reserving or charging for with-profit guarantees.
That changed, however, in the late 1990s. Over the period between 1997
and 2004 several important papers were published on this topic. By the mid-
1990s, the professions internal debates on if and how to use financial eco-
nomic ideas had become increasingly strained and adversarial. This perhaps
reflected a more general malaise that was impacting on the profession and
its traditional areas of focus. Actuaries were facing an unprecedented degree
of scrutiny from others in the financial, regulatory and even legal sectors.
Financial economics was increasingly being viewed by some actuaries as a
threat to their core doctrines rather than as a tool they could utilise. Criticism
emerged, especially in the pensions field (discussed below in Chap. 6), that
suggested some actuarial struggles were arising from their economic illiteracy.
Some actuarial traditionalists now rejected financial economic thinking, not
on the grounds that it was a nice theory with no practical utility (Corby and
Nowells position), or because it needed to be considered more thoroughly
in order to make use of it (Collins and Wilkies position), but because it was
simply wrong and ought to be wholly rejected.
This latter view was the position of Robert Clarkson, who made his argu-
ments in a Journal paper, An Actuarial Theory of Option Pricing, published
in 1997.52 This paper must surely rank highly on the list of the most eccentric
papers ever published by an actuarial journal. Clarkson took the reader on
a journey that pondered the ideas of many great thinkers such as Einstein,
Keynes, Bernoulli, Newton, Adam Smith, Mandelbrot, Halley, Hayek and
himself before concluding that we have to abandon this [Black-Scholes]
methodology completely53 and it seems unscientific in the extreme not
to conclude that a completely new paradigm of option pricing is urgently
required.54 This rejection was based on the argument that its formal math-
ematical derivation is completely detached from reality.55
He rejected the most basic and general insights of well-established option
pricing theory. He argued that it was commonsense that a call (put) option
price must increase (decrease) with the expected return on the underlying
asset.56 This whimsical rejection of the risk-neutral valuation concept would
have been plausibly breathtaking to other financial professionals. Clarkson
proposed a new approach to option pricing. But it was fundamentally the
same one that had been advocated for unit-linked maturity guarantee pric-
ing back in the 1970s by Benjamin and Wilkie: calculate the guarantee price
as the real-world expected cashflow, discounted at a rate consistent with a
static asset mix, and add a loading for the cost of the capital that would be
required to support the risk associated with the (unhedged) option. Clarkson
also argued that financial markets exhibit systematic over-reaction and that
the expected cashflows and capital should therefore be calculated on the basis
of a mean-reverting stochastic process, which again was consistent with the
Maturity Guarantees Working Party. Beyond the hyperbole, his proposed
grand new idea was not new at all. If anything, it was an anachronism. It
failed to produce prices consistent with put-call paritythe most basic (and
model-independent) fundamental property of put and call option pricing.
Clarkson did have many sensible practical observations to make, and his gen-
eral critique of the limitations of economic theory to practical financial prac-
tice was not without some basis, but his overall position was so extreme and
out-of-touch with the modern financial sector that it was an embarrassing
reflection on the British actuarial profession that it provided a platform for
such views in the late 1990s.
52
Clarkson (1997).
53
Clarkson (1997), p.333.
54
Clarkson (1997), p.335.
55
Clarkson (1997), p.367.
56
Clarkson (1997), p.335.
The Faculty discussion of the paper highlighted that by this point in time
an impressive generation of younger British actuaries such as Kemp, Cairns,
Exley, Smith, Mehta, Macdonald, Speed and Bowie had emerged who had
independently developed expertise in financial economics and who worried
that the British actuarial profession was dangerously behind other financial
professionals in their positive utilisation of these ideas. Bowie gave a particu-
larly impassioned speech where he argued that it is nothing short of mis-
placed arrogance to repudiate a perfectly respectable and successful science
on the basis of a criticism [use of abstract or unrealistic models] to which our
own profession is also subject.57 This ultimately bode well for the catching-
up that the profession would go on to undertake in the 2000s.
In the late 1990s with-profit funds were experiencing a period of difficult
publicity that required life offices and the actuarial profession to be clearer
about how the product worked and, in particular, which policyholders were
paying for what. Improving the transparency around guarantee costs and how
they were charged for was increasingly another important factoralong with
PRE and financial reporting requirementsdriving change in actuarial meth-
ods in with-profits. As an appointed actuary commented at an Institute ses-
sional meeting in April 2000:
Recent events, including guaranteed annuity options, potential mortgage

endowment shortfalls, and a public suspicion that life offices have squirreled
away policyholder money over the years in the form of orphan estates, have
made the public and consumer press less inclined to trust life companies and the
black box processes of the traditional with-profits fund.58
A flurry of papers was produced between 1999 and 2004 that significantly
enhanced the actuarial professions thinking on the application of option
pricing ideas to the assessment of the costs and risks associated with with-
profit guarantees. These papers provided new insights into potential trans-
parent guarantee charging and risk management approaches for with-profits.
The first of these papers, A Market-Based Approach to Pricing With-Profit
Guarantees,59 was published in 1999 by the Faculty of Actuaries Bonus and
Valuation Research Group. The group was led by David Hare, who would go
on to become President of the Institute and Faculty in 2013. The paper can
be viewed as a natural, if somewhat belated, development of Wilkies 1987
57
Bowie, in Discussion, Clarkson (1997), p.385.
58
Saunders in Discussion, Hare etal. (2000), p.722.
59
Hare etal. (2000).
ith-profits and options paper and the 1980 Maturity Guarantees Working
w
Party paper. The British actuarial professions 1990s issues with quantitative
techniques and financial economics were referred to by Hare in his intro-
duction of the paper at the Faculty sessional meeting: Mention of the word
stochastic can cause some actuarial eyes to glaze over, and the inclusion of
equations like the Black-Scholes pricing formula can prove major deterrents
to a wide readership of a paper.60
The Hare paper applied the Maturity Guarantees Working Partys approach
to reserving for guaranteesa probability-of-ruin approach using a 1 % prob-
ability and a stochastic equity model with mean-reverting propertiesto
determine the reserves required for with-profit guarantees at various durations
and with various levels of equity backing ratio (which were assumed to be
static). Unsurprisingly given the similarity in methodology, these results were
consistent with the Maturity Guarantees Working Party results. The paper
also considered a guarantee charging approach that had two components: the
expected (real-world) guarantee shortfall plus a loading for the cost associated
with the capital that would need to be held to maintain a 1 % probability
of ruin. As discussed above, this was consistent with the 1970s thinking of
Benjamin and Wilkie, and also with the more recent writings of Clarkson.
The paper then went on to also consider a market-based approach to pric-
ing the guarantee. That is, put options (and their market prices) were used to
match the guarantees (for a given equity asset allocation). They introduced
an interesting wrinkle to the use of optionsthey argued that in order to
compare with the capital-based approach, the strategy should not pay out
on guarantee shortfalls beyond the 1 % probability of ruin level, and so the
strategy involved a put spread where the office sold a put option with a strike
at the equity index value at the 1 % probability of ruin level. They concluded
that the two approaches to pricing the guarantees produced broadly compa-
rable results (though this would naturally be a function of subjective param-
eter choices for the expected-shortfall-plus-cost-of-capital-loading approach).
Like in the Wilkie 1987 paper, these with-profit guarantee costs were gen-
erally higher than was intuitive to with-profit actuaries because it made no
allowance for the cost-mitigation levers that were available to the with-profit
actuary (such as reducing the funds equity asset allocation level when surplus
assets were eroded).
At the start of the 2000s, significant changes in global insurance financial
reporting and UK insurance regulation were afoot. In 1997, the International
Accounting Standards Board started work on developing a new accounting
60
Hare in Discussion, Hare etal. (2000), p.197.
standard for insurance. In the following years its intention became clear: the
principle driving the Standard would be that a balance sheet-driven approach
should be used for insurance firms financial reporting; profit should be mea-
sured as the change in the value of assets and liabilities; and those valuations
should be done at fair value (which, for deep and liquid markets, meant mar-
ket value). The Faculty and Institute Life Board established a working party
in 1999 to consider what a fair value approach to asset and liability valua-
tion might offer for the development of an improved approach to reporting
for prudential supervisory purposes.61 The terms of reference of the working
party provided the motivation:
It is recognised that the existing methods of actuarial valuation for long-term

insurance, particularly for supervisory purposes, are inadequate for their pur-
pose under certain circumstances and economic conditions.
This would be the third actuarial working party in ten years to make an
attempt at improving actuarial methods for life office statutory reserving. It
would be easy to suspect it was a task that was beyond the profession. But this
time was rather different: now the profession merely had to follow the lead
given to it from outside. Historically, financial reporting was based on adjust-
ments to the actuarys statutory solvency valuations (the modified statutory
basis). Now the roles would be reversed. Financial reporting standards would
lead the specification of the profit reporting requirements, and the job at hand
for the actuarial profession was to consider how these developments could help
the profession make their necessary improvements in statutory valuations.
This Fair Value Working Party, chaired by C.J.Hairs, presented its find-
ings at Institute and Faculty meetings in November 2001 and their paper was
published in the British Actuarial Journal in 2002.62 Its findings were fairly
revolutionary in comparison to the progress of the Scott and Wright work-
ing parties. This working party supported the fair value approach to financial
reporting. More fundamentally, it concluded that prudential solvency assess-
ment should also be based on the fair value approach. They advocated a risk-
based capital system for with-profits that made explicit use of probabilistic
models and a defined probability of ruin; the ruin event could be defined not
simply as failing to fund all liability cashflows as they fell due, but as fair value
insolvency over some specified time horizon that should be related to the time
taken to close out risk positions. They recognised that with-profit liabilities
Hare etal. (2000), p.204.

61
Hairs etal. (2002).

62
would require market-consistent stochastic models to estimate fair values, and

these models would likely need to be complex in order to adequately capture
management actions. There was no mention of net premium valuations.
All this was quite revolutionary and it is fascinating to see that relatively
little dissension arose in the Faculty and Institute discussions of the paper.
Thirty years after Benjamins aborted unit-linked maturity guarantee paper
on stochastic reserving approaches, the actuarial profession had now come
to accept that such principles should be implemented in with-profit funds,
the traditionally undisputed home of actuarial judgement and discretion.
But the profession was now dancing to someone elses tune. The UK finan-
cial regulator, the Financial Services Authority (FSA), the International
Association of Insurance Supervisors and the International Association of
Actuaries had all recently suggested a move to a more risk-based capital sys-
tem that could reduce the scale of regulatory arbitrage that prevailed between
countries and between insurers and banks. The emerging New Basel Capital
Accord for banks would use probabilistic models and even encouraged insti-
tutions to develop their own internal models for capital assessment.
Meanwhile, the UKs FSA published consultations in 2002 and 2003 that
would ultimately enshrine these concepts in the statutory solvency system.63
It required market-based measures of with-profit liabilities and guarantee
costs. Whilst the actuarial profession may have been fairly sanguine about
its reserving methods for conventional with-profits business, the UK regu-
lator was not. In particular, the regulator was concerned with the risk that
a firm using the net premium method might fail to keep enough reserves
to meet a policyholders reasonable expectations that bonuses would increase
arkets improve, and asset values increase).64 The FSA proposed a twin
(if m
peaks approach where the firm would need to hold the greater of two values:
the net premium valuation currently undertaken in accordance with EU leg-
islation; and a realistic present value of expected future contractual liabilities
plus projected fair discretionary bonus payments.65
What did this realistic present value imply? The first task was to codify the
projected discretionary bonus payments in a way consistent with PRE.For
this, the with-profit fund would be required by the FSA to set out its with-
profit bonus, smoothing and investment policies to policyholders in a docu-
ment called Principles and Practices of Financial Management. The realistic
present value had to be assessed using assumptions consistent with those
63
Financial Services Authority (2002), Financial Services Authority (2003).
64
Financial Services Authority (2003), p.11.
65
ocumented policies. Once the pattern of these discretionary payments had

d
been specified, they then needed to be valued. The consultations were unam-
biguous in their intentions in this regard: For the purposes of valuing the
contracts and the embedded options and guarantees the methods require
market consistency i.e. the use of model assumptions, or option prices that
replicate the costs of hedging such risks in the market.66
The FSA was the first insurance regulator in the world to attempt to imple-
ment such a reserving method for complex long-term life business such as
with-profits. But the initiative did not come completely out of the blue. It can
be seen as the natural conclusion of a process, led from outside the actuarial
profession, to put in place objective, transparent, measures of risk and capital
in life business that reduced the reliance on actuarial judgement and discre-
tion. An alignment of regulatory reserving methods with financial reporting
standards was also an increasingly important driver (and was a prime motiva-
tion of the abandonment of the net premium valuation method in Canada
and Australia). The FSA proposals aligned with emerging International
Accounting Standards Boards proposals around fair value accounting.
The FSA was also not intent on allowing any undue delay to the imple-
mentation of their radical proposals. For more than a decade, the British
actuarial profession had prevaricated over reserving methods for with-profit
business. The FSAs proposals of 2002 and 2003, which were far more radi-
cal and created quite significant new intellectual and technological demands,
were to be implemented in 2004. The actuarial profession had some work to
do and three papers were published in the British Actuarial Journal in 2003
and 2004 that proposed how these market-consistent or realistic valuations
of with-profit liabilities could be undertaken.67 Up until this point in time,
no stochastic methods had been used in with-profit statutory reserving. The
FSA proposals didnt merely try to bring with-profits reserving in line with
unit-linked maturity guarantee reserving as was discussed by Hare et al. It
went much further, defining a fundamental shift in how to quantify a reserv-
ing requirementthe reserving requirement was to be based on the 99.5th
percentile of the one-year change in the fair value balance sheet rather than
on a percentile of the ultimate cashflow shortfall that could emerge over the
full run-off of the liability outgo (i.e. over 30 or 40 years or more). As Hibbert
and Turnbull wrote in the introduction to their paper:

66
Hibbert and Turnbull (2003), Dullaway and Needleman (2004), Sheldon and Smith (2004).
67
Actuarial philosophy towards the valuation of liabilities has traditionally been

based on the notion of funding the funding approach will tell us what
reserves are required to meet the liabilities [cashflows] with a given level of con-
fidence. By contrast, the thinking behind the economic valuation of a liability is
very different. The economic value is defined as the sum of money required to
establish a portfolio of assets that provided that they are invested in a particu-
lar way will replicate the liability as closely as possible. This special portfolio is
called the hedge portfolio.68
Smith and Sheldons introduction provides further background on the

industry and regulatory context:
The introduction of the realistic balance sheet is, in part, a response to the difficul-
ties that un-hedged guarantees have caused the life industry in recent years. Reliance
on long-term solvency tests runs the risk that we overlook more imminent prob-
lems, compounded by the use of over-optimistic assumptions and models used to
determine capital needs. While there has been some criticism of the transfer of
banking techniques to life assurance, the rate of deterioration in life offices finances
over the last three years demands a greater focus on the short term.69
Hibbert and Turnbulls (2003) paper presented an approach to economic

valuation of with-profit liabilities that could be viewed as an evolutionary
next step beyond the approaches of Wilkie in 1987 and Hare etal. in 2000.
As was noted above, the Wilkie and Hare papers did not attempt to iden-
tify how the unique features of with-profitsthe forms of discretion that the
life office has in managing the productcould impact on the assessed cost
of with-profit guarantees. As a result, they produced counterintuitively high
measures for the guarantee cost. This failing had become well-understood.
The Fair Value Working Partys 2002 paper had noted stochastic models of
with-profits funds often show alarmingly high ruin probabilities because of
inadequate modelling of management actions in adverse scenarios.70
Hibbert and Turnbull suggested that a stochastic simulation approach
with an asset model calibrated to observable market pricescould be used
to fully capture the impact on guarantee costs of the with-profit funds abil-
ity to dynamically revise investment policy and bonus policy. Naturally, any
results would be highly sensitive to the form that this dynamism was assumed
to take, but this point had already been addressed by the FSAs requirement
for the Principles and Practices of Financial Management documentation.
68
Hibbert and Turnbull (2003), p.726.
69
Sheldon and Smith (2004), p.547.
70
Hairs etal. (2002), p.233.
One of Hibbert and Turnbulls points of emphasis was that the market-
consistent valuation methodology could also be used to identify the hedg-
ing required in the estate to mitigate the risks left behind after the available
(PRE-compliant) management actions had been implemented. Simulation
modelling together with a codification of the with-profit funds actuarial
management levers could allow the costs and risks of with-profit guarantees
to be analysed like any other derivative contract, albeit a complex one.
The Hibbert and Turnbull case study results suggested that the economic
cost of guarantees could be as much as halved by allowing for the life offices
ability to dynamically manage investment policy and bonus policy under
assumptions that were typical of the time. Dullaway and Needlemans 2004
paper provided additional illustrative results of the impact on liability valu-
ation of dynamic rules for the discretionary management features of with-
profits business, again highlighting that these effects were material and could
be captured by a simulation model.
The paper by Sheldon and Smith further discussed the technical challenges
of market-consistent valuation of life businessparticularly for the calibra-
tion of market-consistent asset models (choice of risk-free asset, extrapolation
of implied volatilities, and so on). It highlighted the important point that
market-consistent valuation of very long-term, complex liabilities could never
be completely objective, even if that was the desire of the regulator. The paper
also showed how some formulations of dynamic management actions could
be captured in closed-form guarantee pricing formulae, thereby avoiding the
need for the brute-force method of Monte Carlo simulation.
Collectively, these three papers provided a technical grounding for the
actuarial implementation of market-consistent valuation of long-term life
assurance liabilities. Such techniques could be used in solvency reserving,
guarantee pricing and hedging of the market risk created by the guarantees.
The FSAs 2004 timetable was successfully met. The use of market-consistent
valuation as a basis for life assurance reserving went on to be adopted more
widelyfor example, in the European Unions Solvency II system. The expe-
rience demonstrated that whilst the British actuarial profession had arguably
lost its ability to be a master of its own destiny, it was capable of technical
excellence in implementation when it was told what needed to be done.
Guaranteed Annuity Options (19972003)

Guaranteed annuity options (GAOs) options to convert the cash proceeds
of a life policy into an annuity at a guaranteed minimum raterepresent a
particularly difficult episode of post-war British life office history. The GAO
story contains a number of the threads that have already been identified in
our above discussions of post-war British life offices: long-term financial guar-
antees provided under the stress of competition with little initial actuarial
consideration for charging and reserving requirements; a lack of actuarial
interest in modern thinking on economic risk measurement and hedging; an
over-reliance on the assumption that the unique features of with-profits could
mitigate whatever economic risk was created by guarantees; ultimately, in the
face of actuarial prevarication, the intervention of external parties (regulators,
and, in this case, the courts) to compel a resolution.
Forms of guaranteed annuity option were written by British life offices
from the end of the Second World War, and, to a lesser degree, earlier. Long-
term interest rates rose almost continuously between 1950 and 1973 (consol
yields increased from 4 % to 17 % over this period) (Fig. 5.1). GAOs were
valuable to the policyholder in low interest rate environments and they there-
fore did not create any financial stress for life offices over the third quarter of
the twentieth century. From the outset, however, there was some awareness
amongst actuarial thought-leaders that such long-term guarantees were inher-
ently risky for the life office. A particularly notable critic was Dick Gwilt, who
was President of the Faculty of Actuaries from 1952 to 1954 and the principal
executive of Scottish Widows from 1946 to 1960. Speaking at a faculty meet-
ing in 1948, Gwilt voiced his concerns:
16
14
Long-term Government Bond Yield (%)
12
10
0
1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005
Year
Fig. 5.1 Consol yields (19702005)

Some of the guarantees given by offices at the present time [1948] appear to me
to be based on somewhat optimistic assumptions about the level of interest rates
many years hence. I cannot see any justification for giving options on a basis
which may involve offices in serious loss if interest rates are low but do not give
a chance of profit if interest rates are high for, in that event, the policyholders
take the cash and secure the benefit of the high rates. I feel we ought to keep this
matter prominently before us.71
A few years later at the 1952 Faculty meeting where Haynes and Kirtons
interest rate risk management paper was presented, Gwilt again voiced his
concerns about these long-term options:
I think there is no doubt that many actuaries pay lip-service to the dangers of
options, but, unfortunately the stress of competition leads them to grant terms
which in the long run may prove to be a serious embarrassment and source of
loss. I would go so far as to predict that the time will come when offices will
bitterly regret some of the options which have been so light-heartedly granted in
recent years [1952].72
Gwilt was not the only actuary with such worries about these product fea-
tures at this time. In the Staple Inn discussion of Redingtons seminal immun-
isation paper, Haynes commented:
Reserves ought to be set up immediately to meet the chance that the options
might become onerous and that the actuarys assessment of current surplus
should be reduced in some cases very extensively reduced by reason of the
options having been granted to my mind, the fundamental answer was to
restrict the granting of options to an absolute minimum.73
These comments are particularly striking in the context of the life offices
experience of guaranteed annuity options in the 1990s and 2000s, particu-
larly as such options were written up until the mid-1980s and no reserves were
held for them until the 1990s. But they also reflect a more general phenom-
enon that would recur in post-war British life actuarial experience: actuaries
were struggling to keep up with the sales and product development functions
of life offices. Whether it was guaranteed annuity options on with-profit poli-
cies, or maturity guarantees on unit-linked savings products, or perhaps even
71
Gwilt in Discussion, MacLean (1948), p.318.
72
Gwilt in Discussion, Haynes and Kirton (1952), p.213.
73
Haynes, in Discussion, Redington (1952), p.331.
with-profit equity asset allocations, the actuarial profession found it difficult

to adequately influence their firms decision-making in the face of the stress
of competition. In all these cases, some actuarial thought-leadership was evi-
dent that could identify concerns and issues with the levels of associated risk
being generated, but actuaries ability to influence and lead life offices corpo-
rate decision-making had started to wane.
It was the guaranteed annuity options written in the 1970s and first half of
the 1980s that ultimately proved Dick Gwilt correct. At the time they were
originally written, these options were generally far out-the-money. Indeed, they
were viewed as so out-the-money that they need not be charged for or reserved
for in any way. Even in the 1950s, actuaries were aware that out-the-money
options had some value to the policyholder. Haynes and Kirton noted in 1952:
In practice, guaranteed options are always valuable to the policyholder, even

though a margin is retained between the current rate of interest and the interest
rate adopted in calculating the option.74
This observation, however, did not influence the business practices of British
life officesGAOs were considered sufficiently worthless to the policyholder
to be given without charge, yet also to be a product feature that was powerful
in driving sales. Guaranteed annuity rates varied, but typically GAOs written
in the 1970s and early 1980s provided a guaranteed annuity rate of around
11 % for a level annuity on a male aged 65. The a(55) m ortality table typi-
cally in use in 1970s life offices implied a break-even long-term interest rate of
67 % for the GAO.Consol yields in 1975 were around 15 % and hence the
options were deeply out-the-money. However, long-term interest rates had
been below 6 % in the early 1960s so there was little reason for the possibility
of rates again falling to these levels over a 25-year horizon to be considered
completely outlandish.
Two unrelated phenomena emerged over the 1980s and 1990s that trans-
formed GAOS from a deep out-the-money guarantee into a highly material
liability. The heady rises in long-term interest rates of the third quarter of the
twentieth century were reversed over the final quarter. Long rates fell inexo-
rably from the early 1980s through to the end of the century (and beyond):
the UK long-term interest rate fell from 13 % in 1981 to below 5 % by 1999.
Secondly, the mortality assumptions of the a(55) table, which was designed
for immediate annuitants in 1955, proved wholly inadequate for pricing
annuities at the time these contracts matured. Pensioner longevity improved
74
Haynes and Kirton (1952), p.185.
at an exceptional rate over the period from the 1960s to 1990s. For example,
the one-year mortality rate for a 65-year-old male in the Permanent Male
Assurances (PMA) 92 mortality table was less than half of that in the PMA
68 table. This was largely unanticipated by British actuaries. It had a material
impact on GAO liabilitiesthe break-even interest rate of a 11 % guaranteed
annuity rate increased from 67 % to 89 % when a(55) was replaced with a
typical 1990s longevity basis.
What actions were taken by life offices and their actuaries as the double-
whammy of lower long-term interest rates and lower pensioner mortality rates
emerged? Most offices stopped writing GAOs on new business during the late
1980s. By this time long-term government bond yields had fallen below 10 %
and the prevailing market annuity rate approached the guaranteed annuity rate.
Ceasing to write new GAO business could not, however, entirely stop the offices
from accepting further GAO liabilities on to their books: the terms of the exist-
ing business applied the guarantees to future regular premiums and sometimes
to one-off additional single premiums that the policyholder could choose to pay.
By 1993, the combination of further falls in long-term interest rates (by
then below 8 %) and lighter pensioner longevity assumptions brought sig-
nificant volumes of GAOs into-the-money for the first time. Some tempo-
rary respite was provided by the short-lived increase in long-term interest
rates during 19941996 but by 1997 rates had fallen below their 1993 levels.
UK insurance regulators were curiously inactive on the subject of GAOs over
this period. The scale of life offices GAO exposures was perhaps not well-
understood by regulators or indeed the life offices and their actuaries at this
time. In January 1997, the Life Board of the Institute and Faculty of Actuaries
established an Annuity Guarantees Working Party to establish the scale and
nature of the potential GAO problem and how life offices were managing or
planning to manage it. The terms of reference of the working party noted:
Currently there is no accepted practice for reserving for these guarantees and
there is no published research to guide Appointed Actuaries in setting reserves.
The DTI [Department of Trade and Industry, the UK insurance solvency regu-
lator at the time] have not published any guidance or regulations specific to
annuity guarantees.75
The working party published their report in November 1997.76 It provided

an insightful survey and intelligent commentary on the GAO situation at
Bolton etal. (1997), Appendix 1.

75
Bolton etal. (1997).

76
British life offices and the prevailing actuarial practices around it in the key
areas of reserving, hedging and bonus policy (including implications for PRE).
But it stopped short of providing any recommendations. It was never pub-
lished in the British Actuarial Journal or presented at an Institute or Faculty
sessional meeting. With hindsight it seems a curiously understated contribu-
tion from the actuarial profession.
The working party found that around half of British life offices with GAO
liabilities were ignoring them entirely in their statutory reserving at the time
of writing. The other half reserved for the greater of the cash maturity value
and guaranteed annuity value, as implied under the statutory reserving basis.
This reserving adjustment was straightforward in the context of GAOs on
unit-linked business, but the majority of GAOs were written on with-profits
business, and incorporating the GAO into the net premium valuation was less
straightforward. As the working party noted:
Adding in allowance for annuity guarantee reserves is another twist on top of a

series of potential adjustments to standard net premium reserves.77
The statutory net premium valuation included a resilience test that required
firms to calculate the capital required after some specified stresses to valuation
assumptions such as changes in equity values and interest rates. But, as was
noted in Chap. 3, the net premium valuation has a perverse sensitivity to
interest rate changes. In particular, it understates the economic balance sheet
impact of a fall in interest rates as a lower interest rate basis generates a higher
net premium. As a result of this, many life offices found they had greater
exposure to interest rate rises than falls under the net premium valuation. In
those cases, no resilience reserve was required for the contingency that GAO
costs would increase due to further falls in long-term interest rates (though
that only arose because they were required to hold reserves that incorporated
a loading for the imaginary fall in future regular premiums that would arise
if interest rates increased). This was an archaic actuarial feature that reflected
poorly on the professions ability to understand and manage financial risk.
The originators of the net premium valuation never intended it to be used
in solvency resilience testing, and it is an obviously flawed tool for analysing
exposure to future changes in interest rates.
UK long-term interest rates continued to fall substantially over the late
1990sfrom 8.0 % in 1996 to 4.9 % in 2000. GAOS were by this time
unambiguously in-the-money options. Dick Gwilts prediction that such
77
Bolton etal. (1997), Section 3.
options would become a serious embarrassment and source of loss had,

half a century later, finally crystallised. GAOs costs were no longer a possible
contingency. They had arrived and something had to be done. With-profit
funds estates could meet some or possibly all of the losses but the general
first preference of life offices in the late 1990s was to manage the GAO losses
through reductions in with-profit terminal bonus pay-outs. This led to the
much vexed question of whose terminal bonuses should be reduced to fund
the GAO cost, and how this stacked up against PRE.The simplest approach
would be for all current policyholders (both those with and those without
GAOs) to receive the same reduction in terminal bonus that applied to the
(pre-GAO) cash maturity value. This reduction could be calculated to fund
all GAO costs, or the cost could be shared between current policyholders and
future policyholders (i.e. the estate). Such an approach may or may not have
been deemed consistent with PRE, perhaps depending on the specific con-
stitutional arrangements of the given with-profit fund and the policyholder
communications it had made.
However, some life offices, most notably Equitable Life, the venerable
institution whose early history we discussed in Chap. 2, pursued a different
approach. Equitable Life had become something of an outlier in terms of
many life office practices in the second half of the twentieth century. Unlike
virtually all other life offices, it reported on a gross premium rather than net
premium valuation method; it did not sell any policies through independent
advisors but only through a tied salesforce; its appointed actuary was also
its chief executive through the 1990s until 1997. Most strikingly, and most
importantly for our discussion, it had a policy of not retaining any estate. The
idea was that all policyholders would receive an asset share-based pay-out,
and hence there would be no need for the office to hold any additional assets
beyond the assets accrued from the investment of existing policyholders pre-
miums (notwithstanding that the with-profit policies included contractual
guarantees whose ultimate costs could exceed the policyholders asset shares).
The Equitable believed that they had communicated their asset share-based
bonus policy such that the policyholder had effectively and knowingly man-
dated to the directors absolute discretion over each terminal bonus addition.78
The Equitable used this absolute discretion to make each individual policy-
holder bear the final cost of the GAO that they had been grantedthat is,
at an individual policyholder level, the terminal bonus was reduced so that
the post-GAO value of the policy would be equal to the cash maturity value
of the with-profit policy if it did not include a GAO.This meant the GAO
Corley etal. (2001), p.12.

78
was irrelevant to the value of the policy pay-out (except in the circumstance
where the GAO cost implied the required equalising terminal bonus was
negative). This was a fairly striking practice. Importantly from a PRE perspec-
tive, it was not a bonus policy that had ever been explicitly communicated to
policyholders when the policies were sold, or indeed at any time prior to its
implementation.
This bonus policy was questioned by some GAO policyholders through the
pensions ombudsman and the Equitable sought clarification from the courts.
The case went all the way to the House of Lords, the highest court of appeal
in England at the time. In 2000, the law lords found that the Equitables dif-
ferential terminal bonus policy was unlawful. The Equitable immediately cut
bonuses on all policies to zero in order to try to meet their GAO liabilities.
But the Equitable had sailed too close to the wind and was now floundering.
It put itself up for sale and closed to all new business in December 2000.
The House of Lords judgement and the closure of the Equitable to new
business sent shockwaves through the British actuarial profession. Within
weeks of the Equitable closing to new business, the Faculty and Institute
established a committee of inquiry to review its implications for the role of
actuaries in life offices and their actuarial professional guidance. It published
its findings in September 2001.79 It called for a rewriting of key aspects of
actuarial professional guidance, and for external peer review of the work of
the appointed actuary.
The hedging of guaranteed annuity options through the purchase of deriva-
tives from investment banks started in the late 1990s and received a fillip as
the Lords judgement removed other management options. Such hedges were
generally financed by with-profit funds estates, and they secured the financial
safety of the leading British life offices. The hedges cost much more than they
would have had they been purchased in 1997, but nonetheless have proven
valuable investments in the seemingly interminable environment of falling
long-term interest rates of the 2000s.
With the benefit of hindsight, how does the GAO story reflect upon the
British actuarial profession? In 2003, David Wilkie co-authored a paper with
Heriot-Watt Professor Howard Waters and the doctoral candidate S. Yang
that developed some retrospective analysis.80 They considered how actuar-
ies ought to have thought about charging and reserving for GAOs in 1985,
given the state of actuarial thinking at that time. Specifically, they argued
that the Maturity Guarantees Working Party paper of 1980, and the Wilkie
79
Corley etal. (2001).
80
Wilkie etal. (2003).
model presented to the Faculty in 1984, collectively provided actuaries with

an adequate conceptual and analytical framework to charge and reserve for
GAOs in an appropriate way. Their analysis showed that in 1985 the Wilkie
model would have implied that there was a c. 20 % probability of a GAO
attached to a 25-year endowment maturing in-the-money. A probability-of-
ruin reserving approach would have implied that a reserve of around 10 %
of a single premium would be required at a 95 % probability level. Based on
the guarantee charging approach used in the Maturity Guarantees Working
Party (not an option price but the expected cost plus charge-for-cost-of-cap-
ital approach), a charge for the GAO of around 5 % of the single premium
would have been justified. There can be little doubt that had such reserving
and charging parameters been in place, the result for GAOs would have been
similar to that of unit-linked maturity guarantees: a lot less would have been
written.
Wilkie, Waters and Yang were, of course, factually correcta stochastic
modelling approach to pricing and reserving for guarantees had been well-
established in British actuarial thought by the mid-1980s. However, at this
time, these concepts had only been rigorously applied within the domain of
unit-linked business. In the 1980s there was no serious backing from the actu-
arial profession to apply these concepts to reserving for any of the guarantees
in with-profits business. Some leading British life offices were undertaking
internal stochastic analysis of their with-profit business in the 1980s, but few,
if any, were suggesting that statutory reserving for with-profits business be
undertaken in the way that had been implemented for unit-linked maturity
guarantees.
In short, in the 1980s and 1990s there existed a misplaced confidence
within the actuarial profession that with-profits business provided enough
risk management levers to the actuary to make stochastic analysis unnecessary,
difficult to apply and of limited value. Redingtons 1976 statement that the
actuarial analysis of with-profits was a countryside to explore on foot and not
by fast car is the most colourful expression of this perspective. There is doubt-
less some truth in Redingtons view: it is easy to substitute computer model-
ling output for hard thought, and for insight to be lost amongst a plethora of
ill-understood random numbers. But, by the 1980s and 1990s, was the actu-
arial profession doing enough walking of the with-profit countryside? And
was the profession guilty, through hubris or fear, of rejecting a whole body of
thinking that had emerged in the previous decades which could provide the
key to the measurement and management of the increasing financial risks that
it was charged with managing? Boyle and Hardy published a GAO actuarial
retrospective in 200381 which forcefully concluded:
Boyle and Hardy (2003).
81
This entire episode should provide salutary lessons for the actuarial profession.
It is now clear that the profession could have benefited from greater exposure to
the paradigms of modern financial economics, to the difference between diver-
sifiable and non-diversifiable risk, and to the application of stochastic simula-
tion in asset-liability management [this] would have enabled insurers to
predict, monitor and manage the exposure under the guarantee.82
It is evident that, particularly during the 1980s and 1990s, there was a
strong reticence on the part of the British actuarial profession to adopt or
make use of financial economics in their thinking and best practices. With
hindsight, these were partly wasted decades for the profession in this respect.
On a positive note, after taking their direction from regulators, financial
reporting standard-setters and judges, real progress was made in the 2000s in
actuarial education, research and practices that started to rectify this damage.
82
Boyle and Hardy (2003), p.150.
6
British Actuarial Thought inDefined
Benefit Pensions (19051997)
The recognisable British private staff pension scheme, funded by a mixture of

employer and employee contributions based on percentage of salary, paid over
the period of service of the employee and invested in a trust fund, with benefits
defined with reference to the salary history and years of service of the employee,
first emerged over the second half of the nineteenth century. Ad hoc unfunded
pension arrangements arose amongst some large employers earlier in that cen-
tury. The occasional historical precedent of earlier funded arrangements for
post-employment benefits can also be found: the earliest funded scheme for
the provision of widows annuities to employees is thought to be the Scottish
Ministers Fund, which was established in the mid-eighteenth century.1
Actuaries have provided technical and professional advice on the financial
management of defined benefit pension funds since around 1875. The British
profession has been actively engaged in this role continuously from then to
the present day. Whilst the task of funding long-term life-contingent pensions
had many parallels with the actuarial work in the more-established life assur-
ance sector, pensions work has also always had its own unique requirements
and challenges: typically, pension benefits were even longer-term than the
liabilities that arose in the life sector; and the ultimate size and cost of the ben-
efits were a function of a wider array of uncertain variables than the liabilities
of life offices. Whilst rates of interest and mortality tables were similarly vital
to pensions work, the cost of the pension benefit would also depend on other
variables such as long-term salary inflation and the rates at which employees
would leave their employer prior to retirement. The trust arrangement that
Crabbe and Poyser (1953), p.1.
1

DOI10.1007/978-3-319-33183-6_6
formed the bedrock of funded pension arrangement also implied a different

legal and commercial context to that of a mutual life office. This raised subtle
issues about the purpose of funding which had important implications for the
objectives that the actuarys advice should attempt to meet.
Amidst these new challenges, there was a plethora of external drivers of
change in pensions actuarial thought that will be familiar from the history
of actuarial thought in the life sector: economic and political shocks; increas-
ing government regulation in both solvency and consumer protection that
gradually saw statutory responsibilities displace unbridled actuarial judge-
ment and discretion; financial reporting metrics that demanded alignment
with accounting principles; the emergence of financial economics and the
new insights it could provide into financial management and the difficulty in
reconciling them with traditional actuarial approaches.
This chapter identifies some of the key developments in pension actuarial
thought that arose over the long history of actuarial involvement in the field.
Their historical context can help us understand the ways that thinking in this
field meandered, circled and occasionally underwent revolutions. A final word
on terminologythe terms final salary pension scheme, defined benefit pen-
sion fund and so on are used interchangeably in this history, and none gener-
ally provide a perfect description of how these pension arrangements worked.
Most of the schemes had pensions based on final salary, but a minority were
based on salary levels over a longer period of employment. Pension benefits
were generally not always prescriptively defined: for significant periods of the
history early leaver benefits and in-force pension increases included signifi-
cant discretionary elements that would tend to be greater when funding levels
were strong.
arly Thought onDefined Benefit Pension Funds

E
(19051921)
Early actuarial thought on defined benefit pension funds focused on the devel-
opment of the algebra and substantial mechanical computation necessary for
the valuation of a schemes liabilities and the assessment of a required contri-
bution rate under a given set of deterministic assumptions for the schemes
future development. In these early years, there was considerable discussion
of how to set the decrement bases for the liability projection, but relatively
little focus at this time on how to set the valuation interest rate. Asset valua-
tion was rarely, if ever, addressed. Investment was predominantly in long-term
6 British Actuarial Thought inDefined Benefit Pensions (19051997) 235
g overnment bonds, and in the late Victorian and Edwardian eras, long-term
interest rates were remarkably stable by later standards: between 1875 and
1910 the long-term government bond yield moved within the range of
2.03.2 %. Valuing assets at the lower of book value and market value seemed
simple, prudent and adequate.
This actuarial work was fundamentally similar in spirit to the reserving
calculations made in life offices, but the calculations were even more convo-
luted and computationally intensive. Ralph Hardy developed the first defined
benefit pension fund valuations around 1875. In doing so, he originated
the commutation factor as a vital computational aid for this task.2 Hardys
work, however, was never formally published. The earliest British actuarial
professional papers on funded pension schemes were published during the
Edwardian era by disciples of Hardy such as George King (in 1905) and
Henry Manly (in 1911).3 Actuarial papers on funded pension schemes were
nonetheless sparseprior to the conclusion of Second World War, only a
handful of notable papers on pension schemes were published in the British
professional actuarial journals.
Kings paper presented the actuarial calculations that were required for pen-
sion fund valuations and the assessment of the contribution rate that would
fund the ultimate benefitsthat is, to find a contribution rate expressed as
a constant proportion of salary that had a present value equal to the present
value of the liabilities. It was the first paper to set out the detailed calculations
that were necessary to allow for treatment of the various decrements (death,
pre-retirement withdrawals and retirement) and assumptions (mortality rates,
interest rate, withdrawal rates, salary growth) in the valuation of the liabilities
(pensions, death benefits, withdrawal benefits) and the setting of the contri-
bution rate required to fund the projected liabilities.
King suggested that the withdrawal, mortality and retirement rates should be
estimated by each scheme based on their own recorded experience. These could
be suitably graduated using methods such as those developed by Woolhouse for
life office mortality experience tables. He emphasised how these assumptions
should vary across pension funds and railed against the use of assumptions
based on pooled pension fund data. He recognised, however, that scheme-
specific data would seldom be available for pensioner mortality at the older
ages and he advocated the use of one of the published population mortality
tables such as English Life Table, No. 3 for this purpose. He suggested a salary
scale should be specified (i.e. how salaries vary by age) again based on the actual
See King (1905), pp.129 and 175.

2
King (1905), Manly (1911).

3
experience of the scheme membership. Inflation was a relatively subdued eco-

nomic phenomenon in this erathe UK inflation index was at the same level
in 1904 as it had been in 1860; so the salary scale was intended to represent the
growth in salary due to promotion and experience. In the examples he gave,
he assumed that salaries would increase by about 80 % between the ages of 20
and 30, but by only 12 % between 50 and 60.
In considering the setting of the contribution rate, King only considered
the calculation at the inception of the scheme, and hence avoided addressing
the valuation of any pre-existing scheme assets. Nor did he discuss how the
valuation interest rate should be set. It was simply a parameter that was sub-
sumed into Hardys commutation functions.
Manlys 1911 paper further discussed the estimation of the member decre-
ment assumptions, and he also provided some analysis of the sensitivity of
the valuation to those assumptions. Manly was more sanguine about pooling
the experience of different pension funds for the purpose of estimating decre-
ment rates, providing the pension funds reflected a similar profile of mem-
ber. The paper included a Life and Service Tablethe withdrawal, mortality
and retirement rates and salary scale based on scheme experiencefor the
aggregated experience of several large railway schemes. He noted that selec-
tion effects would produce pension scheme mortality that would typically be
lighter than that observed on assured lives. Members who were seriously ill at
pre-retirement ages would tend to withdraw from the scheme as they would
no longer be able to work. The surrender of a life assurance policy, on the
other hand, would tend to be made by a person in good health.
Manly highlighted the sensitivity of the projection of the pension scheme,
and the contributions required to fund the liabilities, to these assumptions
by comparing the results obtained with those produced by assumptions pub-
lished in other early studies. He showed that his life and service table implied
that a 5 % salary contribution would fund a pension of 1.55 % of average
salary for every year of membership whereas alternative previously published
assumptions would support 2.38 % of average salary for every year of mem-
bership.4 Manly also provided detailed year-by-year projections of the pension
scheme asset values, contributions and benefit cashflows that would be gener-
ated over a 90-year projection using his assumptions and the alternative. This
highlighted how the solvency of the pension scheme could go awry if experi-
ence deviated significantly from the basis assumed in setting the contribution
rate, thus highlighting the need for careful and regular monitoring of the
pension funds experience by the actuary.
Manly (1911), p.157.

4
The Staple Inn discussion of Manlys paper generated some interesting

debate on the purpose of advance funding of pension liabilities that illumi-
nated the actuarial views of the time. One speaker, Thomas Tinner, noted that
a primary reason for advanced funding was to provide security to the pension
fund member:
Without a fund, or a guarantee from the employer, there was no security that
when a man who had been paying all his period of service for the pensions of
others came to retire his own pension would be provided by a younger genera-
tion. This was especially the case if the employer was a private firm.5
He also argued that the interest earned on advance funding of the pension
benefits would reduce its cost:
The effect of accumulating a fund was to reduce the cost of the pensions, because
the interest earned would help to pay the outgoings.6
Meanwhile, W.O.Nash argued that the main aim of the pension scheme
was to recognise a smooth accrual of the cost of the pension provision:
What they were aiming at in starting a Pension Funds was an equalization of the
annual expense for the pension.7
As we shall see below, these three reasons for advanced funding would be
revisited and debated numerous times by actuaries, accountants and econo-
mists over the remainder of the twentieth century. The relative importance
attached to each of these reasons would influence the choice and development
of actuarial methods used in pension fund valuation.
The economic consequences of the First World War represented the first
great economic shock to test British defined benefit pension schemes. It
created considerable stress for the financial health and solvency of the pen-
sion funds. It also created new challenges for pension actuaries, particularly
with respect to the treatment and management of inflation. Over the Pax
Britannica periodthe 100 years between the end of the Napoleonic War
and the start of the First World Warinflation rates had at times fluctuated
markedly both up and down but there was no discernible consistent upward
trend in UK consumer prices. Indeed, the UK Consumer Price Index stood
5
Tinner, in Discussion, Manly (1911), p.219.
6
Tinner, in Discussion, Manly (1911), p.219.
7
Nash, in Discussion, Manly (1911), p.222.
some 20 % lower at the outbreak of the First World War in 1914 than it had
been at the end of the Napoleonic Wars in 1815. Thus, the risks associated
with future price inflation did not feature prominently in actuarial thought
on final salary pension schemes prior to 1914. No explicit assumptions for
price inflation or real salary growth featured in the work of King or Manly.
Their age-dependent salary scales were intended to capture the typical path of
career progressions, not economic inflation.
In the years 19151920 inclusive, the UK CPI index more than doubled.
Salaries and wages increased at a similar rate. This substantially increased the
value of pension liabilities without generating an offsetting increase in the
value of pension fund assets (which, as noted above, were predominantly
invested in long-term government bonds at this time). This scenario was radi-
cally different to anything within living memory of actuaries. The possibil-
ity of such a scenario and its consequences for pension funds had, naturally
enough, not been seriously considered by pension actuaries prior to the war.
In 1921, G.S.W. Epps wrote an important, and at the time controver-
sial, paper8 on the implications of this experience for pension fund valua-
tion, noting in his introduction that [due to] enhanced scales of salaries or
wages which have resulted from the change in the value of money it is to
be feared that in many funds a very serious position has to be faced.9 Epps
used case studies to highlight that, for older, established pension schemes, the
impact on the valuation of accrued service benefits and the resulting solvency
levels was very substantial. Furthermore, at this time pension funds were not
required to inflation-link pensions in payment, and Eppss valuations made
no allowance for such increases. But significant pressure bore on the schemes
to provide some increases to pensions in payment in order to mitigate the
significant reduction in real value of pensions which had been inflicted on
pensioners over the previous decade.
This exceptional economic environment also naturally had an impact on
the asset side of the pension fund balance sheet. Between 1900 and 1920,
long-term government bond yields more than doubled from 2.5 % to 5.3 %.
The implications of this interest rate rise for the valuation of the assets and
liabilities of pension funds was a topic of considerable controversy within the
actuarial profession at the time. Epps recognised that the net impact of such
interest rate rises may not necessarily be adverse for a pension fund:
Epps (1921).
8
Epps (1921), p.405.

9
It is, of course, true that materially higher rates than those assumed in past valu-
ations can be looked for in the case of the accruing funds to be invested for
many years to come; but the question as to how far it is prudent to meet past
depreciation by raising the valuation rate of interest is one of extreme difficulty,
the solution of which throws a great responsibility on the actuary concerned.10
The topics of asset valuation and depreciation and the related treatment of
the liability discount rate received much attention at the Staple Inn discussion
of Eppss paper. Some actuaries were reluctant to recognise the market value
impact on bond prices, even if not to do so was evidently imprudent. For
example, R.G.Maudling commented:
As to the general treatment of depreciation, he might be accused of heresy, inas-

much as he could not bring himself to write securities down to their present
market value He preferred to make a reserve for possible loss on realization,
and to give considerable weight to the fact that securities might ultimately
recover. He was not inclined to give much credit to present market values
because they were more or less fictitious, being the values which a man who was
compelled to realize was prepared to take, and he thought it would be extremely
unwise to write everything down to that level.11
J.Bacon expressed a quite similar view to Maudling:
He would certainly not write down investments to their present market price,
but would simply set aside an investment reserve and write off any deficiency
there might be by means of an annuity, believing as he did that in five or ten
years time, when they got to a state of equilibrium, the annuity set apart would
probably not be required, and by that means he would have succeeded in giving
a degree of safety to the fund without unduly crippling the employer.12
So, the above actuaries advocated valuing assets above market value,
though with perhaps a partial write-down from their book value. A reserve
would be included on the liability for future realisation of investment losses,
but this would be smaller than the difference between the assets actuarial
valuation and their market value, reflecting the actuarys confidence that the
assets would subsequently increase in value as rates reverted back to histori-
cal norms. No change in the liability discount rate would be made using this
10
Epps (1921), p.410.
11
Maudling, in Discussion, Epps (1921), p.444.
12
Bacon, in Discussion, Epps (1921), p.448.
approach. The current market yield was treated as an aberration that could be
largely assumed away. However, S.G.Warner, an influential actuary who was
Institute President during the years 19161918, suggested a quite different
approach where assets were fully marked down to their current market values
whilst the liability discount rate was increased consistently to reflect those
market yields:
Depreciation should be severely dealt with, but that there should be little hesita-
tion about using something nearly approaching the rate of interest which that
depreciation revealed as having been secured.13
This is perhaps the earliest actuarial advocacy of a market-based pension

fund valuation. As we shall see below, the topic of consistent pension fund
asset and liability valuation in the presence of significant changes in mar-
ket interest rates would occupy pension actuaries thoughts for much of the
remainder of the twentieth century.
onsistent Asset andLiability Valuation

C
(19481963)
The exceptional inflation experience of the years immediately following the
First World War proved to be a transitory phenomenon. Indeed, over the
course of the 1920s, a significant portion of the price increases of the post-
war years were reversed by a decade of deflation. By 1933, the UK CPI stood
at a lower level than at the end of the First World War. The Second World
War inevitably generated a further bout of inflation, but in its immediate
aftermath the political stancein particular, that of Hugh Dalton, the then
Chancellor of the Exchequerwas to seek very low long-term interest rates.
In 1946, long-term government bond yields stood at 2.6 % (they were 4.6 %
20 years earlier). The position pension actuaries faced was therefore reversed
from the time of Eppss 1921 paper: pension funds bond assets now showed
a considerable market value appreciation, whilst the prospective future avail-
able yield on new money was considerably lower than had been assumed in
pension valuations.
C.E.Puckridge published a paper in the Journal of Institute of Actuaries in
1948 which analysed these circumstances and their actuarial consequences.14
Warner, in Discussion, Epps (1921), p.449.

13
Puckridge (1948).
14
In the late 1940s, as 40 years earlier, standard actuarial practice was to value
assets at book value or, if lower, market value (or perhaps somewhere in
between), with liabilities valued using a broadly static valuation interest rate
assumption. The liability valuation rate could be set with reference to the
investment yield implied by the book value of assets, with some allowance
made for how that yield was expected to evolve in the future. Puckridge, how-
ever, argued for an alternative approach that entailed a more systematic, con-
sistent approach to the valuation of pension fund assets alongside liabilities:
Value assets (including existing investments) and liabilities at the rate of interest
which it is anticipated can be earned on future investments.15
Puckridge did not provide much specific guidance on how to estimate this
interest rate, but he did note that the rate will be
determined by considering the average rate at which investments could be

made at the present time, taking into account the range of investments autho-
rized for the particular fund and the proportion likely to be invested in the vari-
ous classes; adjustments will have been made to allow for any interest margin
that it is desired to incorporate and for the view taken on the probable trend of
interest rates over a long period.16
Puckridges future investment rate was an inherently long-term and subjec-

tive parameter. He likely envisaged that it would result in more stable asset
and liability values than the market-based approach suggested by Warner back
in the discussion of Eppss paper in 1921. But the vital point of Puckridges
method was that the actuarial assessment of the financial health of pension
funds required greater and more transparent consistency between the valu-
ation of assets and liabilities than was found in the then standard actuarial
practice. Whilst this may appear obvious in hindsight, it was not a particu-
lar point of emphasis in the previous half-century of actuarial thinking on
defined benefit pension funds. Though Puckridge preferred to reject the direct
use of market prices, he did recognise the difficulty in placing a value on assets
that exceeded their current market value:
It would in practice be difficult to adopt any value in excess of the total market
value, and the balance would therefore be retained as a margin which would fall
into surplus in the future as the excess interest is received.17
15
Puckridge (1948), p.2.
16
17
Puckridges idea of valuing assets by discounting their future income at an

interest rate that is consistent with the rate applied to the liabilities gained
considerable ground within the actuarial profession over the following 15 or
so years. Eventually, a form of this discounted income approach to the valu-
ation of all assets in a defined benefit pension fund would become standard
actuarial practice. Of course, S.G.Warners advocacy back in 1921 of using
market values of assets along with a market-based liability discount rate could
be viewed as one example of consistent asset and liability discounting. But
pension actuaries were generally reluctant to shift their long-term liability
valuation assumptions by the degree that could be implied by market move-
ments. After all, in the context of open, expanding defined benefit pension
funds, there would never be a need to sell assets, and actuarial judgement
could apparently form a better view of yields available for future investments
than current market prices. Puckridges embryonic notion of setting a stable
long-term liability discount rate and valuing the projected cashflows of both
assets and liabilities at that rate emerged as the pension actuarys solution to
consistent asset and liability valuation that would provide valuation stability
through time.
Three actuarial papers published between the years of 1958 and 1963 were
particularly important in further developing Puckridges idea and establish-
ing its practice in British pension scheme valuation. Once again, economic
conditions provided a spur to this progress. Following the post-war Dalton
era of politically induced low interest rates, economic reality had reasserted
itself. Long-term bond yields reverted from 2.6 % in 1946 to 5.0 % in 1958.
Pension actuaries were once again wrestling with the consequences of substan-
tial bond asset depreciation.
The first of these three papers was written by D.F.Gilley and D.Funnell.18
Gilley and Funnell, like Puckridge a decade before them, argued that both
asset and liability cashflows should be discounted at the average yield on
future investments. However, in doing so, they also noted some of the per-
verse consequences that this may have. In particular, if this approach resulted
in a deficit being identified in the scheme, the amount of money required to
extinguish this deficit would not necessarily be equal to the assessed size of
the deficit. Under the discounted income approach, actuaries were essentially
operating in a parallel currencyactuarial valuation pounds had a different
value to real-life pounds. But they argued that their approach of valuing both
sides of the balance sheet with a single discount rate was preferable to attempt-
ing to find consistency between liability valuation and book asset valuation by
Gilley and Funnell (1958).

18
creating a discount rate that was an ill-defined blend of the current book yield
and the assumed average yield on future investments.
Gilley and Funnell moved beyond Puckridges work when they considered
the valuation of equity assets within their discounted cashflow approach. As
we will discuss further below, equities became an increasingly important asset
class for pension schemes during the 1950s. Any asset valuation method for
British pension funds therefore needed to be directly applicable to equities.
The simultaneous valuation of equities and bonds immediately gave rise to
the question of whether and how the asset mix of the pension fund should
impact on the actuarial valuation of assets and liabilities. Should a portfolio of
equities and a portfolio of bonds, each with the same market value, command
the same actuarial asset value? Should a change in the asset mix result in a
different valuation of liabilities through a different choice of liability discount
rate? In a nutshell, should the higher expected return of equities be capitalised
in the current assessment of the pension fund surplus/deficit?
Gilley and Funnell were unambiguous in their view: it would not be pru-
dent for the actuary to reduce the contribution rate or weaken the basis of
valuation by assuming a higher valuation rate of interest than the expected,
long-term, gilt-edged rate.19
So Gilley and Funnell were clear that the asset allocation choices of the
pension scheme should not impact on the valuation discount rate applied to
its projected asset and liability cashflows. However, that did not mean that
they advocated the use of the market value of the equity holdings. They sug-
gested that the market value of the equity portfolio be scaled by the ratio of
the actuarial valuation of irredeemable consols (as implied by the valuation
interest rate) to the market value of those consols. Or, equivalently, the equity
market value should be scaled by the ratio of the market consol yield to the
valuation interest rate, i. In this approach, the actuarial value of total assets
would be the same for any split of assets between equities and consols. Thus,
the asset allocation choice would not impact on the overall actuarial valuation
of assets or liabilities. However, they added a caveat: the equity valuation may
be reduced by a margin of arbitrary amount if the actuary held a strong view
that they were over-valued in the market.20
These asset valuation ideas were quite controversial at the time. Traditionally,
actuaries viewed the writing-up of asset values above their book values as a
dubious and imprudent practice. Gilley and Funnells approach could see
assets valued significantly in excess of not just book values but market values.
19
Gilley and Funnell (1958), p.50.
20
Gilley and Funnell (1958), p.53.
The minutes of the discussion of their paper highlights the suspicion and hesi-
tancy with which conventional actuarial thought welcomed these ideas. Gilley
and Funnells essential point, however, was that subjective assumptions about
future investment conditions must arise in pension fund valuations, and their
method provided a more coherent and transparent approach to setting those
assumptions.
Heywood and Landers 1961 paper, Pension Fund Valuations in Modern
Conditions,21 was the next significant contribution to British actuarial
thought on the consistent valuation of pension fund assets and liabilities. Like
Puckridge and Gilley and Funnell before them, Heywood and Lander again
argued that the valuation discount rate should be set as the long-term interest
rate that is expected to be earned on new money; and that assets should be
valued consistently with that rate. However, Heywood and Lander differed
from Gilley and Funnell in two fundamental respects. Firstly, whereas Gilley
and Funnell had argued that the discount rate should reflect the expected
future long-term gilt yield irrespective of the actual asset allocation of the pen-
sion scheme, Heywood and Lander believed that the expected return on the
actual asset allocation of the pension scheme should be used. This meant that
the liability valuation would be reduced by increasing the funds allocation to
equities. They suggested that a 100 % gilts allocation would merit a 3.5 %
discount rate, whilst a 50/50 equity/gilts allocation would justify a discount
rate of 4 % or possibly even higher.22
The second, and related, point on which Heywood and Lander diverged
from Gilley and Funnell was on the valuation of equity assets. Equity invest-
ment was becoming ever more material for UK pension schemes. Heywood
and Lander noted that a typical approach at the time of writing was to invest
50 % of assets in equities and 50 % in fixed income. Equities had proved
attractive to pension funds both as an inflation hedge and as a high-yielding
asset class (though the reverse yield gap was just starting to emerge at this
time). Like Gilley and Funnell, they noted that the practice of valuing assets
at the lower of book value and market value was still prevalent. And they
expressed strong unease with the use of market values in general on the
grounds of their volatility:
It seems very difficult to justify a method of valuing assets which places a value
upon the fund which may alter substantially if the valuation date were varied by
only some two or three months.23
21
Heywood and Lander (1961).
22
Heywood and Lander (1961), p.323.
23
Heywood and Lander (1961), p.327.
They also noted that there was no reason to expect book values to pro-
vide an asset valuation that was consistent with the liability valuation basis.
Like Gilley and Funnell, they concluded that a discounted income approach
to asset valuation, using a discount rate consistent with that applied to the
liability outgo, was the only way to obtain a consistent asset and liability
valuation. For bond valuation, Heywood and Lander concurred with Gilley
and Funnell: the projected cashflows of the bond could simply be discounted
at the liability discount rate. For equities, they argued that equity dividends
should be explicitly projected using a dividend growth assumption before dis-
counting at the valuation rate. Thus, equities could be valued as a perpetuity
discounted at the rate (i) where i was the liability discount rate, and was
the assumed dividend growth. The dash was used because was defined as the
assumed rate of salary inflation that should be applied in the liability valua-
tion. Salary inflation was typically not explicitly allowed for in liability valua-
tion at this time, but Heywood and Lander argued that credit for inflationary
dividend growth on the asset side of the balance sheet should only be taken if
consistent treatment of inflation was applied on the liability side of the bal-
ance sheet. Moreover, they specified that should be less than (though they
did not provide any specific rationale for this).
So, under Heywood and Landers approach, the relationship between the
actuarial valuations of equity and bond holdings were not directly constrained
by their relation to their respective market values. This meant that a change
in the equity/bond mix could change the actuarial valuation of assets as well
as the actuarial valuation of liabilities. Shifting from bonds to equities would
typically improve their assessment of the pension fund solvency position.
However, their stipulation that, for reasons of consistency, this improvement
should only be permitted if the liability valuation recognised a level of salary
growth consistent (and not less than) the assumed equity dividend growth,
meant that their approach would also be accompanied by a general strength-
ening of the liability valuation basis.
Finally, perhaps to hedge their bets, Heywood and Lander permitted a scalar
parameter to be applied to the equity valuation as an arbitrary multiplier.
This was intended to allow the actuary to make any special adjustments that
he deemed appropriate in the context of the valuation exercise. Whilst this
might appear as something of a fudge, it prevented their methods from being
cast as mechanical rules to be mindlessly applied by the actuary. It emphasised
the application of actuarial judgement that was valued by the profession. In
the Staple Inn discussion of the paper, Gilley suggested:
Truly, the actuary valuing the fund was in a better position to assess the relative
values to the fund of ordinary shares and irredeemable gilt-edged securities than
was the market.24
Not all participants in the discussion held to the view however that actuar-
ies were better equipped to value equities than the market. J.G.Day, whose
writings we shall discuss below, commented:
I am disturbed by the idea expressed in the paper that an actuary should be

prepared to value ordinary shares in a way quite independent of market value,
and that it should be presumed that his value would be superior to that of the
market.25
The 1963 paper,26 The Treatment of Assets in the Actuarial Valuation of a

Pension Fund, by Day and McKelvey represented the final instalment of this
periods trilogy on consistent valuation of pension fund assets and liabilities.
They once again started from the basis that the assumed value for the valua-
tion rate of interest should be the average rate of interest that is assumed for
future investment over a long future term.27
We saw above that Gilley and Funnell suggested that equities should be
valued as their market value scaled by the ratio of the market consol yield to
the valuation interest rate. Day and McKelvey suggested a variation on this
theme: they argued equities should be valued as their market value scaled
by the ratio of the equity dividend yield to the valuation interest rate. As the
product of the market value and the dividend yield is the current dividend
payable, it can be seen that this formula simply values the current dividends
as a perpetuity at the valuation interest rate i. In what was now becoming an
established actuarial asset valuation tradition, they also allowed an additional
scalar parameter as an arbitrary factor.
Clearly, this approach to equity valuation ignored future dividend growth
and they suggested this valuation formula should only be used when inflation
is ignored in the liability valuation. They then suggested that the discount rate
could explicitly be set as a real rate that could be applied to both sides of the
balance sheet, and suggested a real rate of 3.54 % would be appropriate. This
was fundamentally the same result as Heywood and Lander: it was, in essence,
a dividend discount method where the markets required return on equities
24
Gilley, in Discussion, Heywood and Lander (1961), p.342.
25
Day, in Discussion, Heywood and Lander (1961), p.364.
26
Day and McKelvey (1963).
27
Day and McKelvey (1963), p.108.
had been replaced by the pension funds valuation interest rate (which, in
turn, was to be set to reflect the long-term expectation for the pension funds
asset return). Under these approaches, changes in equity market values that
arose from changes in the dividend yield rather than changes in dividend pay-
outs would have no impact on pension fund valuations of equity assets.
There was a broad consistency in the thinking of this collection of papers.
They all eschewed market asset values on the grounds of consistency with a
liability valuation that was to be made using stable assumptions. But some,
like S.G.Warner many decades earlier, felt like this was tackling the consistent
valuation problem from the wrong end. In the Staple Inn discussion of Day
and McKelveys paper, Plymen commented:
They were assuming a certain figure for the rate of interest for valuing the liabili-
ties, and twisting the valuation of the assets round to be consistent with that
basis. Why not start off with the market value of the assets and try to deduce
from that basis a consistent system for valuing the liabilities?28
Speaking at a Staple Inn meeting over 30 years after the publication of the
paper, McKelveys son, K.J. KcKelvey, also an actuary, commented on the
explanation his father had given him for tackling the consistency problem
this way round:
The sole objective of that 1964 paper was to find a consistent basis for valuing
assets, given an existing methodology for valuing liabilities. The liability valua-
tion basis was off-market, by convention at that time. Therefore the asset valua-
tion inevitably became off-market. The main aim of the authors was to move
away from the valuation of assets by book value, which was still common. They
simply did not think about market values since there were no formal discon-
tinuance tests.29
We will see below how funding objectives such as discontinuance testing

influenced valuation thinking in the latter decades of the twentieth century.
But first, we will review how actuarial thinking on investment strategy for
pension funds developed alongside the above ideas on asset and liability
valuation.
28
Plymen, in Discussion, Day and McKelvey (1963), p.134.
29
K.J.McKelvey, in Discussion, Exley, Mehta and Smith (1997), p.950.
ctuarial Thought onPension Fund Investment

A
Strategy (19571985)
As noted above, the exceptional UK inflation experience that accompanied
the First World War and its aftermath provided a salutary lesson to pension
actuaries on the vulnerability of pension fund solvency to inflation shocks. But
this period of extraordinary inflation proved short-lived: between the years of
1921 and 1941, the UK CPI actually fell. However, the post-Second World
War period saw the return of high inflation, and, perhaps equally importantly,
high levels of inflation volatility and uncertainty. By the mid-1950s, there was
a growing sense that the prevailing political and economic environment made
this inflationary experience the new normal rather than merely a transient
post-war adjustment. The political focus on full employment and the increas-
ing strength of trade unions appeared to make the prospect of a permanent
inflationary environment quite plausible.
Pension actuaries of the post-Second World War era recognised the inflation
risk exposure that arose when investing in long-term nominal fixed interest
assets to back salary-linked liabilities. At this time, there was no index-linked
gilt market. The only available real asset classes were equities and property.
Actuarial thought in pensions started to embrace the use of equities as a form
of inflation hedge in the mid-1950s. As we have seen in earlier sections, this
was also a period of increased equity investment for life offices, and the cult of
the equity was more generally in full flight. Pension funds were not immune
to this trendbetween 1945 and 1954, UK pension schemes average asset
allocations to equities increased from 10 % to around 30 %.30 The inflation
hedging argument may have driven this, or may have simply provided an
intellectual rationale for it.
An actuarial paper by McKelvey, Pension Fund Finance, published in
1957, was amongst the first to argue for substantial equity investment by
pension schemes on the basis of equities inflation hedging ability.31 The qual-
ity of inflation hedging that equities could provide was accepted as a mat-
ter of some uncertainty by McKelvey. He constrained his advocacy of equity
investment to open rather than closed pension funds, and emphasised that it
was the inflation-protection of long-term dividend growth (rather than equity
market values) that he believed was important for such funds. He posited
that the propaganda value of balance sheet asset valuations of open funds
was of no great importance. In essence, McKelvey was arguing that dividends
McKelvey (1957), p.136.

30
McKelvey (1957).
31
provided a form of real cashflow match for salary-linked benefits. Volatility in

real equity price levels could be ignored as an inconsequential market foible
for an open pension fund that would never need to sell them. He emphasised
that, in real terms, long-term nominal fixed income was the risky asset class
for backing salary-linked pension schemes:
The question now is not, as it used to be, dare we put more than 10% in equi-
ties? It is, dare we leave more than 50% in fixed interest investments?32
J.G. Day published a paper in the Journal of the Institute of Actuaries in

1959 that provided further support for substantial equity investment by final
salary pension funds.33 Day emphasised how final salary pension liabilities had
different characteristics to life office liabilitiesspecifically being real rather
than nominal, and having longer duration. He argued that these characteris-
tics made pension funds more suited to equity investment than life offices and
noted the high levels of equity allocation that life offices had made in recent
years. He recognised that equity investment would expose pension funds to
more market value volatility, but, like McKelvey, he argued that this was not
fundamentally important. He was unequivocal in his advocacy of equities as
the basic investment of final salary pension funds:
A pension fund by the nature of its liabilities is so vulnerable to both inflation

and any movements towards higher wages and salaries and a higher standard of
living that, if one takes the view for the normal pension fund, it is income that
matters and provided that income is secure then fluctuations in the value of the
portfolio do not matter, then in the authors opinion one should come down on
the side of equities (or property) as the basic investment for a pension fund.34
The classic papers on pension fund valuation of the early 1960s that were
discussed above also contained further advocacy of equity investment for final
salary pension schemes, primarily on the basis of inflation hedging. For exam-
ple, in the discussion of Heywood and Lander, Hemsted commented:
Rising standards derive from increased productivity which in turn is based on

increasing capital backing per employee. Much of this increased capital arises
from retained earnings by industrial companies and these retentions in a period
of expansion produce increasing equity asset values per share. There is therefore
32
McKelvey (1957), p.121.
33
Day (1959).
34
Day (1959), p.130.
reason to think that ordinary shares have a built-in correction not only for fall-
ing money values [price inflation] but also for increasing standards of living [real
salary growth].35
Day and McKelveys 1963 paper was unsurprisingly supportive of pension

fund equity investment given the positions they had taken in their individual
papers of the late 1950s. They wrote:
It has been widely and authoritatively suggested that, to meet the dangers of
inflation, pension funds should invest in equities. It has been argued that with
inflation a businesss real assets increase in money terms so that, although
inflation may initially cause difficulties and distortions, an equity investment
will eventually have increased in earning power in rough proportion to the effect
of inflation (or very often rather more, as prior charges lose in value).36
All of the above development in pension fund thinking had taken place
under the premise that the raison dtre of the pension fund was to generate
asset income that, together with a stable pattern of contributions from the
employer and possibly also the members, would reliably fund liability outgo
as it fell due. (There was also an implicit assumption that the uncertainty in
long-term real dividends was much less than could be inferred from short-
term equity market price volatility.) The idea of the pension fund as a vehicle
whose primary purpose was to ensure the security of members accrued ben-
efits was recognised by some actuaries, but sat largely in the background of
the development of actuarial thinking on pension funds before the 1970s.
However, this perspective would occasionally be considered and could lead to
substantially different conclusions. Plymen noted in the discussion of Days
1959 paper:
With any but the strongest employers the contingency existed in a slump that
the employer might be bankrupt or there might perhaps be severe staff redun-
dancy. Under those conditions, the employer would be unable to subsidize a
pension fund prejudiced by depreciation of its ordinary shares. He imagined
that that would be regarded as by far the greater threat, so that fixed interest
securities would tend to be preferred. He maintained that the strength and secu-
rity of the employer should be the major factor in determining the equity
percentage.37
35
Hemsted, in Discussion, Heywood and Lander (1961), p.365.
36
Day and McKelvey (1963), p.107.
37
Plymen, in Discussion, Day (1959), p.152.
Plymen was arguing that the nature of the asset risk that mattered to pen-
sion funds was complex: asset price falls did not matter so much if the sponsor
was healthy; what really mattered was how the pension fund assets behaved
in the conditions that were associated with sponsor bankruptcythats when
the assets were really needed. Such thinking would play an increasingly influ-
ential role in actuarial thinking on pension funds, their valuation and their
asset strategy over the following decades. With the benefit of hindsight, it is
perhaps striking that it was never more than a footnote in the thinking of the
1950s and 1960s, especially when sophisticated thinking by leading actuaries
such as Plymen was in circulation.
Equity allocations continued to grow in the UK defined benefit pension
fund sector, both as a result of the strong growth of pension funds equity
holdings over the 1960s, and with the investment of new contributions. By
1975, two thirds of UK final salary pension funds were invested in equities
and property.38 The extraordinary volatility of UK equities in 19731975,
however, provided actuaries with a reminder that equity returns could be
driven by factors other than inflation. In his faculty presidential address of
1977,39 R.E.Macdonald noted:
The market collapse of 1974 finally provided convincing evidence that any con-
nection between the values of shares and the rate of inflation was mainly
fortuitous.40
The 19731975 period certainly opened eyes to the degree of volatility that
was possible in real equity price levels (despite the exceptionally high infla-
tion rate, UK equities fell by over 70 % in 19731974; they then more than
doubled in 1975). But McKelveys and Days papers had been careful not to
suggest that real equity returns were not volatile: rather, they argued it did not
matter if they were. Their argument was that real dividend growth could be
expected to be stable, and that was what mattered for pension funds (open
ones, at least). But the experience of the early 1970s also proved that this faith
in the stability of the real value of dividends was misplaced: UK dividend
pay-outs fell by around 45 % in real terms between 1970 and 1974 and did
not fully recover their real 1970 value until the late 1980s.41
38
Holbrook (1977), p.58.
39
Macdonald (1977).
40
Macdonald (1977), p.9.
41
Dimson, Marsh and Staunton (2002), p.152.
So the UK experience of the early 1970s showed that the much-vaunted

inflation-protection of equity dividend income was a myth. Nonetheless,
actuarial wisdom remained unperturbed. In his Journal paper42 of 1977 on
pension fund investment, Holbrook was sanguine on this episode of equity
market volatility, noting:
The catastrophic fall in UK equity markets in 1973 and 1974 reflected a serious
crisis of confidence, which has not been wholly restored by the subsequent rise.
With the benefit of hindsight it may now be said that the main importance, for
a long-term investor, of the exceptionally low market level in 1974 was the
opportunity it presented to buy shares cheaply.43
Thus, equities and property became the dominant asset class choices of
defined benefit pension funds. This was doubtless in part influenced by the
wider cult of the equity that developed from the second quarter of the twen-
tieth century onwards, and, from a specific actuarial perspective, because they
were the only viable real asset class choice of the era with which to back
undoubtedly real liabilities. The question of how much protection from
short-term inflation shocks or long-term inflationary eras can be provided
by equities (relative to bonds) is a question that has been much contested by
actuaries and economists for decades. Whilst it is intuitive that the real char-
acteristics of an equity claim implies some form of inflation protection, the
question remains of how material a driver of returns (and dividends) inflation
is relative to the many other factors that drive the behaviour of equities (over
the short and long term).
A quantitative framework for analysing liability-orientated investment
strategy was proposed by A.J.Wise in his 1985 paper The Matching of Assets
to Liabilities.44 Wise defined a matching portfolio in a very particular way,
which encompassed a much wider potential set of portfolios than the abso-
lute matching of cashflows that was first defined by Haynes and Kirton back
in 1952: Wises matching portfolio was defined as the asset portfolio that
minimised the volatility of the ultimate surplus of the book of business, i.e.
the residual (positive or negative) asset value after the final liability cashflow
had been paid. Importantly, he added the stipulation that the matching port-
folio could not be re-balanced over the projection horizon.
42
Holbrook (1977).
43
Holbrook (1977), p.17.
44
Wise (1985).
Wises matching portfolio could differ from a cashflow matching portfolio

for a number of reasons: the universe of investable assets may not be granular
enough to exactly cashflow match (i.e. there may not be a bond with a matu-
rity date exactly coinciding with the liability cashflow); the liability cashflow
could be exposed to factors such as salary inflation that may not be directly
obtained in available assets; the liability cashflows may have non-linear fea-
tures that are not present in available assets (and that could not be produced
by a dynamic rebalancing strategy for the asset as such strategies had been
outlawed by Wises matching portfolio definition).
Wises stipulation that the matching portfolio must be a static one came
more than a decade after Black-Scholes-Merton had shown how fundamental
the idea of dynamic rebalancing was to liability matching (which in Black-
Scholes-Merton was just a means to arbitrage-free pricing by replication).
The static assumption also ran counter to the actuarial idea of immunisation
introduced by Redington in 1952 which focused on what might be called
instantaneous matching and recognised the need to rebalance through time
to maintain the match. But Wise believed that there was something inher-
ently imprudent about including dynamic rebalancing assumptions in a risk
modelling exercise. In introducing his paper at the Staple Inn discussion, he
considered practices in the life field where modelling was comparatively more
developed than in pensions, and asked:
When assessing the solvency of a life office is it safe to assume that the office will
always be able to take advantage of investment conditions in the future so as to
keep itself in an immunised position? Would it not be safer to identify a mini-
mum risk matching portfolio for solvency purposes on the assumption of no
future investment activity except as required by new money?45
I.C.Lumsden, a senior life actuary, responded:
It does seem to me unrealistic to consider matching portfolios only on the

assumption of passive investment policies, and a pity in particular to ignore the
one dynamic strategy immunisation which has gained our acceptance in the
past.46
Wise went on to propose that the market value of the matching portfolio as
defined above for pensions liabilities could be considered as a form of liability
valuation. On the surface this seemed quite natural, and not dissimilar to the
45
Wise (1985), p.65.
46
Lumsden, in Discussion, Wise (1985), p.71.
economic idea of arbitrage-free pricing by replication. But Wises definition

of a matching portfolio was profoundly different from an economists notion
of a replicating portfolio. Wises matching portfolio could leave the matcher
significantly exposed to market risksso whilst the portfolio might generate
an expected surplus of zero, it may still have a significant economic cost. The
identification and cost of Wises matching portfolio could also be a function
of subjective assumptions about long-term asset risk premiaagain a feature
that ran contrary to the properties of market-based valuation techniques.
It could be said that much of the confusion that has historically arisen
between actuaries and financial economists has been due to a lack of clarity
between the ideas of expected funding costs (based on expected long-term
asset returns) and replication costs (based on the current market prices of
replicating assets). Wises matching portfolio definition and its use in liability
valuation could only add to this confusion. Wise compared pension fund
liability valuations produced by his matching portfolio concept with the more
straightforward market-consistent approach of discounting liability cashflows
with the observed market yield curves. He expressed the traditional actuarial
discomfort with such direct use of market prices:
This [market-consistent value] result might be considered realistic on the

grounds that current redemption yields on secure fixed interest stocks are the
best available guide to future economic conditions. Whether this is so is a some-
what philosophical point; it is certainly not generally accepted.47
For pension actuaries, uncoupling expected funding costs and economic

values appeared interminably challenging. But the distinction would only
become more relevant.
Funding forWhat? (19721997)

Actuarial thought on pension fund valuation prior to the 1970s was almost
exclusively focused on valuation with the objective of determining the long-
term contribution rate, expressed as a constant percentage of salary roll, that
would be required to fund liability cashflows as they ultimately fell due.
Similarly, thinking on investment strategy was centred on the objective of
generating long-term asset income that would be used together with these
contributions to fund those liability cashflows. Pension funds were managed
Wise (1985), p.55.

47
to provide a stable funding plan for the pension liabilities on a going con-
cern basis. The pension funds function as a source of security for employees
accrued benefits had, of course, been recognised by actuaries since the nine-
teenth century. But it did not figure prominently in the first 100 years of
actuarial thought on how to manage defined benefit pension funds.
Funding with the objective of providing security for current accrued bene-
fits started to command greater prominence in British actuarial thought in the
1970s. This was partly driven by an increased incidence of corporate mergers
and takeovers, and the tendency these had to lead to wind-ups of some of the
involved companies final salary pension funds (in which circumstance, the
sponsor was not obliged to make any further contributions to the fund). In
occasional high-profile cases the pension fund wind-up exposed the inade-
quacy of the asset fund to secure the pension benefits that members expected.
D.F.Gilley, whose work we already encountered above in his 1958 paper
with Funnell, published an important paper on this subject in 1972.48 He
discussed what occurred on the wind-up or dissolution of a defined benefit
pension fund. There were several aspects of the dissolution process that could
place considerable responsibility with the scheme actuary. The trust deed of
the pension scheme would, in theory, set out what the members benefits
would be in dissolution. But these rules would generally be quite vague.
Members might only be entitled to the benefits they would have obtained if
they had withdrawn from the fund on the date of the dissolution of the fund.
If any surplus or, more pertinently, deficit existed relative to the cost of secur-
ing those benefits from an insurance company with the pension funds assets,
the actuarys advice on how it should be equitably distributed amongst the
members would usually be decisive.
The cost of securing the accrued dissolution benefits could diverge signifi-
cantly from the long-term actuarial valuations undertaken to set contribution
rates. The cost of the dissolution benefits was driven by the then-prevailing
market gilt yield (which determined life companies pricing terms for the
deferred and immediate annuities that would need to be bought to secure the
benefits). As we have seen above, this market-based rate could vary substan-
tially from the valuation interest rate used in the pension valuation (which was
based on the actuarys estimate of the expected long-term return on the pen-
sion fund assets and hence could include the actuarys estimates of the equity
risk premium and the long-term expected interest rate). Although market gilt
yields would typically be lower than a pension funds valuation interest rate,
the cost of the dissolution benefits would not necessarily be greater than the
48
Gilley (1972).
long-term ongoing valuationthe withdrawal benefit that would typically be

viewed as the minimum accrued benefit would break the link with real salary
growth, thereby significantly reducing active members liability values (the
degree to which this impacted on the valuation would therefore be a function
of the maturity of the pension fund). Overall, contribution rates and asset
strategies were not set to target dissolution benefit funding, and the funding
position on this basis could therefore vary markedly across pension funds.
Gilley called for the dissolution benefit funding level to be calculated as
part of the regular actuarial management of defined benefit pension funds.
Moreover, he proposed this funding result should be published and com-
municated to members. This was viewed as a fairly radical notion at the time.
Indeed, the entire idea of taking into account the dissolution benefit funding
level when setting contribution rates was provocative to traditional actuarial
thinking. However, the economic turmoil of mid-1970s Britain prompted
actuarial pensions thinking to consider some even more radical questions:
should private sector defined benefit pension liabilities be funded at all?
This far-reaching question was raised in no less august and authoritative a set-
ting than a Faculty of Actuaries presidential addressthat of R.E.Macdonald
in 1977.49 The economic and political environment for defined benefit pen-
sion funds was extremely challenging in the early to mid-1970s. Annual price
inflation reached a peak in excess of 24 % in 1974/1975. This high-inflation
environment naturally fed through to salaries and drove pension liability val-
ues higher. Real investment returns over the first half of the 1970s were nega-
tive. There was political pressure to inflation-protect pensions in-payment,
and to maintain (in real terms) the accrued benefits of withdrawing members.
Taken together, and with the possibility that the high-inflation/negative real
return environment could be a long-term feature of the economy, it was per-
haps unsurprising that radical solutions to defined benefit pension scheme
funding were being put on the actuarial table.
In this context, Macdonald noted the outspoken presence of those who
regard funding [of defined benefit pension liabilities] as an outdated shib-
boleth of the actuarial profession.50 Macdonald expressed some sympa-
thy with this position and, in particular, questioned the reasonableness of
advance funding pension liabilities in a high-inflation, negative-real-interest
rate e conomic environment. His point was two-fold: in such economic con-
ditions, the funding of inflation-linked liabilities would be very demanding
for sponsors, especially if pensioner inflation increases were included, and this
Macdonald (1977).
49

50
would likely be a time when employers could least afford to contribute the
additional cash; and secondly, he found it unconscionable to invest in assets
that offered a negative real interest rate. In his own words, the acquisition of
new assets during such periods [of negative real interest rates] is really impos-
sible to justify.51 Macdonald suggested that, in such circumstances, pension
contributions should be limited to those required to fund immediate benefit
outgo, and no contributions for the funding of the accrual of longer-term
liabilities should be sought.
Macdonalds ideas were quite controversial, especially coming from a leader
of the profession. After all, he was proposing that contribution rates ought
to be reduced at the very time when an economic view of the pension fund
would suggest contributions were required more than everboth because
the funding level would be at its weakest and the credit risk of the sponsor
would be at its greatest. An economist working from the premise that the
primary rationale for advance funding was to ensure a high level of pension
fund member security might be forgiven for thinking the actuarial dark arts
were becoming too clever by half.
But President Macdonald was not the only actuary expressing reservations
about advance funding of pension liabilities during this period. The same year
saw the publication of a paper by J.R.Trowbridge in the Journal of the Institute
of Actuaries which covered similar ground.52 Trowbridge looked overseas for
examples of how unfunded private sector defined benefit pension provision
could function. He found that the French had been running unfunded occu-
pational pension schemes with some success for several decades. Under this
system, which was translated as assessmentism, the employee security concern
of a non-funded approach was mitigated by having industry-level pension
schemes, potentially with hundreds of different employers as sponsors collec-
tively contributing (together with members) whatever contribution amount
was required to meet each years pension outgo as it arose. Thus the single-
name credit exposure that was usually borne by British pension schemes was
diversified (though an industry-level credit exposure would remain). These
ideas naturally met with some suspicion amongst the UK actuarial profession.
There was a concern that much of the French system was based on a voluntary
cross-generational transfer that relied on notions of socialist solidarity that did
not naturally fit with the British culture of self-sufficiency. Above all, there
was a hope and faith that the economic norm of positive real returns in excess
51
52
Trowbridge (1977).
of salary growth would return to Britains pension funds, bringing with it a

return of financial viability for traditional British pension schemes.
The actuarial debate of this fundamental aspect of defined benefit pension
provisionto fund or not to fundcontinued in the following years. Boden
and Kingston followed their Presidents lead and further discussed the relative
merits of advance funding and pay-as-you-go financing of final salary schemes
in a paper published in the Transactions of the Faculty of Actuaries in 1979.53
They agreed with Macdonalds central point that in a negative real interest rate
environment the case for pay-as-you-go was greatly strengthened. However,
Boden and Kingston went on to argue that the 19741975 inflation experi-
ence and the poor real equity return performance of 19681978 were aber-
rations that did not alter their forward-looking expectation of long-term
positive real returns on pension fund assets. They also reviewed the French
assessmentism approach discussed by Trowbridge and concluded: Once we are
reasonably happy that it is possible to accumulate interest on capital at least as
fast as inflation, there seems no compelling argument for going through the
trauma of adopting the French system.54
Boden and Kingston were even sufficiently sanguine about future real inter-
est rates to argue that explicit allowance for post-retirement inflation increases
could and should be made in pension fund valuations and contributions.
This position was supported by influential actuaries such as A.D.Wilkie, who
made the same argument in another Staple Inn discussion a couple of years
later:
I see no reason to assume a long-term real rate of return at present [1982] mar-
ket levels of less than 4% in excess of prices. At that rate of interest, how many
pension funds are in surplus, and how many cannot afford index-linked
pensions?55
The degree of uncertainty and variation in long-term economic expecta-

tions amongst senior actuaries over this period of 19771982 is striking: are
long-term expectations for real asset returns negative or 4 %? There was little
doubt that, whether using advance funding or pay-as-you-go, the economic
cost of defined benefit pension liabilities would be unsustainably high in a
permanent negative real interest rate environment. A 4 % real interest rate, on
the other hand, would make generous inflation-linked increases beyond the
53
Boden and Kingston (1979).
54
Boden and Kingston (1979), p.412.
55
Wilkie, in Discussion, Colbran (1982)), p.410.
level of guaranteed benefits quite affordable. This highlighted the fundamen-

tal challenge of setting a funding objective based on subjective expectations
for the behaviour of economic variables over exceptionally long time horizons.
But the optimists prevailed. British actuarial orthodoxy on the advance
funding of final salary pension funds survived the economic turmoil of 1970s
stagflation and its consequent financial market volatility. Moving from the late
1970s to the 1980s, actuarial debate shifted from whether or not to advance
fund pension liabilities and on to funding methods and what the objectives of
funding should beback to the debate that Gilley started in 1972 which had
lain largely dormant for a decade. Rather like the contemporaneous experi-
ences of their life office brethren, influences and demands from outside the
profession were the main catalysts for pension actuaries to revise and evolve
their traditional approaches. In particular, and again with parallels with the
life office experience, the accounting profession and government regulators
had a telling influence (the Accounting Standards Committee was working
on an exposure draft on treatment of pensions in company accounts; the
Occupational Pensions Board was considering solvency disclosure and infla-
tion protection for early leavers and pensions in payment).
The accounting profession was increasingly active in setting new standards
for the reporting of the pension expenses in financial statements. Historically,
the pension expense may have been accounted for simply as the amount the
sponsor paid in contributions over the accounting period. As contribution
payments became more volatile, accountants became more focused on a mea-
sure of the cost of pension benefit accrued over the accounting period. In
the UK, actuaries preference was generally to keep funding and accounting
expense methods as similar as possible. There was therefore a general prefer-
ence to base this measure of cost on the stable long-term percentage of salary
roll that the actuary assessed would be required to meet the current mem-
bers future benefits net of the existing asset fund (the aggregate method).
However, accountants were increasingly aware that this contribution rate did
not necessarily equate closely with the cost of the liabilities that accrued over
an annual reporting period. In particular, the cost accrued might be expected
to increase later in the life of a pension fund as its membership aged. So
methods such as the projected unit credit method, which focused on the con-
tribution specifically required to meet the increase in the pension liability that
resulted from a years further accrual, were preferred by accountants.
These unit credit methods were in increasingly common actuarial use in
the 1970s but were initially somewhat unpopular with the British actuarial
profession as it challenged the traditional view that the assessed contribu-
tion rate should be a stable long-term percentage of salary roll rather than a
figure that would be expected to increase as the scheme matured. Colbrans
1982 paper reviewed this situation and called for more professional guidance
to be provided by the Institute of Actuaries on pension funding methods.56
He suggested a list of acceptable methods should be created with appropriate
terminology and definitions. This suggestion was acted upon and a Report
on Terminology of Pension Funding Methods was jointly published by the
Institute and Faculty of Actuaries in 1984.57
Colbran also argued in favour of a market value approach to asset valuation
rather than the discounted income approaches that dominated actuarial pen-
sions practice at this time. The Staple Inn discussion of the paper reopened
the pension asset valuation debate that had lain largely dormant since Day
and McKelveys 1963 paper. In the discussion, the established actuarial prefer-
ence for discounted income values over market values was clear. But this view
was not unanimous, and there was a wide recognition that the first priority of
the valuation method should be to ensure consistency between asset and lia-
bility values. This reopened the door to the idea of market valuation of assets
and market-consistent liability valuation, as espoused as long ago as 1921 by
S.G.Warner. In opening the discussion of the Colbran paper, Paul Thornton
(who would go on to become Institute President in 1998) stated:
It is possible to arrive at a set of assumptions which, when used to discount the

future interest and capital proceeds from the investments, would reproduce
market values. There is no reason why a valuation should not be made incorpo-
rating assets at market value if the liabilities are valued on consistent assump-
tions derived in this way.58
However, some senior pension actuaries in the discussion did indeed find
reasons why this approach to consistent asset-liability valuation should not be
adopted. D.E.Fellows noted:
To rely wholly or largely on market values could lead to much volatility in the
rates of interest used from valuation to valuation.59
And C.W.F.Low similarly stated:
This [market-consistent liability valuation] would mean constantly changing

valuation bases to reflect variations in the initial rate of interest obtainable
56
Colbran (1982)).
57
Turner etal. (1984).
58
Thornton, in Discussion, Colbran (1982), p.387.
59
Fellows, in Discussion, Colbran (1982), p.390.
this approach is dangerous, as it would concentrate employers and trustees

minds on short-term investment performance which is not appropriate to the
real investment needs of pension funds.60
1970s stagflation faded into distant memory as final salary pension funds
returned to financial health in some style in the first half of the 1980s. UK
price inflation appeared to have been tamed: in the mid-1980s, the UK annual
inflation rate moved within the range of 35 %. Equities performed excep-
tionally stronglythe FTSE 100 increased by more than 125 % between the
start of 1984 and the October 1987 crash. Real dividend growth was around
5 % perannum. Meanwhile, large-scale industrial redundancies further con-
tributed to pension fund surpluses (redundancies shifted active members
benefits into deferred benefits that were no longer linked to real salary growth
and hence reduced their value).
This turn of events may have afforded actuaries some latitude to think
more broadly about funding objectives and targets. In any event, the funda-
mental objective of funding became a hot topic again. D.J.D.McLeish and
C.M.Stewart were particularly instrumental in reopening the arguments pre-
sented by Gilley back in 1972 for focusing funding on ensuring that the cost
of discontinuance benefits could be adequately covered by the pension fund
at the time of the valuation. McLeish restarted this debate with a Faculty
paper in 1983 which emphasised member security as the paramount rationale
for advance funding of pension liabilities.61 He expressly rejected the Faculty
Presidents suggestion that funding should be deliberately suspended in times
of negative real interest rates:
In private occupational schemes I believe that funding is necessary to provide

security. The higher the yield on assets the better but there is no particular sig-
nificance in the point at which the difference [between investment yield and
salary inflation] changes from positive to negative.62
McLeishs funding-for-security arguments were further marshalled in a

paper63 co-written with C.M.Stewart in 1987 where they argued that:
The prime purpose of funding an occupational pension scheme must be to

secure the accrued benefits, whatever they might be, in the event of the employer
60
Low, in Discussion, Colbran (1982), p.406.
61
McLeish (1983).
62
McLeish (1983), p.275.
63
McLeish and Stewart (1987).
being unable or unwilling to continue to pay at some time in the future. To that
end, the contributions would have to be sufficient both to pay the benefits as
they fell due for as long as the scheme continued, and also to establish and
maintain a fund which would be sufficient to secure the accrued benefits in the
event of contributions ceasing and the scheme being discontinued, whenever
that might occur.64
A practical complexity that arose in consideration of this type of funding

target was the ambiguity that might exist in the pension schemes trust deed
and rules as to what members were entitled to on wind-up of the scheme.
However, by 1987, statutory minimum benefits for early leavers existed that
linked accrued benefits to the lower of future national average earnings infla-
tion and 5 %. This provided an unambiguous minimum benchmark for
wind-up benefits.
The approach of funding for discontinuance benefits was still generally
unpopular with British actuaries at this time. Colbrans 1982 paper that dis-
cussed the need for greater professional guidance on funding methods had
even called for this particular funding method to be outlawed by the profes-
sion on the grounds that it could produce a substantially lower contribution
rate than other methods. But McLeish and Stewart argued that this would
only be the case for immature schemes. They rejected the relevance of the
level contribution rate that would ultimately fund all liability cashflows and
instead focused on ensuring that accumulated assets were always sufficient to
meet the accrued liabilities. They further argued that the level contribution
rate approach would tend to result in the accumulation of unnecessary sur-
pluses relative to the accrued liabilities.
The responses to the paper in its Staple Inn discussion were generally nega-
tive. Traditional actuaries were concerned that the specific targeting of the
security of the accrued benefits would actually result in less security by slow-
ing the pace of funding. But some recognition of McLeish and Stewarts argu-
ments was won. The influential Paul Thornton agreed that the funding of
accrued benefits should be part of the funding objective:
I too believe that it is necessary for the funding policy to be such that the assets
are normally sufficient to cover in full the winding-up priority levels. From the
point of view of security for the members benefits, this defines a minimum
funding level.65
McLeish and Stewart (1987), p.155.

64
McLeish and Stewart (1987), p.221.

65
Like other developments in actuarial thought that emerged in a climate of

traditionalist scepticism, once the genie was out of the bottle, it would never
go back in. The next notable contribution to British actuarial thinking on
the funding of final salary pension schemes arrived in a paper, A Realistic
Approach to Pension Funding, by Paul Thornton and A.F.Wilson in 1992.66
This paper was ostensibly traditionalist in outlook, continuing to advocate a
discounted income approach to asset valuation and funding methods that were
focused on producing long-term contribution rates rather than funding cur-
rent discontinuance benefits. But, rather like McLeish and Stewart, they argued
that typical funding calculations could lead to over-funding. They noted the
embarrassingly large surpluses that prevailed in many pension funds at the
time.67 Their central argument was that these surpluses did not arise because
of any inherent limitation in the traditional pension funding methods, but
because the parameterisation of these methods generally included numerous
prudent margins that were cumulatively unnecessary and sometimes acciden-
tal. They argued that the typical valuation assumption of a real return of 44.5
% was lower than the expected real return on equities that most pension fund
assets were invested in. They believed a 56 % expected real return on equi-
ties was reasonable based on historical analysis and forward-looking economic
arguments. They suggested that salary inflation tended to be lower than the
national average for the age group of over-40s, and it was this group that was
most important to the valuation of active members benefits. They estimated
that this over-stated salary inflation by around 0.5 %. According to Thornton
and Wilson, these and other margins were generating unintended surpluses.
In a similar vein to Thorntons comments in the McLeish and Stewart dis-
cussion, they regarded the discontinuance position as an important reference
point, though not necessarily the central driver of the contribution rate. They
even went a step further, arguing that a margin of 20 % over the discontinu-
ance cost should be maintained in light of the speed with which surpluses and
deficits can build up.68 To paraphrase, they were arguing that if the purpose
of the assets was to meet a market value of liabilities in the near future, it was
not sufficient to have assets that met that value of liabilities todayan addi-
tional buffer would be required to the extent that the asset and liability values
could diverge between now and the possible dissolution event. This logic is
essentially that of the value-at-risk concept that emerged from the banking
sector and, in the twenty-first century, would become the basis for insurance
solvency regulation.
66
Thornton and Wilson (1992).
67
Thornton and Wilson (1992), p.263.
68
Thornton and Wilson (1992), p.264.
A number of speakers in the Staple Inn discussion of the paper noted that
the cost of transferring discontinuance benefits to insurance companies had
increased considerably in recent years (partly due to improvements in early
leaver rights and partly due to lower interest rates). For the first time, the
cost of the non-profit immediate and deferred annuities that would meet the
discontinuance benefits might typically exceed the liability values assessed by
the traditional ongoing funding assessments, and indeed the market value of
assets in the pension fund. Between the publication of the paper and its dis-
cussion at Staple Inn, the Robert Maxwell affair occurred, placing discontinu-
ance security more squarely in the public eye than ever before. Thus, actuaries
embarrassing surpluses had to be simultaneously reconciled with increasingly
transparent discontinuance shortfalls. This was a significant challenge to the
actuarial professions credibility and communication skills. It was a challenge
they struggled to meet. Ultimately, it was taken partly out of the professions
hands when the government legislated the introduction of the Minimum
Funding Requirement (MFR) for final salary pension schemes in 1995.
The MFR was a controversial and contentious requirement from its incep-
tion. It was a result of a committee, the Goode Committee, which was
established by the government in the wake of the Robert Maxwell scandal.
However, the recession of the early 1990s was another driving factor for the
interest in legislation to protect pension fund members accrued benefits. A
number of pension scheme wind-ups occurred during this recession where
members did not receive their expected benefits, and this experience further
contributed to the sense that the security of accrued benefits was not being
adequately protected by the prevalent actuarial funding methods. So the MFR
was first intended as a robust measure of the adequacy of the value of assets in
the pension fund to meet the market cost of accrued benefits. However, once
the financial implications of using the insurance annuity basis in the pen-
sion valuation were realised, a political compromise was made and the MFR
valuation basis was fudged: the expected return on equities would be used
as the discount rate for accrued benefits (except pensions in payment), even
though this would undoubtedly understate the cost of securing the accrued
benefits in the event of dissolution of the scheme.69 Consequently, being 100
% funded on the MFR basis did not mean the pension fund necessarily had
close to sufficient assets to meet the cost of the accrued benefits in the event
of dissolution of the pension fund. The result was unsatisfactory for all con-
cerned: actuaries lost more of their professional independence and ability to
69
See Greenwood and Keogh (1997) for a detailed description and discussion of the actuarial bases of the
MFR.
self-regulate; government had created a costly regulatory framework that did

not meet its intended purpose; members may have been led to believe their
benefits were secure when in fact they were not. As with their life office breth-
ren, the 1990s was proving to be a tough decade for British pension actuaries.
odern Thought onPension Scheme Valuation

M
andInvesting (19901997)
The clash of new and traditional thinking on the funding, valuation and
investment of defined benefit pension funds grew in intensity over the 1990s,
representing one of the more turbulent episodes in British actuarial thought.
The 1990 Journal paper, Actuaries, Pension Funds and Investment, by
T.G.Arthur and P.A.Randall can be viewed as the start of this period.70 The
paper was not as radical or provocative as those that would follow over the
next few years, but it unmistakably advocated a further move away from tradi-
tional actuarial valuation approaches to those that attached greater legitimacy
to market values and fully embraced their use in pension fund valuation.
The central idea set out in the paper was that changes in the market values
of assets that did not match liabilities should be fully reflected in the change
in the actuarial assessment of surplus in the pension fund. They argued that
to do otherwise was implicitly taking an investment view that the market was
wrong and that this is not our [actuaries] job when constructing [pension
fund] valuation models.71 They further elaborated there is no actuarial reason
to suppose that any relative changes in market values that have taken place in
the past, whether recent or otherwise, will be reversed in the future.72
This perspective implied a significant narrowing of the pension actuarys
professional input. It was a departure from the traditional actuarial tendency
to attempt to look through irrational short-term market volatility to the
underlying intrinsic worth of the pension fund investments on a going con-
cern basis. For the previous 40 years, pension actuaries had felt quite com-
fortable placing values on pension fund investments that could differ quite
substantially from their market values. This was partly justified on the basis
of consistency with an off-market and stable liability basis, but underlying
these choices was a philosophical view that short-term market value move-
ments were of no consequence to an open pension funds long-term funding:
70
Arthur and Randall (1990).
71
Arthur and Randall (1990), p.5.
72
Arthur and Randall (1990), p.6.
as a going concern, it would never have to sell its assets, and asset income was
more stable than was implied by market value movements. More widely, actu-
arial asset models such as Wilkies implied that future risky asset returns were
significantly influenced by past returns (mean-reversion in equity returns,
for example). This gave a technical manifestation to the actuarial convention
that market values varied more than could be justified by changes in assets
expected future cashflows, and a justification for removing this variation from
actuarial valuations. Arthur and Randall argued that all of this had no place
in actuarial methods for pension fund valuation.
A natural corollary of their logic was that liabilities should be valued at the
market value of those assets that matched them. In the terminology of finan-
cial economics, this was the fundamental concept of the replicating portfolio.
Irrespective of what pension actuaries chose to call it, it was essentially what
S.G.Warner had called for in 1921. Whilst the Thornton and Wilson 1992
paper discussed above still advocated the use of the traditional discounted
income approach to asset valuation using parameters that need not be market-
based, the actuarial proponents of the market value approach were undoubt-
edly growing in number and influence.
The final implication of the Arthur and Randall perspective was that pen-
sion fund investment strategy should be set at a scheme-specific level that
reflected the specific liability cashflow profile of the pension fund. This was a
fairly novel idea in 1990 that only became well established in the 2000s under
the jargon of liability-driven investing. Their vision of the actuarial role in
pension funds removed subjective investment views from actuarial valuations
and made the valuation independent of the actual asset strategy of the pension
fund. But it suggested a greater role for actuaries in pension funds strategic
asset allocationit made actuarial estimates of liabilities central to pension
fund investment strategy.
In the Staple Inn discussion there were inevitably voices who were uncom-
fortable with Arthur and Randalls modern perspective and its implied dimi-
nution of the role of actuarial judgement in pension fund valuation. There
remained amongst some senior actuaries a strong view that the pension fund
valuation should be an estimate of the amount of assets they expected to be
required to meet the liability cashflows given the actual asset strategy of the
fund, rather than be based on the market cost of a theoretical risk-free match-
ing strategy. In Dr Sissons speech to open the discussion, he argued:
The valuation ought increasingly to reflect the realities of life. Included in this
would be a recognition of the investments expected to be held by the trustees in
the future and the assumed returns thereon.73
This encapsulated the fundamental difference of opinion as to the pur-

pose of the pension fund valuation: was it to assess the current market cost
of matching the liabilities (which should reflect the cost of transferring the
liabilities to a third-party immediately), or to make a best estimate of the aver-
age amount of assets required to meet the ultimate payment of the liability
cashflows if a particular set of investment risks (and corresponding expected
rewards) were borne by the fund until its extinction? Both of these quantities
were, to borrow Dr Sissons phrase, realities of life. Neither was wrong, they
were merely answers to different questions. But this clarity was generally miss-
ing from the debate.
Arthur and Randalls call for a market-based valuation approach to pension
fund assets and liabilities received further support in a paper by Dyson and
Exley that was published in the British Actuarial Journal in 1995.74 Dyson
and Exley emphasised that, by this time, pension fund valuations had a wider
set of applications than their traditional use in the assessment of the long-
term contribution rate required to fund expected liability outgo. And whilst
the assessment of the contribution rate expected to fund long-term expected
liability outgo necessarily required assumptions about asset risk premia, a
number of other valuation purposes had emerged that did not. In particular,
the assessment of whether the fund had sufficient assets to fund the transfer
of the accrued liabilities to a third-partythe discontinuance solvency assess-
mentwas fundamentally an assessment of a market cost. It did not require
subjective actuarial estimates of long-term asset risk premia. Such applications
were naturally more suited to market-based approaches to asset and liability
value. The traditional approach of using a discounted income asset valuation
coupled with a valuation interest rate based on a long-term expectation for
investment returns that was developed in the early 1960s did not answer the
discontinuance solvency question.
Dyson and Exley also expressed misgivings about the use of the discounted
income approach even in the context of long-term contribution rate assess-
ment. In particular, they highlighted that the equity valuation produced by
the discounted income approach was extremely sensitive to the actuarys
long-term dividend growth assumption, and that this was a parameter around
73
Sisson, in Discussion, Arthur and Randall (1990), p.28.
74
Dyson and Exley (1995).
which there was significant uncertaintythe level of confidence in any esti-

mate of m [the long-term rate of real dividend growth] is likely to be very
low indeed.75 But there was no escaping thisthe estimation of the expected
amount of contributions required to meet ultimate liability outgo in the pres-
ence of a risky asset strategy was an inherently subjective, difficult-to-estimate
judgement.
Dyson and Exley also emphasised that the existence of index-linked gilts
(from the early 1980s) permitted a market-consistent approach to the valu-
ation of inflation-linked liabilities that was not available to actuaries in the
1950s and 1960s era when discounted income methods were pioneered. They
argued that pension liabilities should be valued using the nominal and real
term structures of interest rates implied by gilt and index-linked gilt prices
at the given valuation date. Predictably, these proposals led to discomfort for
some actuaries who saw their valuation expertise being reduced to mechani-
cally plugging in the markets latest bond prices. In opening the Staple Inn
discussion of the meeting, G.J.Clark suggested:
We must not allow ourselves to become so mesmerised by market information

that we exclude one very valuable source of information the professional
judgement of the actuary.76
Interestingly, whilst market-based approaches to valuation had increasingly

been advocated in actuarial thinking in recent years, their proposals were still
viewed as quite drastic. R.C.Urwin commented:
It [the paper] has quite a radical conclusion which is iconoclastic to current

actuarial practice.77
Concerns were also raised as to the wisdom of relying on index-linked bond

prices when the volume of issuance was low relative to the scale of pension
fund assets. C.D.Daykin, the Institute President of the time, stated:
It has been indicated by several people that the real return on index-linked gilts
is not necessarily a very good indicator [of the market expectation of the real rate
of return] the low liquidity and the way in which they are influenced by a
75
Dyson and Exley (1995), p.489.
76
Clark, in Discussion, Dyson and Exley, p.543.
77
Urwin, in Discussion, Dyson and Exley, p.544.
particular sector of the investment market means they are of less value than they
might be in determining the market value of liabilities.78
But the biggest protest against the extensive use of market prices was a phil-
osophical one. Many actuaries retained the view that market prices were irra-
tionally volatile, and that actuarial judgement should be able to look through
this. S.J.Green rhetorically asked:
What new piece of information became known on 19th October 1987 which
was not known three days before? It must have been very significant to lead to
falls of more than 25% in a few days, and that is the answer to the authors point
where they say that the market provides rational prospective expectations.79
The influential Professor Wilkie struck a similar chord in defence of dis-

counted income valuation approaches:
Using this methodology [discounted income approach], the actuary is, indeed,
saying that the market has temporarily got it wrong, but that, in due course, it
will get it right, especially by the time the liabilities fall due.80
Wilkies view of the role and breadth of actuarial judgement in asset valua-
tion was the polar opposite of that advocated by Arthur and Randall five years
earlier. But, as noted above, these differences were perhaps not as irreconcil-
able as they appeared: different assumptions were required to answer different
questions. A rigorous discontinuance solvency assessment would be market-
based; an assessment of the amount of contributions required to, on average,
meet liability cashflows as they fell due would inevitably require someone
somewhere to make an assumption about the prospective long-term risk pre-
mia offered by the pension funds asset strategy. The two were not theoretically
inconsistent or mutually exclusive, they were merely distinct.
Throughout their paper, Dyson and Exley were careful to position their
arguments in as inoffensive a manner as possible to actuaries schooled in tra-
ditional methods. They were quite prepared to concede that the discounted
income methods were appropriate or even impressive at the time they were
conceived. Their position was that recent circumstances dictated it was time
for the actuarial profession to continue to improve and modernise its meth-
ods. An important paper by Exley, Mehta and Smith that followed in 1997
78
Daykin, in Discussion, Dyson and Exley, p.552.
79
Green, in Discussion, Dyson and Exley (1995), p.546.
80
Wilkie, in Discussion, Dyson and Exley (1995), p.549.
adopted a markedly different tenor.81 By contrast, its tone barely concealed a

relish in smashing all actuarial shibboleths in sight.
Like Dyson and Exley before them, Exley, Mehta and Smith again high-
lighted the broader range of contemporary applications of pension fund
valuations and how these applications were better served by market-based val-
uation approaches. This paper developed some further sophistication around
the market-consistent valuation method for pension liabilities, tackling com-
plexities such as the valuation of the complex non-linear inflation exposures
of deferred member liabilities that arose in the form of limited price index-
ation. The paper also presented some empirical evidence from UK economic
data to argue that national real salary growth was fairly stable. This supported
the argument that index-linked bonds could be regarded as a good hedge for
active liabilities (and hence helpfully avoiding the need to consider equities or
other risky assets as a hedging asset within liability matching and valuation).
The main contribution of the paper was, however, in the analysis of the
investment strategy of a defined benefit pension scheme from the perspective
of the corporate sponsor. Little had been written in actuarial journals on the
management of pension funds from the specific perspective of the sponsor.
Since the 1980s, however, consulting pension actuaries often formally advised
sponsors as well as pension fund trustees. Exley, Mehta and Smith pointed out
that from an economic perspective, the sponsor of a defined benefit pension
scheme should, in theory, be indifferent to the pension funds investment
decisions. The assets of the scheme could be considered as off-balance sheet
assets of the corporation and, from a shareholder value perspective, bonds
with a market value of 1 had the same economic worth as a portfolio of
equities that had a market value of 1. They argued that shareholder wealth
therefore could not be created or destroyed by merely switching from one to
the other, in much the same way that Modigliani-Miller showed that share-
holder wealth could not be created or destroyed by changing the corporate
financial structure of the firm.
However, this conclusion came with some caveats. Firstly, it only strictly
applied in the case of a pure defined benefit scheme. If, for example, the
members participated in discretionary benefits that would be increased in the
event of a pension scheme surplus arising but which could not be reduced
beyond a floor in the event of a deficit arising, then in economic terms the
shareholder was writing a call option to the member. The greater the volatility
of the pension funds assets relative to liabilities, the greater this option would
be worth. In such circumstances, the shareholders interest would be served
Exley, Mehta and Smith (1997).

81
by minimising this volatility, which implied matching the pension liabilities

as well as possible.
The second complication was that their argument ignored the pension
fund members exposure to sponsor defaultif the sponsor went bust when
there was a deficit, the shareholder would never make good on the pension
promises. In economic terms, this could be viewed as a put option that the
pension fund member had written to the shareholder. It was a complicated
option because its value was a function of the sponsor default probability and
the correlation between the default event and the size of the pension fund sur-
plus/deficit (as well as the volatility of the pension fund surplus). Nonetheless,
it was an option that would increase in value (to the shareholder) as the vola-
tility of the pension fund surplus increased. An increasingly mismatched asset-
liability position in the pension fund would therefore increase this options
value to the shareholder. These two effects were therefore impacted in oppo-
site directions by changes to the pension funds degree of asset-liability match-
ing. The specific circumstances of the pension fund and its sponsor would
determine which effect dominated at a particular point in time.
The Staple Inn discussion of the paper unsurprisingly included much
defence of the traditional actuarial approaches to the management of final
salary schemes. A.J.Wise argued that discretionary benefit features such as
early retirement pensions remained an important plank of scheme benefits,
and more important than suggested by the authors. The ability to reduce such
benefits in times of poor equity performance meant that for shareholders
the downside to equity investment is less onerous when that equity is in the
pension fund82 (though presumably the shareholders exposure to the upside
would be similarly dampened).
Thornton pointed out that the shareholder value analysis was not a priority
for trustees who were responsible for managing the pension fund assets. He
also could not accept the removal of the equity risk premium from the active
members liability discount rate, though he could not articulate a rational
reason why not:
I am suspicious of their arguments that pension fund trustees should not be

investing so heavily in equities, and that valuation bases should factor out any
excess return expected. This conflicts with economic reality!83
82
Wise, in Discussion, Exley, Mehta and Smith (1997), p.942.
83
Thornton, in Discussion, Exley, Mehta and Smith (1997), p.952.
As ever, economic reality meant different things to traditional pension

actuaries and the modernisers. The traditionalist view was that a realistic mea-
sure of liability cost meant the expected amount of (risky) assets required to
fund the cashflows as they fell due; to the moderniser, economic reality was
the current market cost of transferring the liabilities to a third party. This basic
difference in perspective got lost in a debate of increasing passion that could
verge on animosity. Nonetheless, it appeared that the use of market values for
asset valuation had now reached close to universal acceptance amongst the
actuarial profession. As L.P.Tomlinson noted in closing the discussion:
I was astounded by the uniform acceptance of the use of market values. Speaker
after speaker endorsed the principle of using market values for assets, and then
valuing the liabilities consistently. My actuarial background was in the 1970s
when no actuary would endorse market values, so we have moved a very long
way since that period.84
Final Thoughts onthePensions Story

In the final analysis, the actuarys traditional objective in pension fund
management of estimating the stable long-term contribution rate that was
expected to be required to fund ultra-long-term and economically complex
forms of liability cashflow under a given risky asset strategy was an extremely
challenging task. Their conclusions were highly sensitive to guestimates of
non-stationary economic variables such as price inflation, real salary growth
and long-term asset returns. Different economic environments came and
went, each time prompting fundamental reviews of actuarial practices and
assumptions. In times of particular economic uncertainty, this could become
publically embarrassing. We saw how the five-year period between 1977 and
1982 produced actuarial estimates of the long-term real asset return ranging
from less than zero to more than 4 % perannum. The implications of these
different assumptions for the health of pension funds and their ongoing fund-
ing needs was so great as arguably to be a discredit to actuarial expertise.
Given the scale of inevitable uncertainty in the above estimates, it is per-
haps surprising in retrospect that no coherent risk management framework or
philosophy emerged as an intrinsic part of the apparatus of actuarial theory
and practice in defined benefit pensions. By the late 1950s, actuarial thinking
Tomlinson, in Discussion, Exley, Mehta and Smith (1997), p.955.

84
in the life sector, propelled by the papers of actuaries such as Redington,

Hayes, Kirton, Anderson and Binns, was producing and applying new tech-
nical ideas about asset-liability risk managementmatching, immunisation,
portfolio insurance-style thinking in dynamic asset allocation and the stress
testing of risky asset values relative to guaranteed liabilities all received serious
consideration in this rich era of development of actuarial thought on finan-
cial risk management in the life sector. Pension actuarial thinking of that era
appeared oblivious to such ideas and unwilling to consider how they could be
applied in the particular context of defined benefit pension funds. It is worth
pausing to ask why this was the case. It is perhaps linked to the historical lack
of clarity and consensus on what advance funding of pension liabilities was for
in the first place. In particular, a lack of focus on member security as the over-
riding purpose of the pension fund and on sponsor credit risk as a potential
threat to that member security meant that pension actuaries perhaps did not
have the same clarity of purpose as their life brethren when it came to think-
ing about managing financial risk.
The actuarial focus on the estimation of the stable long-term contribution
rate arose in nineteenth-century actuarial practice. It was accountants and
regulators that drove actuaries to consider other relevant financial objectives
and metrics such as funding for security against near-term sponsor insolvency
and measuring the cost of liability accrual. The profession struggled to estab-
lish in its thinking how these different demands were distinct questions that
could be consistently reconciled within a wider technical actuarial framework.
Again with the benefit of hindsight, it now seems clear that the path the
actuarial profession took to ensuring consistency between asset and liability
valuationoff-market liability valuations with consistently off-market asset
valuations, first proposed by Puckridge in 1948 and enthusiastically refined
over the following 15 yearswas a choice that created unnecessary confusion
both inside the profession and in its dealings with others for half a century.
The treatment of equity investments permeated much of the above think-
ing, and was a particular source of complexity and, at times, controversy. Prior
to the 1960s, the valuation treatment of equities was relatively straightfor-
ward, if basic. Equities would be valued at the lower of market value and book
value, and the liability valuation paid no heed to the pension funds asset allo-
cation. The papers of Heywood and Lander, and Day and McKelvey in the
early 1960s strove to create an equity valuation framework that was consistent
with the established off-market liability valuation basis. Things were made
more complicated still when Heywood and Lander proposed that equity allo-
cations should permit the liability discount rate to be adjusted upwards from
the actuarial estimate of the long-term risk-free rate in anticipation of the
long-term excess returns of equities. This contributed to the actuarial percep-

tion of a free lunch in long-term equity investment that was ultimately only
dispelled in 1997 by Exley, Mehta and Smith.
The UK dividend experience of the 1970s showed that the original actu-
arial rationale for pension fund equity investmentstability in real growth in
dividendswas misplaced. But actuarial confidence in their ability to look
through short-term irrational market value volatility for the benefit of their
long-term clients was difficult to entirely dispel, despite the arguments of
Arthur and Randall and others in the 1990s. However, this actuarial per-
spective was not so far away from contemporary financial economics as the
actuarial modernisers would lead the profession to believe. The empirical pro-
gram of financial market research of the 1980s developed by leading academic
financial economists such as Jensen, Shiller and Roll could lend some serious
intellectual weight to the traditional actuarial disdain for short-term market
price volatility. Curiously, this point was entirely missed in the actuarial tradi-
tionalist/financial economist moderniser debate of the 1990s. That is perhaps
reflective of how this debate could generate more heat than light, and could at
times enflame personal passions at the expense of objectivity and intellectual
openness.
Rather like with-profits in the life sector, from the late 1990s defined ben-
efit pension funds would start on a path of terminal decline, a structure that
no longer met the needs of the modern workforce, and whose untenable eco-
nomic cost was becoming increasingly apparent to employers. By the twenty-
first century, British actuarial debate on how best to manage open defined
benefit pension funds had become increasingly moot.
7
British Actuarial Thought inGeneral
Insurance (18511994)
General insurancethat is, the insurance of risks other than life

contingenciespre-dates life assurance in its development as an established
business activity of critical importance to Britains economy. Marine insur-
ance first emerged as a significant commercial activity in early thirteenth-
century Italy in ports such as Genoa. The earliest historical records of British
marine insurance date from sixteenth-century London. Over the following
200 years, the City of London and its insurance institutions such as Lloyds of
London developed into a leading global centre of maritime insurance. These
institutions delivered dependable insurance to the cargos and sailing ships
whose perilous sea voyages drove Britains maritime trading economy and the
expansion of its Empire.
The actuarial role in the British general insurance industry has, however,
been most notable by its absence for most of this history. It was only in the
1970s that the actuarial professions engagement with the industry moved
significantly beyond the occasional passing observation. General insurance
first appeared on the syllabus of the Institute of Actuaries in 1978. Actuarial
professional guidance was not proposed in Britain until 1986. This contrasted
with experience elsewhere in the world, most notably in the USA.General
insurance actuaries even had their own actuarial professional organisation:
the Casualty Actuarial Society, which was established in 1914. This greater
actuarial role in US general insurance was prompted mainly by state regula-
tion of premium setting, which created a demand for objective analysis and
professional certification.

DOI10.1007/978-3-319-33183-6_7
In comparison with life assurance, most forms of general insurance typi-

cally have several fundamental features that, to a greater or lesser degree, com-
plicate and confound the use of the statistical methods pioneered by the likes
of Richard Price and William Morgan in mortality modelling and the valua-
tion of life contingencies. These include:
Defining the unit of exposure. In life assurance, the unit of exposure is

straightforward: person-years. It is unambiguous, intuitive and quite easily
observed and recorded. In general insurance, the unit of exposure can be
much less obvious: for example, in motor insurance, it could be vehicle-
years or vehicle-miles; for fire insurance, it could be building-years, or
building-floors-years, or something else. And if there is a theoretically
appropriate unit, it may not be easily recorded.
The above point is related to the heterogeneity of general insurance experi-
ence. There may be a very large number of risk factors that drive general
insurance experience. Identifying and observing these risk factors may be a
significant practical challenge. And once they have been identified, the
resultant homogenous sub-sets of data may be very small.
The claims experience incurred in a given insurance period may take many
years to discover. In some forms of general insurance, especially liability
business, it may only be many years after the end of the insurance period
that it transpires that a claim exists. So-called incurred but not reported
(IBNR) claims are inherently challenging to predict or estimate. For some
classes of business, the final settlement of the claim size may only occur a
further many years from when it is reported.
Most classes of general insurance business are inherently less stable or sta-
tionary than life assurance and its mortality rates. Social, judicial, techno-
logical, economic and commercial changes can drive changes in the
expected claims behaviour of general insurance that are typically faster and
more material than those in life contingencies. This can place substantial
limitations on the usefulness of historical experience data, even where it is
homogeneous, plentiful and complete.
As a result of these challenges, general insurance pricing and reserving have

historically often been determined on a case-by-case basis by underwriters
and other general insurance specialist practitioners using their expert judge-
ment and individual experience. The use of analytical methods and statistical
modelling based on empirical data has often been very limited. In this setting,
British actuaries have historically been uncertain of what role they can and
should play in the industry. We will see below, however, that British actuar-
7 British Actuarial Thought inGeneral Insurance (18511994) 277
ies have been, at least sporadically, considering how to apply their technical
and professional skills to general insurance pricing and reserving since the
mid-nineteenth century, if not before. This engagement grew in conviction
from the early 1970s, and by the 1990s actuaries in general insurance, hav-
ing finally established an uncontested industry role, were faced with similar
challenges and questions as their life assurance and pensions peers: did their
relative lack of technical training in cutting-edge stochastic modelling and the
new ideas in financial economics represent an existential threat to their estab-
lished role? Could they absorb some of these concepts into the mainstream of
actuarial thought and practice? First, however, we go back to the 1850s and
how that generation of innovative actuarial thinkers tackled the unique chal-
lenges of general insurance.
Fire Insurance (18511880)

In the UK, fire insurance first emerged as a significant commercial activity
in London in the decades following the Great Fire of 1666. A century later,
it was well-established across the country, and increasingly covered industrial
and agricultural fire risks as well as residential properties and retailers. By
the nineteenth century, some of the leading British fire insurance companies
had become successful multinationals, writing substantial business in North
America and the British colonies.
Eighteenth- and nineteenth-century fire insurance practices did not make
use of the statistical methods that were being contemporaneously developed
and successfully applied in life assurance. The broad challenges of applying
statistical and analytical techniques to general insurance were wholly appli-
cable to fire insurance business: the classification of historical claims experi-
ence into homogenous groups; the modelling of claim size as well as claim
frequency; making allowance for social effects such as fraud and its propensity
to vary with the prevailing economic climate. These challenges led to a deep
scepticism of the potential usefulness of statistical analysis of historical experi-
ence as an alternative to underwriting by case-by-case estimation.
However, some of the pioneering actuaries of the mid-nineteenth cen-
tury were attracted to the challenge. Samuel Brown, who would go on
to become the third President of the Institute, and who wrote one of
the first actuarial papers on investment, also wrote one of the first actu-
arial papers on general insurance.1 The mid-nineteenth-century challenges
Brown (1851).
1
in developing statistical analysis of fire insurance started with a lack of

available data on historical claims experience. The pooling of life assur-
ers mortality data was a well-established practice by the mid-nineteenth
century but fire insurers showed no inclination to follow that collaborative
practice. This was both due to commercial sensitivities and because the
insurers recognised that their data records were generally inadequate for
this purpose. Brown encouraged the fire offices to cooperate on the devel-
opment of statistical claims analysis by providing him with historical data
that he would pool but he found the course of his investigations on this
subject was stopped by the determination expressed by the fire offices to
refuse him all information whatever.2
Brown struck upon an alternative rich source of data. The fire offices had
formed the London Fire Engine Establishment in the 1830s to protect its
insured properties. Its superintendent, Mr Baddeley, compiled comprehen-
sive records of every fire to which its fire engines were called. These records
included the extent of damage to the property, the occupations of the prem-
ises, the cause of the fire, and what proportion of the value of the building was
insured. Browns 1851 paper presented a tabulation and analysis of Baddeleys
records for the years 1833 to 1849. Baddeleys records, however, could only
tell half the story. In order to deduce some useful insights for fire insurance
premium-setting, some measure of the exposed-to-risk was also required. For
this purpose, Brown turned to the Post Office Directory, which provided an
estimate of the total number of buildings that lay within the area covered by
London Fire Engine Establishment and, for commercial properties, the trades
undertaken at the premises.
Armed with these estimates of the number of buildings exposed and the
historical fire experience, Brown then calculated the one-year probabilities of
fire for commercial properties as a function of the occupation undertaken in
the property. He estimated, for example, that there was a 0.34 % probabil-
ity of fire in a given year at a grocers premises, whereas the corresponding
probability for Lucifer match-makers was a rather disconcerting 30 %. By
observing the distribution of the damage categories recorded by Baddeley,
and making some assumptions about the cost associated with each degree of
damage, he calculated the premium rates required for each class of occupa-
tion. His calculations implied an average annual premium rate of around 5 %
of sum insured with significant variation by use of premises.
Brown himself recognised how the inherent complexity of fire insurance
business inevitably placed significant limitations on the usefulness of his pre-
mium rate estimates:
Bailey in Discussion, Brown (1851), p.34.

2
The locality, the construction of the buildings, the distance from a fire engine
station, and, still more, the manner in which an adjoining building may be
injured by a fire breaking out in a neighbours house, would have to be
considered.3
Nonetheless, he had made resourceful and practical use of whatever lim-

ited data was available, and had handled it competently to provide some
illuminating insights and new quantitative reference points for insurance
premium-setting.
Thomas Miller wrote two short papers on the application of statistical tech-
niques to fire insurance that appeared in the Journal of the Institute of Actuaries
in 1857 and 1880. Unlike in life assurance, the claim size in the event of a
general insurance claim is usually not a fixed amount. Assumptions are there-
fore required about the behaviour of claim size as well as claim frequency. In
the case of fire insurance, the claim size would be based on the cost of the fire
damage incurred, up to the size of the policys sum insured. Furthermore, fire
insurance policies could insure either the full value of the property or a speci-
fied sum that was only a fraction of the value of the property.
Millers first paper considered some of the difficulties in interpretation of
claim experience that resulted from these features of fire insurance.4 He argued
that fire insurance claim data would be more useful if it reported the size of
loss as a proportion of the value of the property rather than the usual prac-
tice of recording it as a proportion of the sum insured. This would allow the
behaviour of claim size and hence insurance cost to be better understood as a
function of how the sum insured varied as a proportion of property value. He
argued that the profitability of the fire office and the adequacy of its premium
bases should be analysed not by considering the actual profits that had been
generated by the business, but by considering what the profits would have
been if the size of all its policies was standardised to the same sums insured.
This would remove the impact of the variation in policy size that occurred
due only to the incidental preferences of a particular group of policyholders
accidental circumstances that were not controlled or targeted by the insurer.
This could be considered as an early example of actuarial thinking on the
analysis of surplus.
Millers second paper considered what the appropriate unit of risk exposure
should be for fire insurance.5 He suggested it should be based on the number
3
Brown (1851), p.51.
4
Miller (1857).
5
Miller (1880).
of building floors insured rather than simply the number of buildings. He

then developed an algebraic framework for calculating fire probabilities based
on the two possible ways of a floor catching fire: by a fire originating on that
floor; and by a fire spreading to the floor from another floor. This allowed him
to produce probabilities of fire for a given floor of a building as a function of
the total number of floors in that building. From this he produced the result
that the probability of a given building floor catching fire increased with the
number of floors in the building. Miller produced some numerical examples
of his algebra, but made no attempt to calibrate his model to empirical data. It
is hard to believe that fire insurance practitioners would have recognised any
practical value in his approach.
These early actuarial contributions had little to no impact on British fire
insurance practice. In 1878, the historian and actuary Cornelius Walford
noted in a Journal paper,6 It will no doubt be thought remarkable that, after
two centuries of experience, the rates of fire insurance are still deduced by a
method entirely non-scientific But if there be no scientific base for fire
insurance premiums, it is not from the want of suggestions in this direction
[from actuaries].7
British actuarial interest in fire insurance and property insurance subse-
quently waned. No paper of note on fire or property insurance was published
in the Journal or Transactions between Millers 1880 paper and the end of the
Second World War.
Employers Liability Insurance (18821931)

A series of Acts of Parliament came into effect during the final two decades of
the nineteenth century that extended British workers rights to compensation
in the event of workplace injury or death. These laws meant a step-change
in the magnitude of legal liabilities that employers would be exposed to in
the event of workplace accidents. In nineteenth-century industrial Britain,
this could represent a substantial business risk, particularly in the accident-
prone manufacturing and extraction industries. Employers liability insur-
ance quickly emerged as a significant new line of general insurance business.
Its complexity, potentially long-term nature and exposure to mortality risks
naturally attracted strong actuarial intereston of the face of it, it certainly
appeared a more actuarial type of general insurance than fire insurance.
Walford (1878).
6
Walford (1878), p.4.

7
In Great Britain, as in several countries across Europe, the common law

accorded injured parties some general rights to compensation from those who
caused the injury. These rights were vague and limited and were not specifi-
cally related to injuries caused at work. The Employers Liability Act of 1880
gave explicit rights to workmen who were injured in the workplace due to
defective machinery or the negligence of superiors. The compensation was
limited in value to three years wages and was payable in the form of a lump
sum. William Whittall wrote the first British actuarial paper to tackle the spe-
cific challenges raised by insurance of this form of liability. It appeared in the
Journal of the Institute of Actuaries in 1882.8 Like much of early, and indeed
later, actuarial work in general insurance, many analytical compromises were
required as a result of the limited quality and quantity of historical claims
data. Whittall set out to analyse the rates of both fatal and non-fatal accidents
and how they varied by occupation, but a sparsity of data constrained him to
the statistical analysis of fatal accidents only.
In a similar spirit to Samuel Browns fire insurance analysis, Whittall had
to splice together data from disparate sources, neither of which arose directly
from insurance claims data. He used the 1871 census to obtain estimates of
the numbers employed in different occupations. For accident numbers, he
made use of the records of fatal accidents (in 1870, 1871 and 1872) that were
maintained at the governments General Records Office. These records had
the helpful feature of recording the occupation of the deceased (although they
did not distinguish between those accidents that occurred in the workplace
and those that did not). From these data sources Whittall compiled tables of
fatal accident rates for a comprehensive array of occupations. These results
showed, for example, that only 1.3 authors per 10,000 died annually from a
fatal accident, whereas horse breakers experienced 47.9 such deaths. Clearly
occupation was a material risk factor for fatal accident rates!
The actuarial audience at the Staple Inn discussion of the paper, however,
proffered a great deal of scepticism towards the reliability and relevance of
these statistics. The President, A.H.Bailey, cautioned against the usefulness of
the statistics for setting liability insurance premiums given that fatal accidents
made up a relatively small proportion of liability claims and were not neces-
sarily strongly correlated with the rate of non-fatal accidents. Whittall himself
acknowledged that the granularity of the occupations meant that the numbers
of deaths in some of the estimates resulted in unreliably small sample sizes. He
also acknowledged that the fast-changing nature of some occupations could
quickly render some historical rates obsolete. F.G.P.Neison, who had some
Whittall (1882).
8
experience of working with liability claims data from the coalmines and rail-
ways industries, highlighted that the records of the General Records Office
may not provide a reliable record of the occupation of the deceased, and did
not provide any satisfactory basis for the determination of the rates of fatal
accidents.9 As in fire insurance, the application of actuarial techniques con-
tinued to be frustrated by the lack of availability of relevant, reliable historical
claims experience data.
The Workmens Compensation Act of 1897 extended the rights of employ-
ees to compensation for workplace injuries. This act entitled the employees
dependents, if he had any, to three years wages in the event of a fatal acci-
dent, and in the event of a non-fatal accident that rendered the employee
unable to work, 50 % of his weekly earnings were payable during the period
of incapacity after the second week. Whereas the 1880 act only entitled the
employee to compensation in the event of a form of negligence on the part
of his employer, the 1897 act entitled the employee to compensation for any
workplace injury, providing it was not wilfully committed by the employee.
The 1897 act was also notable for establishing injured employees right to
weekly incapacity compensation rather than only a lump sum benefit. Further
Workmens Compensation Acts of 1900 and 1906 extended the industries
in which the act was applicable. The 1906 act made its application almost
universal. Similar forms of legislation were introduced around continental
Europe and Scandinavia during the same period (though not in the USA,
Canada or Australia).
A behemoth, prize-winning paper by John Nicoll, published in 1902,
represented the next notable British actuarial contribution on the subject of
employers liability insurance.10 Some 20 years of liability insurance business
experience had been accumulated in Britain by this time, but no actuarial
basis for premium-setting had been established, even in theory, beyond the
early efforts of Neison and Whittall on accident rates. Standard industry prac-
tice was to charge a premium amount set as a percentage of the total wages of
the employees of the business.
Nicoll derived formulae for the value of fatal and non-fatal accident com-
pensation, including the new feature of weekly compensation during the
period of incapacity, from actuarial first principles. For the valuation of fatal
accident compensation, Nicolls formula was a function of the rate of fatal
accident of a given occupation (which was specified as a function of age);
and the probability of the employee having dependants (also specified as a
Neison, in Discussion, Whittall (1882), p.212.

9
Nicoll (1902).
10
function of age). For the valuation of non-fatal accident compensation, his

formula used probabilities of different degrees of incapacitation (total and
partial) occurring for each occupation as a function of age; mortality rates
for the incapacitated as a function of age (treating partial and total incapac-
ity separately); and the rate at which non-fatal accidents recover from their
incapacity. Assumptions were also required about wage levels and how they
varied as a function of age.
Whilst the actuarial community at the Staple Inn discussion of the paper
voiced appreciation for the application of actuarial techniques to this complex
form of insurance, a visiting non-actuarial practitioner in employee liability
insurance was much more sceptical:
Actuaries could not tell, any more than they [general insurance underwriters]
could, what the rate should be, because they had no data the time, he
thought, was not ripe for actuarial knowledge to be called in.11
Thus, by the start of the twentieth century, some recurring themes were
already well-established in the British actuarial relationship with general
insurance: a lack of data rendered statistical and analytical methods of limited
use, and the actuarial role in general insurance was therefore at best unclear.
William Penmans Journal paper of 1911 was the first to focus specifically on
actuarial approaches to reserving for outstanding claims rather than premium
setting.12 Penman was another example, like Samuel Brown, of a life actuary
who dabbled in general insurance thinking, as well as investment thinking,
and who would go on to reach a leading position in the profession (Penman
was President of the Institute of Actuaries during the years 19401942).
The reserves required in a general insurance business can be considered to
have a few high-level components whose relative size will vary with the type
of business that is written. At the highest level, reserves can be required for
two basic categories of claims: claims that arise due to future events that may
occur during the outstanding period of insurance covered beyond the valua-
tion date by the premiums already received; and claims for events that have
already occurred but have yet to be settled.
The reserve for the first of these categories of claims is generally referred
to as an unexpired premium reserve. It reflects the portion of the insurance
premium received that is charged to cover the period of insurance beyond
the valuation date. Alternatively, an unexpired risk reserve may be set up that
11
Green, in Discussion, Nicoll (1902), p.559.
12
Penman (1911).
is intended to reflect the portion of exposed-to-risk that is still outstanding

on the insurance cover (these two quantities may differ for reasons such as
seasonality effectsfor example, more motor accidents might occur in winter
than summer).
Reserves for claims that have already occurred are required due to delays in
the claim reporting and claim settlement processes. These two forms of delay
each has its own form of reserve: incurred but not reported (IBNR) reserves are
for claims that have already occurred but are not yet known to the insurer
(that is, there is a delay between the occurrence of the event that causes the
claim and the notification of the insurer of the claim); and outstanding claims
reserves are for claims that the insurer has been notified of but which are yet
to be settled. Settlement delays may arise due to the time taken to agree the
claim size. This process may simply involve negotiation between the insurer
and the policyholder, or may have to be settled by the courts.
Penmans paper focused on a particular form of outstanding claim reserve
that arose from the 1906 act: the reserve that arose not from settlement delays,
but simply because the form of settlement for incapacity compensation was
in the form of an annuity for the period of incapacity rather than an imme-
diately payable lump sum. The standard industry practice was to reserve for
these outstanding claims by making individual case estimates of the required
reserve based on medical opinion on the duration of the injury. To the trained
actuarial mind, this approach appeared arcane, expensive, unnecessarily sub-
jective and open to deliberate manipulation. Penman contrasted the case esti-
mate approach with life office practice and argued it was surely contrary to
all the principles of valuation to deal with individual cases on their merit.13
Penmans paper adopted a pattern that was similar to Nicolls a decade earlier.
He developed an actuarial method for valuing the reserves based squarely on
standard life office techniques, only for general insurance practitioners to tell
him that there was no adequate data with which to apply his formula. Penmans
approach was intuitive enough as an actuarial approach: the outstanding claim
was a weekly annuity that would remain payable as long as the employee was
alive and remained incapacitated. It could be valued using a mortality table for
incapacitated workers together with a table for the rate of recovery from inca-
pacity (which could be set as a function of age) and a valuation interest rate. In
referring to the calibration of the recovery rates, Penman conceded:
The only way in which these items of information can be obtained is by a very
complete analysis of a large number of claims and such an analysis has not been
possible for me to make.14
Penman (1911), p.112.

13
Penman (1911), p.121.

14
Penman did acquire some claims experience data from two companies and
he used this to develop illustrative bases and valuations. He showed that the
rate of recovery from incapacity tended to be lower at older ages and argued
this should be explicitly allowed for in the reserving basis.
The Staple Inn discussion was broadly negative. The themes were famil-
iar. Several speakers voiced strong doubts about the availability of sufficient
relevant data to implement the actuarial approach set out by Penman. There
were so many factors affecting the duration of incapacity that the separation
of claims into homogenous groups would produce such small groups that
the result would not differ significantly from the case estimation approach
that the author had strongly criticised. The business was in a state of flux,
with factors such as judicial interpretation rendering historical experience less
relevant to the future. One speaker pointed out that valuing life business by
case estimation may not be so absurd if the information was readily available
to support it.
It was almost two decades before another paper appeared in either of the
British actuarial journals on the topic of employee liability insurance (or any
other form of general insurance). Then, in 1929 and 1931, two papers were
published, partly motivated by the 1925 and 1926 Workmens Compensation
Acts. Hugh Browns paper,15 published in the Transactions of the Faculty of
Actuaries in 1931, reviewed the implications of the 1925 and 1926 Acts of
Parliament. A notable feature of these acts was that they allowed the weekly
incapacity compensation payments to be commuted to an immediately payable
lump sum at the behest of the employer, providing the payment was not less
than 75 % of the prevailing market value of an immediate life annuity. This
provided a natural benchmark for reserving for outstanding claims, though the
actuary had discretion to attach a different value to the liability in the regulatory
board of trade returns providing a justification was provided along with details
of the basis used. However, an actuary was only required for the valuation of
weekly incapacity payments that had been in duration for five years or more.
For other outstanding claims, the practice remained to value those liabilities on
an individual case basis using available medical opinion. Hence, 20 years after
Penmans paper and almost 30 years after Nicolls, liability insurance practices
remained largely unaltered by actuarial input: outstanding claims were either
valued on an individual case basis, or were essentially assumed to be life annui-
ties. No actuarial method for the estimation of the duration of the claims based
on statistical analysis of rates of recovery from incapacity was implemented in
practice.
15
Brown (1931).
Brown advocated a pragmatic compromise. Instead of valuing the out-

standing claims using intricate assumptions about incapacity recovery rates
and how they varied as a function of age, together with incapacity mortality
assumptions, he came up with something rather counter to actuarial conven-
tion: that for claims of less than five years duration, age could be completely
disregarded from the annuity valuation. Instead, he suggested grouping the
claims by type of employment and then applying an average duration factor
to each employment group that could be based on the analysis of historical
data. He did not provide any analysis of such data and how these factors
might vary by employment category, however. His suggestion, whilst prag-
matic in theory, remained somewhat enigmatic in practice.
The motivation for the second of this pair of papers, written by
R.G. Maudling and published in the Journal in 1929,16 was related to the
authors own experience of analysing a large amount of liability insurance
claims data. The data was specifically related to the mining industry and
was sourced from the experience of one particular Colliery Owners Mutual
Indemnity Fund. Mutual indemnity funds were a popular approach to co-
insuring liability exposure amongst the employers of some industries, and
the mining industry was one of the biggest examples. He considered the data
for the period from 1908 to 1923, and his data set included 11,000 claims.
Maudling categorised his data by the type of injury or disease that was the
cause of the claim, and by age at date of accident. He found that the rate of
discontinuance of incapacity was a function of both. He used the data to pro-
vide annuity valuation tables as a function of age and type of injury.
This was, in spirit, the implementation of the actuarial frameworks advo-
cated by Nicoll and Penman some 30 and 20 years earlier, finally brought
to life with a meaningful dataset. However, Maudling also noted that the
observed rates of incapacity discontinuance had varied over the data period. In
particular, the discontinuance rates tended to be lower when wages were lower
and vice versa. Data heterogeneity appeared inescapable in general insurance.
Maudlings papers represented the final notable contributions of British actu-
arial thought to general insurance business until after the Second World War.
Risk Theory (19541971)

British actuarial thought on general insurance over the one hundred years
from 1850 to 1950 was primarily focused on attempting to apply the actu-
arial techniques then in successful use in life assurance to fields of general
Maudling (1929).
16
i nsurance. These attempts were continually stymied by the complexities found

in general insurance business: in particular, by the heterogeneity in its claims
experience and by the lack of stability engendered by its greater sensitivity to
unpredictable social and economic change.
Risk theory offered a different analytical approach. Traditional actuarial
techniques in life assurance and pensions were predicated on the idea that
insurance risks were diversifiable and that insurers wrote business in such a
way that these risks were indeed diversified away. Thus actuarial work in life
assurance and pensions was traditionally set within a deterministic, riskless
framework. Risk theory started from a different premise: it applied stochastic
modelling techniques to model the random occurrence and impact of indi-
vidual general insurance risks. This explicitly captured the variability that
could arise from one claim to the next due to fluctuations in both claim size
and frequency. It employed sophisticated mathematical models that went far
beyond the quantitative techniques used in British actuarial science in mid-
twentieth-century life and pensions.
The germs of risk theorywhich we could loosely define as the stochas-
tic modelling of individual insurance claim outcomeswere developed in
Scandinavia and particularly in Sweden during the early decades of the twen-
tieth century.17 This initial work was focused on the modelling of mortality
risk in life assurance, but it was found to have no practical use there as its
application to the large sizes of homogeneous policies in life business left
little random fluctuationrisk theory essentially produced the same results
as the standard deterministic actuarial models. Following the end of the
Second World War, actuarial researchers began increasingly to consider risk
theory in the setting of general insurance. This largely occurred outside the
UK. Researchers were again very active in Scandinavia during this period,
most notably in Finland.18 Somewhat later, technical actuaries such as Hilary
Seal further developed and applied these ideas in the USA.19
The essence of the post-war general insurance risk theory framework can
be summarised as follows. Individual claim events and individual claim sizes
were each modelled as random variables. Claim events were typically assumed
to be generated by a Poisson process. This could be generalised into a mixed
Poisson process, where the Poisson rate was itself assumed to be a random
variable. The claim size was typically assumed to have a skewed probability
distribution such as the lognormal or gamma distribution. In individual risk
17
See Lundberg (1909) for example.
18
See, for example, Pentikainen (1952).
19
See, for example, Seal (1969).
theory, each policy in the portfolio had its own specified probability distribu-
tion for claim frequency and claim size. In collective risk theory, the models
and calibrations were specified at the portfolio level without distinguishing
which policies gave rise to the claims (but individual claim events were still
modelled).
This analytical framework could provide many theoretical insights. At the
most basic level, it was one way of calculating the expected level of claims,
and hence the pure premium for insurance. It could also be used to model
claims with and without reinsurance and to set the pure premium for a par-
ticular form of reinsurance. For a given level of premium and starting capital,
the evolution of the risk reserve could be modelledthat is the stochastic
projection over time of assets and premiums less claims and expenses. The
modelling of this reserve could be used to estimate probabilities of ruin over
various time horizons for a given starting reserve. Equivalently, it could be
used to establish the starting level of reserve required to support a given prob-
ability of ruin.
This analysis of ruin probabilities was fundamentally very similar to the
approach pioneered by Sidney Benjamin and adopted by the Maturity
Guarantees Working Party for unit-linked investment guarantee reserv-
ing during the 1970s. In maturity guarantee reserving, the ruin probability
modelling focused on the impact of stochastic variation in non-diversifiable
(financial market) risk. Risk theory focused on the stochastic impact of what
might be called under-diversification of risks that were theoretically diversi-
fiablethat is, the impact of random fluctuations in claims processes that
were usually assumed to be independent (claims may also be highly correlated
due to exposure to a common event or risk factor). As we shall see below,
Benjamin also played a significant role in developing British actuarial thought
in general insurance.
Whilst the theoretical framework of risk theory was intuitive and powerful
in the context of general insurance business, it did not necessarily provide a
solution that was any less challenged than traditional actuarial techniques by
the age-old practical general insurance modelling issues of data, heterogeneity
and stability. Could the models of risk theory be reliably calibrated such that
it could produce dependable quantitative output?
Risk theory never reached the mainstream of the British actuarial profes-
sion. Its primary British actuarial exponent during the post-war decades was
Robert Beard. He liaised extensively with international actuarial colleagues
from the late 1940s onwards, especially the leading Finnish thinkers in the
discipline, to apply the subject to British general insurance business. Beards
interest in risk theory was mainly stimulated by his experience in q uantitative
operational research during the Second World War rather than by his actu-
arial training.20 He appealed to the Institute of Actuaries as early as 1948 to
engage more actively in general insurance, but he received little encourage-
ment. Faced with a less-than-enthusiastic institute, Beard helped to estab-
lish ASTINActuarial Studies in Non-Life Insuranceas a section of the
International Actuarial Association.21 Its journalthe Astin Bulletincarried
much of Beards substantial research output of the 1950s and 1960s.
Beard did manage to get a couple of papers on risk theory and its applica-
tion to general insurance published in the Journal of the Institute of Actuaries
in 1967. He also had a greater number published in the perhaps more liber-
ally minded Journal of the Institute of Actuaries Student Society. Beard also
co-wrote a book,22 first published in 1969, on risk theory with two of the
leading Finnish actuarial thinkers on risk theory, Professor Pentikainen and
Dr Pesonen. The book was widely used in European actuarial university
courses and subsequent editions were published in the 1970s and 1980s.
A notable Beard paper was published in the Journal of the Institute of
Actuaries Students Society in 1954.23 Beard gave British actuarial students an
overview of risk theory and demonstrated its application in general insur-
ance with an example based on a large US fire insurance claims dataset. The
paper showed that the distribution of claim size could be well-fitted with a
log-normal distribution, except for a handful of exceptionally large claims out
of a dataset of more than a quarter of a million. It also showed how an excess-
of-loss reinsurance treaty impacted on the insurance portfolios net claims
probability distribution. He went on to analyse how the probability of ruin
behaved over various time horizons for differing levels of starting reserves, and
how the excess-of-loss treaty impacted on these results. It was an accessible
and comprehensive practical case study on the application of risk theory to
general insurance reserving, solvency assessment and risk management.
Beard was not a mere theorist. He was a working actuary who recognised
the practical limitations of the mathematical framework and the data that
was used to fuel it. For example, he noted that other US statistics showed
that higher claim frequencies tended to be experienced in years of economic
depression, and less in boom years, and that this could be an important source
of heterogeneity and non-stationarity. He also noted that claims would tend
not to be independent due to geographical exposures. Nonetheless, he held a
20
Beard in Discussion, Plackett (1971), p.355.
21
Beard in Discussion, Abbott etal. (1974), p.277.
22
Beard etal. (1969).
23
Beard (1954).
deep conviction that there were significant practical insights for general insur-
ance premium-setting and reserving that could be obtained from a sensible
utilisation of available data within the sophisticated mathematical modelling
of risk theory.
The editors of the Journal of the Institute of Actuaries permitted Beard a full
seven pages in its September 1967 edition.24 He argued that the risk theory
approach of modelling claim frequency and claim size as separate statistical
processes could make it easier to detect changes in patterns of claims experi-
ence and thus allow for faster premium rate adjustments to be made by the
insurance office. This was a topical issue for the general insurance industry at
the time, especially in motor insurance, where a deterioration in experience
had occurred over a number of years without premium bases adequately react-
ing to avoid a sequence of multiple years of loss.
Although it was not accepted into the mainstream of the professions think-
ing at the time, Beards work of the 1950s and 1960s laid the foundations and
provided the inspiration for broader quantitative research in general insurance
by the British actuarial profession. Meanwhile, progress started to be made
in the collation of relevant claims data in some general insurance lines, most
notably motor insurance. In 1967, the British Insurance Association estab-
lished the Motor Risk Statistics Bureau to pool claims experience data for
several member insurers. This data, whilst challenged by differences in rating
structures and policy types across member firms, provided a new source for
the application of statistical techniques to pricing and reserving.
The 1970s saw more actuaries become engaged in general insurance prac-
tice. As noted above, the Institute eventually added general insurance to its
examination syllabus in 1978. Some new British actuarial thought-leaders
emerged in the 1970s who were seeped in practical experience in general
insurance business. G.B.Hey was one such example and he helped to break
new ground in 1971 when he and P.D.Johnson wrote the first ever paper on
motor insurance to be published in the Journal of the Institute of Actuaries.25
Their paper was not as mathematical as typical risk theory works, but it was
strongly influenced by Beard and his work in ASTIN.
Johnson and Hey analysed the efficacy of experience rating in motor insur-
ancethat is a system by which the premium of the individual risk depends
upon the claims experience of this same individual risk.26 In British motor
insurance, this was known as the No Claims Discount (NCD) system, and
24
Beard (1967).
25
Johnson and Hey (1971).
26
Johnson and Hey (1971), p.202.
it was a well-established feature of that market by the mid-1960s. The indi-

vidual risk data used in an experience rating system still faced the same chal-
lenge as a risk rating approach based on broader historical claims experience:
a longer period of history was required to effectively differentiate the risk
heterogeneity amongst different policyholders, but older data may be of less
relevance for the projection of future claims behaviour. For example, in NCD
systems, the advances in the policyholders driving skills that would typically
accompany increasing age and driving experience could render earlier claims
experience irrelevant to expected future claims behaviour.
A typical UK NCD system of the time allowed the maximum claims dis-
count to be obtained after four or five consecutive claim-free years. Johnson
and Hey analysed how effective this chosen length of period was in differen-
tiating different risks (under the assumption the risks were indeed station-
ary). They considered a theoretical pool of policyholders where 75 % of the
group had a claim frequency of 0.1 and the remaining 25 % had a higher
claim frequency of 0.25. Their analysis showed that a policyholder in the low-
claim-frequency group had a two in three chance of obtaining the maximum
discount rating after four years, whilst the high-claim-frequency group poli-
cyholder had a corresponding probability of two in five. NCD systems made
some contribution to distinguishing policyholder heterogeneity, but they
were a blunt and limited tool that could only be highly effective with a long
history of stationary data. This analysis highlighted that the source of the poor
motor insurance industry experience in the years preceding the paper may not
only have been due to high claims inflation as was generally suspected (and
which was partly driven by changing judicial treatment of third-party injury
claims). Another factor may have been that the average policyholder who
earned the maximum NCD discount was not actually as good a risk as they
had been assumed to be.
Aside from this theoretical analysis, Johnson and Heys paper focused on
analysing the data made available through the Motor Risk Statistics Bureau
(though they only considered the experience of a single member firm to avoid
complications with comparability and consistency). They fitted an eight-
factor regression model of claims frequencies using a least squares optimisation
(factors were intuitive motor insurance risk factors such as age of policyholder,
car rating group and NCD category). They found that the NCD category was
a significant variable in the regression, even in the presence of the other fac-
tors, and hence concluded it was still an important element in the premium
rating process, despite its noted limitations.
Despite such attempts to illustrate the practical application of risk theory
ideas and statistical methods, it was generally viewed through the 1950s,
1960s and even 1970s as too theoretical and quantitative to be of much prac-
tical value to the British actuary. This perspective is well-represented, though
perhaps a little overstated, by the following passage in a Journal paper by
Ryder published in 1976:
Risk theory is a rather esoteric branch of actuarial theory which has been exten-
sively developed by the more theoretical continental actuarial tradition. The
practical actuary, however, finds that he hardly ever uses this theory.27
The application of risk theory, however, did become an established element

of British actuarial practice in general insurance in the 1980s and 1990sa
period when the profession developed a more prominent role in the sector
and the actuarial approaches employed there became increasingly technical.
Claim Reserving (19741996)

Between the years of 1850 and 1970, the British actuarial profession made
little headway in establishing itself as an influential and important component
of Britains highly successful, global general insurance industry. It had tried to
import its life assurance methods into general insurance and had learned that
they were not adequate for the task of pricing and reserving for general insur-
ance. It had seen and broadly rejected the stochastic approach of risk theory
that had been developed and applied by overseas actuaries. Was there a role
for British actuaries to play in general insurance? If so, what was it and what
kinds of actuarial skills would it employ?
As noted above, over the course of the 1970s, there was an increase in the
numbers of British actuaries working in general insurance business. A General
Insurance Study Group was formed by the Institute in 1974. It produced
many research papers, though most of these were merely deposited at the
Staple Inn library and were not deemed worthy of publication in the Journal
or discussion at sessional meetings. The profession recognised that there may
be an opportunity for actuaries to play a more formal role in statutory reserv-
ing for general insurance business. The freedom with publicity regime of
actuarial life assurance reserving, which had been in place in some form or
another since the 1870 Life Assurance Companies Act, provided an explicit,
professional and statutory role for the actuary that gave them considerable
discretion to exercise their expert professional judgement. No such equivalent
Ryder (1976), p.71.

27
role existed in general insurance reserving in the 1970s. General insurance

firms had to calculate statutory reserves, but no actuarial role was mandatory
in reserving and no detailed disclosure of the methods use in assessing the
reserves was required.
The collapse in 1971 of Vehicle and General, a British motor insurance
firm, provided a further opportunity for the actuarial profession to press for
a more active and statutory role in general insurance reserving. An Institute
working party was established in the early 1970s with a remit to consider the
statutory reserving regime for general insurers with reference to uniformity
and the verification of non-life reserves.28 Perhaps unsurprisingly, its report,29
published in the Journal in 1974, concluded that only professional certifi-
cation could address the challenges inherent in general insurance reserving.
Nonetheless, even amongst actuaries there was scepticism that this could be a
cure-all. In the Staple Inn discussion of the paper, J.L.Manches noted:
It was perhaps an oversimplification to believe that actuarial techniques were the

best answer to accurate claims reserving. Life valuations were based on the appli-
cation of established claim probabilities (i.e. mortality tables) to homogeneous
groups of policies which generated claims of known amounts. None of that
applied to non-life business, and there was bound to be the utmost difficulty in
reaching agreement on any standard approach to reserving.30
There is certainly a hint of arrogance in the professions position that it

could improve standards in general insurance reserving when it had done so
little over the previous decades to encourage actuarial research and education
in the field. However, it is important to note a shift in emphasis. No longer
was the profession positioning itself as the leading provider of statistical or
analytical methods. By the early 1970s it was clear that advanced techniques
in mathematical statistics were emerging faster than the professions appetite
to apply them. On the contrary, the professions argument was increasingly
based on its ability to meet the need for sound professional judgement and
broader financial acumen in the setting of prudential insurance reserves.
From the early 1970s until his death in the early 1990s, Sidney Benjamin
whom we met earlier as the frustrated pioneer of stochastic risk-based
approaches to maturity guarantee reservingrelentlessly argued for a statu-
tory role for actuaries in general insurance reserving that was equivalent to
their position in life assurance. This argument was based on the inadequacies
28
Abbott etal. (1974), p.217.
29
Abbott etal. (1974).
30
Manches, in Discussion, Abbott etal. (1974), p.268.
in general insurance statutory reserving that he perceived arose from the

absence of a professional actuarial role:
Historically, the first job of the actuary is to safeguard the interests of policy-
holders. The life actuary determines the amount of risk capital which should be
set aside to give an acceptable level of safety to the policyholders. In non-life
insurance, that does not happen. The amount of backing solvency capital for
any volume of business is set vaguely, according to an informed perceived wis-
dom, with: no scientific justification; no explicit public justification; no pub-
lished standards of consistency within any one company from year to year; and
no apparent standards of consistency between companies in any one year.31
Benjamins paper, Profit and Other Financial Concepts in Insurance,32

appeared in the Journal in 1976 and presented this case with typical Benjamin
gusto. He discussed the universal applicability of fundamental actuarial con-
cepts such as the use and disclosure of actuarial bases that provided transpar-
ency and consistency to the reserving process; the difference between reserving
and premium bases and its implications for new business strain and the emer-
gence of surplus; implicit and explicit reserving margins for prudence; asset
mismatching reserves; differences between provisions and reserves; prospec-
tive and retrospective reserving approaches. Benjamin argued that these actu-
arial concepts were as applicable to general insurance as to life assurance, and
that they were the unique domain of the actuarial profession.
Benjamins paper also offered some specific criticisms of the practices that
then prevailed in general insurance reserving. Like other actuaries before him,
he viewed reserving by case-by-case estimation as fundamentally subjective
not inherently stable in the way a fixed basis will be and hence an inad-
equate substitute for a reserving basis.33 He was also critical of the standard
industry practice of not discounting projected claims cashflows when setting
general insurance reserves. Whilst this provided a form of implicit reserving
margin, its size was arbitrary and it distorted the stated form and timing of
the release of surplus. He also argued that required solvency margins should
be risk-based and should be a function of the asset mix of the business, which
was again not standard practice at the time.
His most forceful argument, however, was reserved for the use in g eneral
insurance of the professional framework of freedom with publicity in
statutory reserving: the idea that the actuary, as the professional expert, should
31
Benjamin in Discussion, Ryan and Larner (1990), p.658.
32
Benjamin (1976b).
33
Benjamin (1976b), pp.25253.
have the freedom to use his judgement to set appropriate assumptions and
methods for the business he is reserving for, providing he disclosed his assump-
tions sufficiently such that another actuary could reproduce his results and
opine on the reasonableness of the approach taken. To Benjamin, this was the
fundamental reserving discipline that actuaries could bring which was lacking
in British general insurance. Whilst the Staple Inn discussion of the paper was
broadly supportive of Benjamins thinking, there were some general insurance
practitioners who felt the need to highlight that the intrinsic differences in
general insurance and life assurance could limit the direct applicability of some
of his arguments and that actuarial advice [in general insurance] would
continue to be sought if, and only if, actuaries showed a proper understand-
ing and humility of approach to an industry which had operated successfully
without actuaries for many years.34
In the years following Benjamins 1976 paper, a flurry of papers appeared
in the Journal which did indeed attempt to develop a proper understand-
ing of an approach to claim reserving. These papers were some of the
most technical and quantitative papers ever to appear in the Journal, and were
mostly written by actuaries, and in some cases non-actuaries, with PhDs in
advanced quantitative fields. They were mainly concerned with developing
improvements in methods for the projection of claims from a run-off triangle
(which tabulated claims paid in rows of the year of insurance, which was usu-
ally referred to as the underwriting year and sometimes as the origin year; and
columns of development year, i.e. the number of years after the origin year
when the claim was paid). The concept of a run-off table of general insurance
claims had been around for a long time. For example, at the Staple Inn discus-
sion of Penmans 1911 paper, W.R.Strong noted:
If the claim payments in respect of the business accepted in any given year were
traced separately until the claims of the year were finally disposed of, the total
volume of payments year by year fell into a somewhat regular sequence of
diminishing amounts It might perhaps be practicable when a sufficient
period had elapsed to construct a table by means of which, given the claim pay-
ments up to the end of the year, an estimate might be formed of the ultimate
cost of disposing of the liability in respect of the policies of the first year.35
The standard method of estimation of the ultimate cost from the run-off
data was the ubiquitous chain ladder method. The method assumes that future
34
Scurfield, in Discussion, Benjamin (1976b), p.300.
35
Strong, in Discussion, Penman (1911), pp.13738.
claims for each underwriting year will accumulate over their outstanding
development years in proportion to how claims paid up to that point in the
origin years development differ from the average observed across all the origin
years in the run-off table for which the outstanding development period has
been observed. Its greatest limitation is embedded in its basic assumption: it
implies that an unexpected claim size in an early year of development impacts
proportionally on all future development years for that underwriting year.
This sensitivity could result in noisy, unstable estimates for business in its early
years of development.
The first of this series of technical papers was written by D.H.Reid and
published in the Journal in 1978.36 Reid began by noting that case-by-case esti-
mation was still the prevalent industry practice for setting outstanding claims
reserves in general insurance. He voiced the usual actuarial concerns with
this approach, but particularly highlighted that whilst the approach might
sometimes have some merit in reserving for claims that had been reported but
not yet settled, it was entirely inapplicable to IBNR claims and their required
reserves. The reserving approach for these claims must involve some form
of statistical method as there was no case-by-case information that could be
used in the reserving assessmentby definition, the claim was completely
unknown to the insurer at this point in its development.
Reid developed a mathematical framework for the emergence of claims
payments over time by specifying a cumulative joint probability function for
claims paid and development year. He then fitted this function to the available
data for the claims experience from the earliest available underwriting year of
the run-off data only. His model was essentially smoothing the observed expe-
rience of a single complete sample path for the claims development. Once this
function had been fitted, it was then transformed into a function for use in
projected future claims via parameters for inflationary changes in claim size
and changes in the rate of settlement that were assumed to be experienced
between the time of the first underwriting years claims and the projected
times of future claims. These parameters could be fitted to the claims run-off
data for the sequence of subsequent underwriting years for which data was
available. Reid applied his framework to example claims data for a variety of
lines of business such as employers liability, fire and motor insurance. The
approach was somewhat aligned to risk theory in that it provided a full proba-
bilistic description of (aggregate) claims. But its formulation was complex,
and the model contained a very large number of parameters that needed to be
fitted to typically very limited data. His presentation was rather impenetrable
Reid (1978).
36
for the typical British actuary of the time. Crucially, it was extremely difficult
to ascertain from his analysis whether his complex modelling would result in a
more accurate or reliable estimate of outstanding claims than a much simpler
modelling approach.
D.H. Craighead produced a more accessible paper for the Journal in
1979.37 Craighead was an experienced actuarial practitioner in the Lloyds
of London market. His paper gave a broad overview of its business practices
and institutional arrangements. It also contained an important section with
his views on how to model the run-off of claims and hence establish claim
reserves at any point of time. He proposed fitting a formula for the incurred
loss ratio of a given underwriting year as a function of the development year.
He applied this approach to the proprietary claims data of an anonymous
reinsurance company using a three-parameter exponential form of function.
This produced fits of varying degrees of quality for different lines of underly-
ing business and forms of reinsurance.
The fitted curves provided a smoothed description of how claims had histori-
cally run-off over their development period. It did not provide an explicit statis-
tical predictive model. The fitted parameters could then be used to extrapolate
the claims run-off of the underwriting years that were not yet fully developed.
This was essentially a parametric form of the chain ladder method and, in simi-
larity with that method, Craighead noted the sensitivity of the reserve estimate
to unexpectedly large claims that arose early in the claim development period.
In the Staple Inn discussion, this theme was expanded upon by J.P. Ryan, a
general insurance actuary who made notable contributions to actuarial research
in the 1980s and 1990s. Ryan highlighted that the work done in the USA by
Bornhuetter and Ferguson might provide the solution to this problem.38 The
Bornhuetter-Ferguson approach reduced the sensitivity of projected ultimate
claims to the claims experience data by mixing the estimate implied by the raw
claims data with some specified prior expectations for the claims development
pattern. It was essentially Bayesian in spirit, and its inherent limitation was in
the potentially arbitrary specification of the prior estimates.
Three further technical papers on statistical methods for general insur-
ance claim reserving appeared in the Journal in 1982 and 1983. The first of
these was written by J.H.Pollard.39 Pollard focused on the run-off behaviour
of the aggregate claims of a large book of business. He invoked the Central
Limit Theorem to justify assuming the aggregate claims would be normally
distributed.
37
Craighead (1979).
38
Bornhuetter and Ferguson (1972).
39
Pollard (1982).
Pollard moved away from the direct manipulation of the run-off triangle and
instead set up a matrix algebra to describe how claims paid behaved through
their development years: he specified vectors for the mean and variance of the
claims paid in each development year, together with a covariance matrix to
capture the correlations between claims paid in different development years
(for example, the correlation between claims paid in development year 1 and
development year 3). This, almost tautologically, allowed the expected claims,
and indeed the whole probability distribution of future claims to be deter-
mined for a given claims development period to date.
The second useful feature of Pollards set-up was that it provided a statistical
basis for assessing whether statistically significant changes in claim settlement
patterns had arisenthe multivariate normal framework allowed Chi-Square
significant tests to be used to assess if the claims development was statistically
different to the previously fitted distributions. Pollards framework was mathe-
matically elegant and intuitive, but, as ever, it relied entirely on how the model
was parameterised and the reliability of the available data for that purpose.
The two claim reserving papers published in the Journal in 1983 were
the most statistically complex. The first of these was written by de Jong and
Zehnwirth.40 Their paper applied recent technical developments in the sta-
tistical modelling literature to the claim reserving problem, particularly the
times series modelling of Box and Jenkins.41 This involved state-space models
and the Kalman filter, which was essentially a recursive application of Bayes
Theorem. This mathematical statistical technology allowed claim reserve esti-
mates to be dynamically updated in a Bayesian way as new data arose. The
second paper,42 by G.C.Taylor, drew parallels with the claim projection prob-
lem and invariance problems in physics and used this observation to apply
variational calculus to the stochastic modelling of run-off triangles. The invari-
ance assumption was that the expected amount of outstanding claims, deflated
to current values, is invariant under all variations of future speed of finaliza-
tions.43 Taylor recognised that such an assumption would not hold if a change
in rate of settlement arose, for example, due to a change in negotiating stance
of the insurer. But of course, no quantitative method could readily incorporate
such factors into claims projections. Neither of these papers were presented at
Staple Inn and it is hard to believe they resonated strongly with either general
insurance practitioners or the British actuarial profession at large.
40
de Jong and Zehnwirth (1983).
41
Box and Jenkins (1970).
42
Taylor (1983).
43
Taylor (1983), p.211.
A working party was established in 1982 by the Institutes General Insurance

Study Group to formally consider general insurance solvency and the meth-
ods and bases used for the valuation of assets and liabilities.44 This was partly
inspired by a recent study of general insurance solvency by Finnish actuaries
which was particularly notable for its application of simulation modelling as
a solution to risk theory problems.45 The profession had embraced simulation
modelling methods in its recently-published and influential maturity guar-
antee research. It is certainly easy to see why it seemed like a more appealing
route for the profession to pursue than further developing the work of Reid,
De Jong and Zehnwirth, and Taylor.
The report of the working party was published as a Journal paper in 1984.46
No significant regulatory or professional guidance applied to general insurance
reserves at this time other than that valuations should be made in accordance
with generally accepted accounting principles or other accepted methods.
Statutory reserving required the assessment of two key quantities: the tech-
nical provisions (i.e. liability valuation) and the solvency margin. The stan-
dard industry practice was for the technical provisions assessed in solvency
reporting to also be used as the provisions shown in financial statements.
Conceptually, the solvency margin provided an asset buffer over the cost of
meeting the liabilities. The regulator had powers to intervene in the running
of the insurer in the event that assets were insufficient to cover the solvency
margin as well as the provisions.
The working party argued that, for the purposes of solvency reporting,
technical provisions should contain a prudent margin over the best estimate
of the cost of meeting the liabilities. This margin should be set such that there
was a relatively low risk of [the technical provisions] proving inadequate.
Meanwhile, the solvency margin would protect against the more remote
adverse contingencies of the run-off.47 This approach was difficult to recon-
cile with accounting principles for the valuation of provisions (which required
the provisions to be a true and fair estimate of the liability value). It also
ran counter to the applicable European Commission regulatory framework,
which assumed technical provisions were best estimates, with a solvency
margin calibrated to produce a risk-of-ruin of one-in-1000 over a three-year
horizon. The working party was satisfied with the solvency margin target but
suggested that technical provisions should be set at a one-in-200 risk-of-ruin
44
Daykin etal. (1984), p.279.
45
Pentikainen and Rantala (1982).
46
Daykin etal. (1984).
47
over the three-year horizon (which was clearly substantially different to the
one-in-two risk-of-ruin loosely implied by a best estimate approach).
The working party also advocated an approach similar in spirit to life assur-
ance reservings freedom with publicity concept: insurers and their actuaries
would be free to choose their own methods to determine the technical provi-
sions and solvency margin, but disclosure of those methods and professional
certification by a loss-reserving specialist would be required. However, in the
Staple Inn discussion of the paper, some concerns were raised that this free-
dom was fraught with danger unless the Department [of Trade and Industry,
the insurance solvency regulator of the time] is able to bring to bear a strin-
gent monitoring of the results.48
The working party recommended that the above ruin probabilities should
take account of asset-side risks as well as liability risksin essence, a mis-
match reserve should be included in both the technical provisions and sol-
vency margin. It may be recalled that Sidney Benjamin first proposed that
general insurers solvency margin should include a mismatch reserve in 1976,
but it remained a significant departure from prevailing general insurance
practice in 1984. Such a change in reserving method could imply a significant
increase in reserving levels. However, the working party attempted to water
down the ramifications of this proposed change with a caveat:
If asset values fall only temporarily, the problem may be largely presentational,
and the supervisor would not need to withdraw the authorization of companies
unable to meet the solvency requirements at a particular date if the position had
subsequently been rectified. Only with a prolonged shift in market values would
the effects be serious.49
Quite how the supervisor, or the certifying loss-reserving specialist, was

supposed to ascertain at a particular date whether experienced asset value falls
were temporary or not was not addressed by the working party.
The working party, again led by Chris Daykin, produced a further paper
that was published in the Journal in 1987.50 Whereas the first paper had dealt
mainly with principles and a conceptual framework, the second had a greater
focus on implementation and the modelling challenges involved therein.
The second paper did, however, make one conceptual U-turnafter profes-
sional and industry feedback, the working party now expressed its ambiva-
48
Hart, in Discussion, Daykin etal. (1984), p.320.
49
50
Daykin etal. (1987).
lence on the question of the statistical standard for technical provisions,

arguing that it was the total solvency level that was really relevant. This effec-
tively rescinded their earlier proposal that technical provisions should include
a margin for prudence and hence accepted a best estimate definition. This
allowed closer alignment of provisions with accounting principles and true
and fair valuation.
The 1987 working party paper proposed that the required solvency mar-
gin be calculated using a simulation model that projected asset and liability
cashflows over the run-off of the existing business. It adopted a pragmatic
approach that did not attempt to reach the heights of statistical ambition that
had been explored by Reid, Taylor, de Jong and Zehnwirth over the previ-
ous decade. It suggested that variations in the real amounts of claims should
be modelled at an aggregate level. As in Pollards work, the Central Limit
Theorem could be used to support the use of a normal distribution to describe
variation in aggregate claims. They suggested that the standard deviation of
the claims paid in a given year should be a function of the size of that years
expected aggregate claim. Variations from year to year in claims paid were
assumed to be independent. Asset variation and inflation were modelled using
a version of the Wilkie model that was recalibrated with the objective of being
more appropriate to the shorter-term horizons of general insurance business
than the long-term projections for which the model was originally intended.
This modelling framework could be used to determine the starting amount
of assets in excess of the technical provisions that was required to support a
given probability of ruin. The working party again suggested that an actu-
ary should write a public report on the financial strength of the company
that presented these findings. The paper showed that illustrative calibrations
of this modelling framework could generate intuitive assessments of solvency
marginfor example, the central case of their example suggested a solvency
margin of around 10 % of the (best-estimate) technical provisions would be
required. Of course, such results were entirely predicated on their assumptions
about the scale of variability in the claims run-off. Their example calibration
was developed in a very heuristic way. The paper did not offer any substantial
guidance on how these parameters could be robustly calibrated to reflect the
specific features of a given general insurance business.
Whilst the Solvency Working Group was developing its vision of a solvency
framework and the role of the actuary within it, other actuarial researchers
continued with the investigation of quantitative techniques for estimating
claim reserves. The most intuitive and accessible paper of the 1980s on general
insurance claim reserving methods was a Journal paper written by Sidney
Benjamin and Ian Eagles, published in 1986.51 Benjamin and Eagles analysed
historical claims run-off patterns in Lloyds syndicates and found that the ulti-
mate loss ratio tended to have a strong linear dependence on the year one paid
claims ratio. This linear dependency relationship had already been alluded to
by Pollards development year correlation matrix but Benjamin and Eagles
presentation implied a very simple mathematical rendering that appeared to
produce good empirical fits: ultimate cumulative claims, and hence required
current reserves, could be estimated by simple linear regression of the cumula-
tive claims of a given development year with the ultimate cumulative claims
paid. The empirically observed variation around the line of best fit could also
provide some heuristic indication of the extent to which ultimate cumulative
claims could deviate from the extrapolated estimate. These regression rela-
tionships could be fitted to different business lines and years of development.
Naturally, the later the year of development, the more confidence could be
had in the ultimate loss estimate. It was classic Sydney Benjaminavoid-
ing unnecessary statistical niceties, it cut to the chase and delivered powerful
practical actuarial insight. In a world increasingly characterised by rocket sci-
ence and advanced computing technology, Benjamin had a knack of making
original and informative use of the back of an envelope.
The working partys normal distributions and Benjamins linear regressions
may have led the British actuarial profession to exhale a collective sigh of relief
at the potential accessibility of new general insurance reserving methods. The
investigation, however, of the use of advanced statistical techniques in claim
reserving was far from over. Three further papers of a high statistical ambi-
tion were published in the Journal in 1989 and 1990. The first two of these
papers were the fruits of a seminar Applications of Mathematics in Insurance,
Finance and Actuarial Work, jointly sponsored by the Institute of Actuaries
and the Institute of Mathematics and its Applications. These two papers, by
R.J.Verrall52 and A.E.Renshaw53 respectively, covered broadly similar statis-
tical ground. They both observed that the ubiquitous chain ladder method
could be considered as a form of two-way analysis of variance (ANOVA).
From this observation, a similar recursive Bayesian estimation approach as
that developed by de Jong and Zehnwirth could be developed. It could also
be shown that, under the statistical assumptions of the ANOVA model and
an assumed lognormal distribution for claims, the chain ladder method did
not produce the maximum likelihood estimates for the expected claims. This
51
Benjamin and Eagles (1986).
52
Verrall (1989).
53
Renshaw (1989).
more statistically sophisticated analytical approach could also shed some light
on the parameter stability (or lack thereof ) of the chain ladder method, espe-
cially for the most recent underwriting years.
T.S. Wrights paper, the third of this series of highly technical investiga-
tions of advanced statistical techniques for claim reserving, appeared in the
Journal in 1990.54 Like the work of de Jong and Zehnwirth, this again used the
Bayesian Kalman filter statistical technology to produce stochastic projections
of claims run-off. However, unlike the papers by Verrall and Renshaw, Wrights
approach did not rely on the assumption that claims were log-normally distrib-
uted. This was an assumption which Wright regarded as an untenable descrip-
tion of general insurance claims distributions. Several technical assumptions
about the distributional characteristics of the claims process were still required
by Wrights approach, but it allowed the statistical insights to be placed in a
more general family of distributions than the previous research.
Whilst these advanced statistical methods may have had some application
in insurers internal assessments of claims experience and profitability, they
were too exploratory and complex for use in 1990s statutory solvency assess-
ment. As has so often been the case in the history of British actuarial engage-
ment in general insurance, further inspiration was sought from overseas. A
group of British actuaries authored a paper reviewing the recent US regulatory
developments which appeared in the British Actuarial Journal in 1996.55 The
US regulatory authorities introduced a Risk-Based Capital (RBC) system in
the early 1990s. The essential idea was that the capital requirement (similar
to the solvency margin in the UK) would be calculated using a series of pre-
scribed factors that were applied to a defined metric of volume of business
(such as premiums earned or reserves net of reinsurance). Hence the capital
requirement would be determined as some percentage of net reserves, and
the percentage would be determined formulaically as a function of the mix
of insurance business and asset risks on the balance sheet. The underwriting
and claim reserve risk factors were set on a rolling basis to reflect the worst
industry experience of the previous ten-year period. This mechanical calibra-
tion approach was open to the criticism of being too retrospective for the
fast-changing world of general insurance.
The authors highlighted some of the limitations of this simplified one-size-
fits-all formula application, most notably its inability to capture the aggre-
gation of exposure to a single underlying risk event. They suggested that a
Dynamic Solvency Testing (DST) approach, implemented within a statutory
54
Wright (1990).
55
Hooker etal. (1996).
professional framework, would be preferable to a formula-based approach.

DST meant the insurer developing their own financial model of their business
and using it to project and analyse the health of the business under a range
of selected adverse stress scenarios. They also considered the use of dynamic
financial analysis within this system, where the stress modelling is replaced
with a full set of stochastic scenariosin essence, what had been proposed by
the Solvency Working Party in 1987. The application of the DST approach
had since been pioneered in statutory reserving for Canadian general insur-
ance in the early 1990s. The Canadian Institute of Actuaries had been active
in developing professional guidance on the professional role actuaries could
perform in a DST statutory framework. As might be expected, this call for a
broad-ranging role that involved substantial actuarial judgement and freedom
was generally welcomed in the Staple Inn discussion of the paper.
Yet the elephant in the room still remaineddid the British actuarial pro-
fession have the skills and experience required to perform this type of role in
general insurance? In the discussion, the influential general insurance actu-
arial practitioner D.H.Craighead tried his best to take a positive stance: I
expressed reservations then [several years ago], but I think that we are now
beginning to be ready for such a role.56
ctuaries andtheFinancial Management

A
ofGeneral Insurers (19811994)
In the 1980s and 1990s, technical actuarial research in general insurance
broadened beyond the assessment of claim reserves and solvency margins to
consider the wider horizons of the financial management of insurance firms.
Sidney Benjamin was ahead of his time in this field as he was in maturity
guarantee reserving, and his 1976 paper Profit and Other Financial Concepts
in Insurance was an early forerunner to this strand of thinking. But the 1981
Journal paper by Abbott, Clarke and Treen, Some Financial Aspects of a
General Insurance Company, with its focus on the measurement and man-
agement of shareholder returns, arguably marked the real departure point
from traditional actuarial ground.57
Like Benjamin in 1976, Abbott et al. advocated the use of discounted
reserves, providing inflation was consistently applied to the claims projections.
The paper particularly focused on how undiscounted reserves resulted in a
Craighead in Discussion, Hooker etal. (1996), p.313.

56
Abbott etal. (1981).

57
misleading profit emergence pattern over the life of the business. It high-
lighted how material this effect could be in the high inflation environment
that prevailed at the time the paper was written and how changes in inflation
and interest rates over the run-off of the business could further distort the
emergence of profits when using the undiscounted reserving approach. The
paper also discussed how to assess the return on shareholder capital that had
been earned by a general insurance business. It did not, however, venture to
answer what return ought to be required by general insurance shareholders
other than to suggest that an arbitrary amount in excess of the real risk-free
interest rate should be a form of profit objective.
A short paper by G.C.Taylor published in the Journal in 1984 considered
the capital that should be required to write new business and from where this
capital should come.58 In particular, in general insurance in the 1970s, the
notion had developed that insurance business should be self-financing: that
is, solvency margin requirements should be funded entirely from premium
loadings. Taylor argued that solvency margin requirements should be consid-
ered as part of the working capital needs of a general insurance business and
should therefore be funded by the shareholder rather than the policyholder. He
further argued that whilst these funds could be invested and would generate
investment return for shareholders, this return would be inadequate compen-
sation for shareholders due to effects such as double-taxation. Hence premi-
ums should be loaded to generate an acceptable expected return on that capital
(which would be less than the loading for full self-financing). He then devel-
oped some algebra to show how to calculate this premium loading, though the
required return on shareholder capital was taken as an arbitrary parameter that
was dictated by the equity market.59 This work can be seen as a step towards
a more economically coherent perspective on insurance premium setting. The
concept of loading premiums to include a cost of capital has endured.
A 1990 paper by Daykin and Hey,60 two leading general insurance
actuaries that we already met earlier in our general insurance discussion,
discussed how the stochastic asset-liability cashflow modelling that had been
developed by the Solvency Working Group in 1987 for the purposes of sol-
vency assessment could be extended for use in the wider financial manage-
ment of a general insurance firm. The 1987 approach to solvency assessment
had explicitly assumed that the insurer ceased to write new business. The 1990
paper considered how the simulation model could be extended to include the
58
Taylor (1984).
59
Taylor (1984), p.178.
60
Daykin and Hey (1990).
long-term projection of new business and its financial impacts. This naturally
required a modelling framework to describe the behaviour of new business in
terms of business volumes, profitability, competitiveness of pricing relative to
the wider market, and how these variables behave jointly with each other and
with the other stochastic variables and outputs in the model. For example, the
business may price more aggressively when its balance sheet is strong relative
to the required solvency margin. Daykin and Hey produced results for the
ten-year stochastic projection of asset and liability cashflows, profits and bal-
ance sheets under a variety of different assumptions for new business writing
and pricing strategies. Their objective was to show that technical actuarial
skills could be employed as an intrinsic part of business strategy development
and planning in a general insurance firm.
Daykin and Hey also considered the potential use of financial economics
in general insurance business management. Once again, this topic of gen-
eral insurance research had first been explored overseasin this case in the
USA.The particular stimulus for the use of financial economics in US gen-
eral insurance had been in determining the return required by investors for
funding general insurance business. This had relevance beyond the commer-
cial management considerations of a given companyin the USA, premium
rates were controlled by government regulation, and this required an objective
and rigorous framework for establishing a fair premium rate. This, in turn
demanded an assumption about the level of return reasonably required by the
providers of insurance capital.
The application of the Capital Asset Pricing Model to determine required
returns on general insurance equity capital was initially explored in the USA
in the late 1970s and early 1980s.61 This work highlighted that where gen-
eral insurance claims are uncorrelated with market returns, no risk premium
should be required by shareholders for bearing this (diversifiable) risk. Daykin
and Hey expressed discomfort with this implication for general insurance
required returns, arguing that it suggests that the CAPM is missing some
important aspects.62 A natural candidate for these missing aspects was the
frictional cost of capital argument that had been presented by Taylor in 1984,
but this line of thought was not pursued by Daykin and Hey.
Daykin and Hey also briefly considered the potential use of Merton-style
option modelling of an insurers capital structure, where shareholder equity is
modelled as a call option on the net assets of the general insurance company.
See, for example, Biger and Kahane (1978); Hill (1979).

61
Daykin and Hey (1990), p.197.

62
This had also recently been explored in the USA.63 By explicitly capturing the
leverage implied by insurance business (where policyholders are essentially
debtholders), higher required shareholder returns (and hence premium rates)
could be implied, which Daykin and Hey welcomed. They suggested that
their simulation modelling approach could be used to analyse more realistic
forms of optionality in shareholder returns than the simplistic analytical mod-
els presented in the literature.
Daykin and Hey widened the actuarial perspective on general insurance
from solvency to include profitability, return of capital and new business pric-
ing strategy. Later in 1990, a paper by Ryan and Larner was published in the
Journal that widened the actuarial horizons of general insurance further to
include company valuation.64 This paper discussed the application to general
insurance of the appraisal value method that actuaries had recently started
to apply in merger and acquisition work in both life and general insurance
business. The essence of the appraisal value was that it involved an explicit
projection of the cashflow earnings of the business, using assumptions set fol-
lowing actuarial investigation. These cashflows would then be discounted at
appropriate risk discount rates to obtain the companys appraisal value.
Ryan and Larner considered the appraisal value in three components: the
adjusted net asset value (the current balance sheet net asset value, adjusted
for the cost of those assets being locked-in to the insurance balance sheet);
other value arising from past written business (expected release of surplus
from insurance reserves); and the value arising from future written business.
The adjusted net asset value would be based on the market value of assets, and
the deduction from the net asset value for capital lock-in would be based on
the shareholders cost of capital. They explained the rationale for the cost of
capital as follows:
They [shareholders] will require a larger return on their funds [if invested in an
insurance operation] than if they invested them separately. This arises, partly
because the capital is being exposed to the risk of loss in the insurance business
and partly because it could be used elsewhere. We use the term cost of capi-
tal to mean the value of the shortfall in net earnings between the risk return
required by shareholders and the actual [expected] investment return.65
Their cost of capital represented the incremental additional return that

arose from the consequences of investing via an insurance entity rather than
63
Doherty and Garven (1986).
64
Ryan and Larner (1990).
65
Ryan and Larner (1990), pp.6034.
directly in financial markets. Taylor had taken a similar position in his 1984
paper on loading premiums for the cost of capital, though he had argued
that this additional cost arose from frictional sources such as double-taxation
rather than because of exposure to insurance risk.
The risk discount rate would reflect the riskiness of the shareholder cashflow
streams. The authors suggested different risk discount rates should be applied
to different elements of the projected earnings stream to reflect their risk char-
acteristics (for example, a higher discount rate would be applied to earnings
from future business than that used for future profits emerging from existing
business). Whilst the authors were aware of financial economists approaches
to valuation, some of their suggestions were not necessarily consistent with
those ideas. In particular, their suggestion that the risk discount rate should
be a function of the extent to which the business was exposed to diversifiable
insurance risk ran contrary to a fundamental principle of financial economics.
This idea ran throughout the paperfor example, later the paper suggested
that a well-diversified insurance business could command a lower risk dis-
count rate and, hence, higher appraisal value than a less-diversified business.
A 1994 Journal paper by Bride and Lomax66 provided a perspective on the
financial management of general insurance business that was more closely
aligned to financial economics. Their starting point was that the appraisal
value implementation that had been developed by actuaries to support share-
holder valuations in merger and acquisition activity fails to capture the
operational dynamics of non-life insurance business and obscures the issues
surrounding the nature of shareholders, policyholders and other creditors
claims on the assets of the firm.67 However, they did not propose to reject the
entire framework of discounted cashflow projections of the appraisal value
rather, they attempted to show how the risk-adjusted discount rate and cost
of capital adjustments could be rigorously set.
They emphasised that the idea that shareholders did not require a reward
for bearing diversifiable risk was the most fundamental and least controversial
result produced by financial economics. This had implications both for the
setting of the risk discount rate and the cost of capital adjustmentRyan
and Larner had argued that one of the reasons for the cost of capital adjust-
ment was that shareholders required compensation for their exposure to the
(diversifiable) risk of future insurance losses. This was categorically rejected by
Bride and Lomax.
Bride and Lomax (1994).

66
Bride and Lomax (1994), p.363.

67
They also highlighted what they viewed as other actuarial misconceptions

in valuation. In particular, they argued that the choice of asset strategy of a
general insurance company should have no impact on the value of the firm:
an investment in gilts and an investment in equities have equal risk-adjusted
total yields from the perspective of a shareholder with a diversified portfolio
of assets.68
In short, Bride and Lomaxs 1994 paper was the general insurance equiv-
alent of Exley, Mehta and Smiths 1997 defined benefit pension paper. It
attempted to bring actuarial practice on the measurement and management
of shareholder value in general insurance fully in line with core financial eco-
nomics concepts. This was arguably less of a challenge in general insurance
than in defined benefit pensions, with its legacy of 100 years of satisfied actu-
arial service. But the authors still wryly noted in their response to the Staple
Inn discussion that the subject of financial economics continues to raise tem-
peratures in the profession.69
Actuaries in general insurance as well as pensions found it difficult to accept
many of the insights and implications of financial economics. A reflex to reject
them as academic theories based on unrealistic assumptions remained preva-
lent. In opening the discussion, Duffy commented:
It is questionable whether the shareholders, let alone the management, of the

major UK composites and other non-life firms follow the precepts of financial
economic theory.70
The influential general insurance actuary J.P.Ryan expressed his opinion

at Staple Inn in unambiguous terms: The authors use a number of oversim-
plified economic assumptions when applying their model. It is largely these
oversimplified assumptions that give rise to some of the odd results in the
paper.71
Incorporating financial economics into the practices of any British actu-
arial field was no straightforward task in the 1990s. But in general insurance,
as in life assurance and defined benefit pensions, the genie was out of the
bottle. The process of embracing and incorporating the financial economic
insights of the previous half-century into core actuarial thinking and practices
was irrevocably underway.
68
69
70
Duffy, in Discussion, Bride and Lomax (1994), p.421.
71
Ryan, in Discussion, Bride and Lomax (1994), p.429.
Conclusions: Looking Back, Looking Forward
The intellectual foundations of actuarial methodsprobability, statistics,

mortality modelling, the valuation of life-contingent claimswere first laid
in the latter stages of the Scientific Revolution by some of the most esteemed
thinkers of that age such as Pascal, de Witt, Halley, Bernoulli, de Moivre, Bayes
and Laplace. The history of British (and indeed global) actuarial thought,
however, started in earnest in the second half of the eighteenth century with
Richard Price. It was Price who developed the actuarial disciplines of reserv-
ing, assessment of surplus and distribution of bonus, as well as furthering
technical methods for mortality modelling. Prices clarity of thought brought
James Dodsons original conception of a new form of financial institution to
life and ensured it was built on sustainable ground. His actuarial ideas on the
management of with-profit business supported a product that prospered in a
fundamentally similar form to his design for the next 200 years.
Relatively speaking, the economically serene nineteenth century was a
period of incrementalism for actuarial thought. Mortality modelling contin-
ued to develop in areas such as graduation and the development of tables
based on the pooled experience data of life assurers. Actuaries debated differ-
ent approaches to life assurance liability reserving, and there was a growing
recognition that valuations served two distinct purposessolvency assess-
ment and the equitable distribution of surplusand that these purposes
could make use of different valuation methods. British actuaries attempted to
apply the techniques they had developed in life assurance to general insurance
business (particularly fire and employers liability insurance), but without any
discernible impact on industry practices.

DOI10.1007/978-3-319-33183-6
312 Conclusions: Looking Back, Looking Forward
Nineteenth-century actuaries showed relatively little interest in the asset

side of the balance sheet. From William Morgans time onwards, however, life
actuaries were often doubtful of the relevance of market values. At the start
of the twentieth century, the emerging field of defined benefit pensions was
broadly aligned to their life assurance colleagues. Ultra-long-term estimates
of liability valuation parameters, including interest rates, were used alongside
similarly off-market asset valuations to produce stable assessments of pension
fund surplus.
In the late nineteenth century, actuarial thought began to actively con-
sider investment strategy for the assets backing life assurance business. Most
notably, in 1862 Arthur Bailey recognised the advantages that the long-term
illiquid liabilities of life offices provided by enabling investment in highly
illiquid assets that offered additional yields relative to equivalent more widely
traded instruments. This thinking was reflected in the British life office asset
allocations of the late nineteenth centuryin 1890, some 80 % of life office
funds were invested in illiquid asset classes such as mortgages and loans. A sig-
nificant rotation of life office assets out of illiquid asset types and into traded
securities followed in the subsequent decades as liability liquidity became
more transparent and the availability of illiquid assets diminished.
Broader thinking on asset strategy for life business did not emerge until
the First World War shattered the economic climate of Edwardian stability,
resulting in unprecedented volatility in interest rates and inflation. The eco-
nomic volatility of the 1920s was followed by a decade of falling long-term
interest rates. By the mid-1930s, long-term gilt yields were alarmingly close
and sometimes even below the 3 % rate assumed almost universally in British
with-profit premium bases of the time.
This challenging environment encouraged actuaries to support greater lev-
els of investment in equities by both life offices and pension funds. This was
partly because the extraordinary inflation experience of post-First World War
period increased the appetite for real assets that could offer some form of pro-
tection from future inflation shocks. Whilst life offices did not have liabilities
that were guaranteed in real terms, there was a policyholder expectation that
with-profit bonuses would offer some element of inflation protection to their
pay-outs. Similarly, pension fund trustees often aimed to provide pensioners
of defined benefit schemes with discretionary increases in their pension that
offset the impact of inflation on the purchasing power of their pension. At
this time, decades before the launch of index-linked bonds, the equity market
was arguably the only liquid, real asset class available to these institutions. The
potential inadequacy of long-term gilt yields to meet with-profit guarantees
provided some offices with a further rationale to allocate to equities as part
Conclusions: Looking Back, Looking Forward 313
of an increase in investment risk appetite. The 1930s also represents the start
of the period of the cult of equities, and actuaries and their institutions were
not immune to its siren call. Leading actuaries of the era such as H.E.Raynes
were heavily influenced by contemporary economists such as Edgar Lawrence
Smith and John Maynard Keynes, and indeed by the indisputably excellent
performance of equity markets over the preceding decades (especially relative
to long-term bonds). Thus, from the 1920s onwards, the financial market
risks embedded in British life office balance sheets started to materially grow.
The economic volatility of the post-First World War era also further disin-
clined pension actuaries to use market values in assessments of pension fund
surplus and required long-term contribution rates. By the early 1920s a general
actuarial consensus was reached that the historically high levels of prevailing
long-term market bond yields should not be fully reflected in pension fund
valuations. Assets were valued at above market value, though with perhaps a
partial write-down from their book value. Liability discount rates were left
unchanged from their pre-war assumptions. The prevailing market yield was
treated as an aberration that could be assumed away. The consensus was not
unanimous, however. Most notably, S.G.Warner, the senior actuary who was
President of the Institute in the years 19161918, argued in 1921 that assets
should be valued at market value and liabilities should be valued using a dis-
count rate based on the market yield. This is perhaps the earliest actuarial call
for a market-based approach to pension fund valuation, but it went unheeded.
The quarter-century following the Second World War was a period of
strong economic growth and rapid technological change. For actuaries, devel-
opments in computer science and financial economics pointed to an increas-
ingly quantitative environment for their work. More advanced quantitative
concepts were introduced to British actuaries, or indeed developed by British
actuaries, during this period and found varying degrees of interest amongst
the profession. The application of risk theory to general insurance business
found few supporters within the British profession, despite the unrelent-
ing efforts of R.E. Beard during the 1950s. On the other hand, the work
of Redington (1952), Haynes and Kirton (1952) and Anderson and Binns
(1957) introduced a new technical sophistication to thinking on asset-liability
management in life assurance. Duration and immunisation were concepts
introduced and mathematically codified by Redington that would eventually
permanently resonate with the wider financial world. The work of Haynes
and Kirton, and Anderson and Binns tentatively pointed towards ideas such
as dynamic hedging and portfolio insurance that would again become part
of mainstream financial risk management in the decades to follow. Whilst
British life actuaries had historically displayed an ambivalence towards market
values, all of these risk management ideas were based on the objective of man-
aging exposure to changes in market prices. Defined benefit pensions think-
ing, however, remained resolutely detached from market values and indeed
from any explicit framework of risk management thinking.
These post-war decades also saw the development in US academia of
the fundamental theories of financial economicsportfolio theory (1952),
ModiglianiMiller (1958), the Capital Asset Pricing Moody (1964), the
Efficient Markets Hypothesis (1970) and BlackScholesMerton (1973).
ModiglianiMiller pioneered the use of arbitrage in the derivation of finance
theories; the other theories were built to various degrees on contemporary
developments in computer science and quantitative analysis: the expanding
computational capabilities of computers, the mathematisation of finance and
economics, and the rigorous statistical analysis of empirical market data were
the underlying building blocks of this stream of research. The British actuarial
profession was largely oblivious to these ideas in the years following their
emergence and when it did start to read about them, its predominant reaction
was to dismiss them as irrelevant theory.
Nonetheless, it is at least a notable historical curiosity that a number of the
technical ideas that dominated financial economic theory (such as portfolio
diversification, dynamic hedging, portfolio insurance) and financial economic
empirical research (excess volatility and predictability of returns) were at least
heuristically explored by twentieth-century British life actuaries ahead of their
rigorous academic development by professors of financial economics. It is also
notable that this actuarial research output tended to emerge from life actuar-
ies rather than pensions actuaries. Actuarial work in pensions was ultimately
an exercise in long-term financial planning and budgeting; pension actuaries
generally did not focus on the proactive risk management of the financial
security of members accrued pension benefits with the same sense of disci-
pline and purpose that life actuaries, as custodians of policyholder promises,
attempted when measuring and managing financial risks of life offices.
Whilst the British profession of the 1970s displayed a general disdain for
the actuarial application of the ideas of financial economics, it did, however,
attempt its own application of computer science and statistical modelling to
financial risk management. These efforts had natural applications given the
market risk-laden balance sheets that actuaries presided over by this time
full of equities on the asset side and long-term guaranteed returns to life poli-
cyholders or pension fund members on the liability side. In the life sector, the
provision of guarantees in the form of unit-linked business rather than in the
opaque with-profits format made them more transparent to those outside the
actuarial professionthey could not be cloaked in the mystique of actuarial
management. Regulators worried and agitated for the actuarial profession to

provide a reserving solution for these guarantees that was evidently prudent
and objective.
The actuarial solution took ten years to deliver and generated much inter-
nal debate along the way. The 1971 sessional meeting that discussed Sidney
Benjamins guarantee reserving paper provoked such a response that the
President converted the sessional meeting into a private discussion so as to
avoid documenting it. The paper was never published by the Institute. What
was the cause of such controversy? Benjamin proposed a risk-based reserv-
ing approach that explicitly recognised the non-diversifiable nature of the
financial market risk that was embedded in life assurers products and balance
sheets. Up until this time, actuarial reserving worked in a deterministic setting
that assumed the risks written by the office could be largely diversified away.
Some margins for prudence were applied to reserving bases relative to pre-
mium bases, but these were of a small scale relative to Benjamins risk-based
reserve. The Maturity Guarantee Working Partys report of 1980, with its rec-
ommendation of a risk-based reserved assessed using a stochastic simulation
model, was therefore a watershed moment in the history of British actuarial
thought. It also resulted in the ending of the British life office practice of
offering maturity guarantees to unit-linked business.
The working partys risk assessment methodology, however, rejected the
insights available from financial economics option pricing literature. Instead
of reserving for guarantees on the basis that they were put options that could
be transferred to a third party at a market price, the risk methodology con-
sidered the amount of assets required to fund the ultimate losses generated
by the guarantee with a high level of probability, under the explicit assump-
tion that no form of hedging could take place. This method therefore did
not provide any incentive to life offices or their actuaries to consider how to
manage the risks they had written. Moreover, the method was highly sensi-
tive to assumptions about the very long-term behaviour of equity returns.
Here again, the approach rejected ideas from financial economics: instead
of using a random walk-style model for equity returns that was consistent
with market efficiency, the modelling, which was essentially a prototype of
the Wilkie model, assumed equity market returns had a predictable compo-
nent that resulted from market prices being more volatile than was rationally
implied by assumed dividend variability. Interestingly, however, the empirical
research produced by financial economics academia in the 1980s by leading
academics such as Shiller, Roll, Fama and French provided some intellectual
support for this approach (although this strand of research was ignored in
the later traditionalist versus modernist debates within the British actuarial
debates of the 1990s).
The stagflation of late 1970s Britain generated even more existential ques-
tions for actuaries in defined benefit pensions: should pensions be advance
funded in an economic environment of negative real interest rates? The
improving economic conditions of the 1980s made this question moot,
but there remained a philosophical debate about the purpose of funding.
Specifically, was it to produce a stable set of asset cashflows with a similar
profile and character to the liability cashflows; or was the central purpose of
advance funding to provide a current pot of assets with a market value that
was sufficient to secure the accrued pension benefits in the event of sponsor
insolvency? The profession slowly moved from the former towards the latter
over a period of several decades, but never fully reached a consensus. When,
in the early 1990s, the government tried to enforce a funding standard that
would secure accrued benefits in the event of insolvencythe Minimum
Funding Requirement (MFR)the result was a fudge that failed to meet its
stated objective and otherwise created cost and confusion.
Whilst the 1970s and early 1980s was a period of profound challenge for
British life and pensions actuaries (challenges that arose from legacy balance
sheets, difficult economic conditions and the circulation of new ideas that ran
counter to actuarial convention), in contrast this was a period when actuaries
profile and role in British general insurance notably increased. This increased
role capitalised on the accumulating volume and improving accessibility of
relevant historical claims data in sectors such as motor insurance, and on actu-
aries traditional skills in the analysis of claims data for the purposes of pric-
ing and reserving. However, the claim for a greater actuarial role in general
insurance was also strongly based on the argument that statutory reserving in
general insurance needed the same professional judgement that was present
in life assurance reserving.
The 1990s was a remarkably tumultuous and difficult decade for the British
actuarial profession. Again, the economic environment played its part in
exposing the limitations of the financial risk management approaches adopted
in actuarial balance sheets. In particular, long-term interest rates steadily fell
throughout the decade. By the middle of the decade, the fall in rates had
triggered the Guaranteed Annuity Option crisis in the life sector, which
claimed the scalp of Equitable Life, the storied institution of Richard Price
and William Morgan. There were many institutional factors that contributed
to the Equitables downfall, but from the perspective of actuarial thinking,
there was at least one fundamental lesson. The profession chose not to apply
to with-profit guarantees the risk-based reserving methods it developed in the
1970s for unit-linked maturity guarantees. This was largely because actuaries
were confident that the various powerful levers they had at their disposal to
steer with-profit businessesspecifically, wide discretion in bonus policy and

in investment strategywould allow them to avoid engaging in the rigours of
risk-based reserving, market risk management and hedging. The exposure of
GAO business to low interest rates had been fully understood by actuaries for
many yearsDick Gwilt, the Faculty President of 1952 to 1954, had argued
as early as in 1948 against exposing life offices to this form of policyholder
optionality. But the sales and marketing pressure to provide such product
features held sway. And actuaries held on to the belief that with-profit funds
had the means to manage the risks they created. Ultimately, this proved to be
a misplaced confidence. The entire episode substantially damaged the public,
government and regulatory confidence in actuarial judgement and discretion.
These turbulent episodes left the profession embattled as never before.
Painful lessons were learned, however, and by the late 1990s there was a per-
ceptible change in the professions intellectual outlook and its openness to
change. Renewed impetus was given to modernising actuarial thinking that
finally embraced the fundamental ideas of financial economics and attempted
to incorporate them into standard actuarial methods. Across each of the three
main practice areas of general insurance, defined benefit pensions and life
assurance, important papers were published in the 1990s and early 2000s
that, whilst at times stirring controversy and discontent amongst traditional-
ists, paved the way for a more open-minded and technically trained era of
British actuarial thought. In life assurance in particular, albeit at the behest
of an impatient regulator reacting in the aftermath of the GAO debacle, new
market-based, risk-based reserving methods for with-profit business were
implemented by actuaries quickly, successfully and permanently. These meth-
ods were amongst the most sophisticated implemented in any life assurance
market in the world, and were arguably influential in establishing the blue-
print for the European Solvency II framework that was first conceptualised in
the early 2000s.
It is no easy task, from todays vantage point of hindsight, to sit in objective
judgement of this story and the events and characters that shaped it over hun-
dreds of years. Perhaps the best we can do is point to where notable progress
evidently did and not occur, and where particular events and decisions were
clearly consequential.
Since the early nineteenth century, the very long-term nature of life and
pensions business had philosophically inclined actuaries to use stable mea-
sures of cost and value. Market values have, on the most part, been treated
with suspicion, disdain or indifference. There are a couple of possible kinds of
explanations for this: in the case of with-profit life business, it could be argued
that this lack of use of market values merely reflected the way the product
was supposed to work. That is, with-profit bonuses were not supposed to
reflect market value changes; at the heart of the product was the concept
of intergenerational cross-subsidy and smoothing of returns between policy-
holders. There is good evidence to support this: there is much recognition in
the historical actuarial literature of the need for different types of valuation for
different purposes. Classically, a net premium valuation with off-market asset
valuation for the equitable distribution of surplus; and a gross premium valu-
ation with market asset values for the assessment of solvency. There is a second
kind of explanation: that actuaries held the belief that market values were
often wrong; that they were too volatile and the changes in market value
were often irrational and irrelevant to the assessment of the expected cost of
funding long-term liability cashflows. Pension fund valuation methods have
arguably been based on this perspective. And much of modern actuarial risk
methodology (such as the Wilkie model) has also been consistent with this
outlook. It also arguably was a factor in the increasing appetite for market risk
on the asset side of life and pensions balance sheets over the twentieth century.
Whilst such an outlook was less inconsistent with financial economics
research than actuaries were often led to believe, it nonetheless inhibited the
professions willingness to embrace new ideas and techniques that could man-
age market risk over shorter-term horizons. In hindsight this was unfortunate,
as governments and regulators became increasingly interested in solvency
measures that were based on the short-term resilience of the balance sheets of
long-term business on a market value basis (in both life and pensions). When
reviewing the twentieth-century actuarial literature, it is quite striking how
a rich stream of market value-based risk thinking emerged from 1950s life
actuaries which was not successfully built upon in the actuarial research of the
1960s and 1970s. One can only speculate whether it was that philosophical
actuarial disdain for managing short-term market value solvency risks that lay
behind this.
To quote the clich, history does not repeat, it rhymes. It also never ends.
This historical account closed around the start of the twenty-first century.
Does it point to some important trends for the future of British actuarial
thought? And how has the extraordinary decade of financial turmoil that fol-
lowed the end of this historical account impacted on the future direction of
actuarial thought?
The global financial crisis of 20072008 systemically challenged financial
institutions (and their regulators) in a way that had not occurred since the
1930s. This was, first and foremost, a global banking crisis. Widening credit
spreads and rising doubts about the quality of the opaque assets that played
an increasingly prominent role on bank balance sheets resulted in a banking
solvency and liquidity crisis that started in the USA and quickly reverberated
around the globe. With the notable exception of AIG (with its notorious
financial products division that was far removed from conventional insurance
business), the financial institutions that relied on actuarial advice were not in
the front line of the crisis and they generally weathered its immediate impact
successfully. Insurance companies and pension funds had equity, credit and
other risk asset investments which performed poorly over the period of the
crisis, but there was no widespread immediate set of failures in these institu-
tions that resulted from the unfolding of the crisis over 2007 and 2008. This
may have been a source of some satisfaction and vindication for actuaries
around the world: if they could weather a once-in-a-century financial storm
such as this, clearly they were doing something right.
But the full picture is more complex than this: the long-term nature of life
and pension liabilities means that they are exposed not only to the short-term
gyrations of financial market corrections but also to the longer-term economic
consequences that a financial crisis such as this may trigger. In the years fol-
lowing the financial crash of 1929, long-term UK government bond yields fell
from 4.6 % to below 3 %. The years following the financial crisis of 20072008
produced a similar pattern. We saw in the Guaranteed Annuity Options dis-
cussion that long-term government bond yields fell from 15 % to 5 % over the
three decades between 1975 and 2005. One of the economic consequences of
the global financial crisis of 20072008 has been to further extend this trend:
by 2015, British government long bond yields had fallen below 3 %, to levels
not seen since the Dalton era at the end of the Second World War.
For UK life assurers, whilst unwelcome, this fall in rates did not present an
existential solvency threat: the regulatory changes of the early 2000s, which
were directly motivated by the GAO crisis of the late 1990s and the industrys
exposure to further interest rate falls that it highlighted, had strongly encour-
aged UK life offices to reduce the financial risk exposures in their back books
and to stop writing new business with significant financial guarantees (at least
without pricing it on a market basis and managing its risk actively). The same
conclusion could not, however, be applied to UK defined benefit pension
funds, many of which were left with very substantial levels of funding deficit
on a market value basis by the post-crisis falls in long-term rates. At the time
of writing it remains to be seen how the cost of meeting these deficits would
be shared amongst corporate sponsors (via additional deficit-funding contri-
butions), the pension fund members (through reductions in future pension
payments) and the government (to the extent that the Pension Protection
Fund cannot be adequately financed via pension fund levies). And the pos-
sibility remains that a substantial rise in long-term nominal and real interest
rates could yet reverse these deficits.
These falls in long-term interest rates between 2008 and 2015 have not been
a UK-specific phenomenon, but have been the experience of most developed
economies around the world. Indeed, long-term interest rates in Europe have
fallen even more precipitously than in the UK: by 2015 German long-term
bond yields had fallen below 2 %. Unlike their UK counterparts, regulators
in northern European countries such as Germany did not put a market-based,
risk-based regulatory solvency system in place for insurers prior to the global
financial crisis and the insurers generally did not put financial risk manage-
ment strategies in place that would protect them from the asset-liability con-
sequences of falls in the long-term interest rate to the levels of 2015. Some of
these institutions may struggle to meet their long-term contractual liabilities
from their existing reserves if interest rates do not revert to something like
their pre-crisis levels in the coming years.
Since around the start of the twenty-first century, the European Union
has been working on the implementation of a harmonised regulatory sol-
vency system for insurers that is market-based, risk-based and principle-based
(and which has similar fundamental features to the system implemented
very quickly by the UK regulator in the early 2000s). This regulatory system,
known as Solvency II, was finally implemented at the start of 2016 after more
than a decade of political horse-trading. The result is arguably rather remi-
niscent of the British attempt at improved pension fund supervision in the
1990s. The Minimum Funding Requirement was a regulatory system driven
by an ambition to provide a high standard of protection to the beneficiary
based on short-term market-based solvency measures and the cost of transfer-
ring liabilities to a third party; but it was fudged in implementation so as to
avoid the economic reality of the scale of capital that that protection implied.
The MFR ultimately produced, at significant cost, numbers that were not
particularly useful or meaningful, and that certainly did not have the meaning
they were originally intended to have. This is arguably what Solvency II looks
like in parts of the life assurance sector of northern Europe.
One might argue that it is unfair to ask actuaries to have anticipated these
unprecedented economic conditions. Moreover, since the global financial cri-
sis, government and central bank policy across the developed world has been
to regard the impact of extraordinary low interest rates on long-term savings
institutions such as life assurers and pension funds as mere collateral damage
that is necessarily incurred in efforts to protect banks and the wider economy.
This policy places insurance and pension institutions in a strong headwind
that is not of their own making. But one might also look at the interminable
decline in long-term interest rates since the mid-1970s and ask at what point
it would be reasonable to view further significant falls as something that ought
to be considered and proactively managed against. The UK life offices showed

that this could be done (albeit at the impatient behest of their regulator).
Their proactive management of interest rate risk at the start of the twenty-first
century probably preserved the future solvency of several financial institutions
and could perhaps be regarded as a highlight of post-war British actuarial
achievement.
What might this all mean for the future of actuarial thought? Thought-
leadership will continue to be driven to a large extent by the role that actuaries
play in financial institutions. In the UK, the gradual but irreversible demise of
with-profits life assurance and defined benefit pensions makes this future role
more uncertain than it has been at perhaps any point in the last 200 years.
Looking ahead, there are many possibilities and few inevitabilities. Whilst tra-
ditional with-profit business is no longer written in any meaningful volume
in Britain and increasingly less is written in continental Europe, it has been
selling successfully over the last decade in fast-growing life assurance markets
such as China and India. Traditional life actuarial skills may be in ongoing
demand in these increasingly consequential markets. The British profession
is enthused by the prospect of actuaries contributing in new fields such as
climate change that may make use of actuaries quantitative skills and their
experience in applying them to long-term complex socio-economic problems.
This is quite plausible but it still remains to be seen if professional actuarial
skills can make an impactful contribution in such fields.
Perhaps there is a further positive possibility: British life actuaries suc-
cess in implementing a principle-based reserving system in the early 2000s
prompts echoes of Sidney Benjamins historical call for the application of
actuarial professional judgement to provide an independent, objective, statu-
tory and expert role in assessing solvency and setting prudential reserving
levels. In the 1970s Benjamin tirelessly argued that solvency assessment ought
to go beyond the application of exploitable regulatory rules and should work
outside the direct influence of the institutions managers and shareholders. He
was speaking of extending the classical freedom with publicity role of British
life actuaries to include general insurance as well as life business. This has now
largely happened. The experience of the global financial crisis perhaps high-
lights that such a role could also be valuable across the wider financial sector.
As any good actuary will tell you, trying to forecast the future is even harder
than trying to explain the past. Future long-term socio-economic problems
are unlikely to be any less complex and consequential than historical ones.
Demographics, pensions, healthcare and insurance all have the potential to
pose profound long-term societal challenges. Perhaps, like the work of James
Dodson in the mid-eighteenth century, new conceptions of risk-sharing can
be developed by actuaries that create new products, new solutions, and even
new industries to tackle these challenges. For new ideas to have longevity,
they require a robust implementation. The best actuarial worksuch as that
of Richard Price and Frank Redingtonhas often embraced more theory and
technical complexity than the broader profession has been comfortable with at
the time, but has been presented in a way that demonstrates it can be applied
in accessible, practical, enlightening and highly consequential ways. This is
likely to remain true of the most important actuarial thought-leadership of
the next century. I look forward to seeing it.
Bibliography
Abbott, W.M. et al (1974), Some Thoughts on Technical Reserves and Statutory

Returns in General Insurance. Journal of the Institute of Actuaries, Vol. 101,
pp.217283.
Abbott, W.M. etal (1981), Some Financial Aspects of a General Insurance Company.
Journal of the Institute of Actuaries, Vol. 108, pp.119209.
Anderson, J.L. (1944), Notes on the Effect of Changes in Rates of Interest on the
Bonus-Earning Power of an Office Paying a Uniform Compound Reversionary
Bonus, Transactions of the Faculty of Actuaries, Vol. 17, No. 162 (19381945),
pp.137173.
Anderson, J.L. and J.D.Binns (1957), The Actuarial Management of a Life Office,
Journal of the Institute of Actuaries, Vol. 83, No. 2, pp.112152.
Andras, H.W. (1896), On the system of bonus distribution to policyholders as a
percentage perannum for the Valuation period on the sum assured, or on the sum
assured and existing bonuses, considered in relation to some recent influences on
Life Assurance Finance, Journal of the Institute of Actuaries, Vol. 32, No. 5 (April
1896), pp.320371.
Ansell, C. et al (1843), Tables Exhibiting the Law of Mortality Deduced from the
Combined Experience of Seventeen Life Assurance Offices.
Arthur, T.G. and P.A. Randall (1990), Actuaries, Pension Funds and Investment,
Bachelier, L. (1900), Theorie de la speculation, Annales Scientifiques de lcole
Normale Suprieure 3 (17), pp.2186.
Bailey, A.H. (1862), On the Principles on which the Funds of Life Assurance
Societies should be Invested, Journal of the Institute of Actuaries, Vol. 10,
pp.142147.

DOI10.1007/978-3-319-33183-6
324Bibliography
Bailey, A.H. (1878), The Pure Premium Method of Valuation, Journal of the Institute
of Actuaries, Vol. 21, No. 2 (July 1878), pp.115136.
Barnwell, R.G. and F.Hendricks (1852), A Sketch of the Life and Times of John De
Witt, Grand Pensionary of Holland, To which is added, His Treatise on Life Annuities.
(Note De Witts original treatise was published in 1671.) 2012 re-print published
by General Books LLC.
Bayes, T. (1764), An Essay Towards Solving a Problem in the Doctrine of Chances,
Philosophical Transactions of the Royal Society of London, 53.
Bayley, G.V. and W.Perks (1953), A Consistent System of Investment and Bonus
Distribution for a Life Office, Journal of the Institute of Actuaries, 79, pp.1473.
Beard, R.E. (1954), Some Statistical Aspects of Non-Life Insurance, Journal of the
Institute of Actuaries Students Society, Vol. 13, pp.139157.
Beard, R.E. (1967), On the Compilation of Non-Life Insurance Statistics, Journal
of the Institute of Actuaries, Vol. 93, No. 2, pp.271277.
Beard, R.E., T. Pentikainen and E. Pesonen (1969), Risk Theory: The Stochastic
Basis of Insurance, Chapman and Hall.
Benjamin, S. (1976), Maturity Quarantees for Equity-Linked Policies, 20th
International Congress of Actuaries.
Benjamin, S. (1976), Profit and Other Financial Concepts in Insurance, Journal of
the Institute of Actuaries, Vol. 103, No. 3, pp.233305.
Benjamin, S. and Eagles (1986), Reserves in Lloyds and the London Market,
Benz, N. (1960), Some Notes on Bonus Distributions by Life Offices, Journal of the
Institute of Actuaries, Vol. 86, No. 1, pp.129.
Bernoulli, Daniel (1738), Specimen theoriae novae de mensura sortis, Commentarii
academiae scientarum imperialis Petropolitanae 5.
Bernoulli, Jacob (1713), Ars Conjectandi, Basle.
Biger, N. and Kahane, Y. (1978), Risk Considerations in Insurance Ratemaking,
Journal of Risk and Insurance, Vol. 45, pp.121132.
Black, F., T.Derman and W.Toy (1990), A One-Factor Model of Interest Rates and
its Application to Treasury Bond Options, Financial Analysts Journal, January/
February 1990, pp.2432.
Black, F. and P.Karasinski (1991), Bond and Option Pricing when Short Rates are
Lognormal, Financial Analysts Journal (July/August 1991).
Black, F. and M.Scholes (1973), The Pricing of Options and Corporate Liabilities,
Journal of Political Economy, Vol. 81, No. 3 (MayJune 1973), pp.637659.
Boden, D.E. and T.D.Kingston (1979), The Effect of Inflation on Pension Schemes
and Their Funding, Transactions of the Faculty of Actuaries, Vol. 36, No. 256,
pp.399468.
Bolton, M.J. etal (1997), Reserving for Maturity Quarantees, Report of the Annuity
Guarantees Working Party.
Boole, George (1854), An Investigation of the Laws of Thought. Watchmaker
Publishing.
Bibliography 325
Bornhuetter, R.L. and R.E.Ferguson (1972), The Actuary and IBNR, Proceedings of
the Casualty Actuarial Society, Vol. 59, pp.181195.
Box, G.E.P. and Jenkins, G.M. (1970), Time Series Analysis: Forecasting and Control.
Holden-Day.
Boyle, P.P. (1978), Immunization Under Stochastic Models of the Term Structure,
Journal of the Institute of Actuaries, Vol. 105, No.2, pp.177187.
Boyle, P.P. and Schwartz, E.S. (1977), Equilibrium Prices of Guarantees Under
Equity-Linked Contracts , The Journal of Risk and Insurance, XLIV, 4, pp.639660.
Boyle, P.P. and M.Hardy (1978), Guaranteed Annuity Options, Astin Bulletin, Vol.
33, No. 2, pp.125152.
Brennan, M.J. and E.S. Schwartz (1976), The Pricing of Equity-Linked Life
Insurance Policies with an Asset Value Guarantee, Journal of Financial Economics,
Vol. 3, pp.195213.
Brennan, M.J. and E.S. Schwartz (1979), A Continuous Time Approach to the
Pricing of Bonds, Journal of Banking and Finance, Vol. 3 (1979), pp.133155.
Brennan, M.J., and E.S.Schwartz (1979; 2), Alternative Investment Strategies for
the Issuers of Equity Linked Life Insurance Policies with an Asset Value Guarantee,
Journal of Business, Vol. 52, pp.6393.
Brennan, M.J. and E.S.Schwartz (1982), An Equilibrium Model of Bond Pricing
and a Test of Market Efficiency, Journal of Financial and Quantitative Analysis,
Vol. 17, pp.301329.
Bride, M., and Lomax, M.W. (1994), Valuation and Corporate Management in a
Non-Life Insurance Company, Journal of the Institute of Actuaries, Vol. 121, No.
2, pp.363440.
Brown, Hugh (1931), Employers Liability Insurance, Transactions of the Faculty of
Actuaries, Vol. 13, No. 115, pp.166.
Brown, Samuel (1851), On the Fires in London During the 17 Years from 1833 to
1849 Inclusive, Showing the Numbers which Occurred in Different Trades, and
the Principle Causes by which they were Occasioned, The Assurance Magazine,
Vol. 1, No. 2, pp.3162.
Brown, Samuel (1858), On the Investments of the Funds of Assurance Companies,
Journal of the Institute of Actuaries, Vol. 7, No. 5 (April 1858), pp.241254.
Cardano, Gerolamo (1663), Liber de ludo alae. In Opera Omnia, Stuttgart-Bad
Cannstatt, 1966.
Clarkson, R.S. (1997), An Actuarial Theory of Option Pricing, British Actuarial
Journal, Vol. 3, No. 2, pp.321409.
Clayton, George and W.T.Osborn (1965), Insurance Company Investment. George
Allen & Unwin.
Cochrane, John (2005), Asset Pricing, Revised Edition. Princeton University Press.
Colbran, R.B. (1982), Valuation of Final Salary Pension Schemes, Journal of the
326Bibliography
Collins, T.P. (1982), An Exploration of the Immunization Approach to Provision for

Unit-Linked Policies with Guarantees, Journal of the Institute of Actuaries, Vol.
109, No. 2, pp.241284.
Corby, F.B. (1977), Reserves for Maturity Guarantees Under Unit-Linked Policies,
Journal of Institute of Actuaries, Vol. 104, pp.259296.
Corley, R.D. et al (2001), Report of the Corley Committee of Inquiry regarding the
Equitable Life Assurance Society.
Cowles, A. (1933) Can Stock Markets Forecasters Forecast? Econometrica, Vol. 1,
No.3 (July 1933), pp.309324.
Cox, J.C., J.E.Ingersoll and S.A.Ross (1985), An Intertemporal General Equilibrium
Model of Asset Prices, Econometrica, Vol. 53, No. 2, pp.363384.
Cox, J.C., J.E.Ingersoll and S.A.Ross (1985), A Theory of the Term Structure of
Interest Rates, Econometrica, Vol. 53, No. 2, pp.385408.
Cox, J.C. and S.A.Ross (1976), The Valuation of Options for Alternative Stochastic
Processes, Journal of Financial Economics, Vol. 3, pp.145166.
Cox, J.C., S.A. Ross and M. Rubinstein (1979), Option Pricing: A Simplified
Approach, Journal of Financial Economics, Vol. 7, No. 3.
Cox, P.R. and R.H.Storr-Best (1962a), Surplus: Two Hundred Years of Actuarial
Advance, Transactions of the Faculty of Actuaries, 28 (19621964), pp.1960.
Cox, P.R. and R.H.Storr-Best (1962b), Surplus in British Life Assurance: Actuarial
Control over its Emergence and Distribution during 200 years. Cambridge University
Press.
Coutts, C.R.V. (1908), Bonus Reserve Valuations, Journal of the Institute of Actuaries,
Vol. 42, No. 2 (April 1908), pp.161177.
Crabbe, R.J.W. and C.A.Poyser (1953), Pension and Widows and Orphans Funds.
Cambridge University Press.
Craighead, D.H. (1979, Some Aspects of the London Reinsurance Market in World-
Wide Short-Term Business, Journal of the Institute of Actuaries, Vol. 106, No. 3,
pp.227287.
Daston, Lorraine (1988), Classical Probability in the Enlightenment. Princeton
University Press.
Davis, Mark and Alison Etheridge (2006), Louis Bacheliers Theory of Speculation; The
Origins of Modern Finance. Princeton University Press.
Day, J.G. (1959), Developments in Investment Policy During the Last Decade,
Day, J.G. and K.M. McKelvey (1963), The Treatment of Assets in the Actuarial
Valuations of a Pension Fund, Journal of the Institute of Actuaries, Vol. 90,
pp.104147.
Daykin, C.D. etal., (1984), The Solvency of General Insurance Companies, Journal
of the Institute of Actuaries, Vol. 111, pp.279336.
Daykin, C.D. et al., (1987), Assessing the Solvency and Financial Strength of a
General Insurance Company, Journal of the Institute of Actuaries, Vol. 114,
pp.227325.
Bibliography 327
Daykin, C.D. and G.B.Hey (1990), Managing Uncertainty in a General Insurance

Company, Journal of the Institute of Actuaries, Vol. 117, pp.173277.
Deane, Phyllis and W.A.Cole (1967), British Economic Growth, 16881959, Second
Edition. Cambridge University Press.
De Jong, P. and B.Zehnwirth (1983), Claim Reserving, State-Space Models and the
Kalman Filter, Journal of the Institute of Actuaries, Vol. 110, No. 1, pp.157181.
De Moivre, Abraham (1718), Doctrine of Chances.
De Moivre, Abraham (1724), Annuities on Lives.
De Morgan, Augustus (1838), An Essay on Probabilities and On Their Applications to
Life Contingencies and Insurance Offices. Longman, Orme, Brown, Green &
Longmans.
De Morgan, Augustus (1839), Mortality, Penny Cylopedia.
De Wittt, Johan (1671), In A Series of Letters to the States-General, In R.G.Barnwell
and F.Hendricks, A Sketch of the Life and Times of John De Witt, Grand Pensionary
of Holland. NewYork, 1856.
Deuchar, David (1890), The Progress of Life Assurance Business in the United
Kingdom during the last Fifty Years, Journal of the Institute of Actuaries, Vol. 28,
No. 6 (October 1890), pp.442463.
Dimson, E., P.R.Marsh and M.Staunton (2002), Triumph of the Optimists: 101 Years
of Global Investment Returns. Princeton University Press.
Dodds, J.C. (1979), The Investment Behaviour of British Life Insurance Companies.
Croom Helm.
Dodson, James (1756), First Lectures in Insurance.
Doherty, N.A. and J.R. Garven (1986), Price Regulation in Property-Liability
Insurance: A Contingent Claims Approach, Journal of Finance, Vol. 41, No. 5,
pp.10311050.
Dullaway, D. and P. Needleman (2004), Realistic Liabilities and Risk Capital
Margins for With-Profits Business, British Actuarial Journal, Vol. 10, No. 2,
pp.185222.
Dyson, A.C.L. and C.J.Exley (1995), Pension Fund Asset Valuation and Investment,
British Actuarial Journal, Vol. 1, No. 3, pp.471557.
Edmonds, T.R. (1832), Life Tables, Founded upon the Discovery of a Numerical Law
Regulating the Existence of Every Human Being. James Duncan.
Edmonds, T.R. (183738). On the Influence of Age and Selection on the Mortality
of the Members of the Equitable Life Insurance Society During a Period of Sixty-
Seven Years, Ending in 1829, The Lancet, Vol. 1, No. 739 (183738), pp.154162.
Elderton, W.Palin (1934), An Approximate Law of Survivorship and Other Notes
on the use of Frequency Curves in Actuarial Statistics, Journal of the Institute of
Actuaries, Vol. 65, No. 1 (March 1934), pp.136.
Epps, G.S.W. (1921), Superannuation Funds. Notes on Some Post-War Problems,
Together with an Account of a Pensioners Mortality Experience (Civil Service
Pensioners, 19041914), Journal of the Institute of Actuaries, Vol. 52, No. 4,
pp.405453.
328Bibliography
Euler, Leonhard (1749), Recherches sur la question des inegalites du movement de

Saturne et Jupiter, sujet propose pour le prix de lannee 1748, par lAcademie royal
des sciences de Paris.
Exley, C.J., S.J.B.Mehta and A.D.Smith (1997), The Financial Theory of Defined
Benefit Pension Schemes, British Actuarial Journal, Vol. 3, No. 4, pp.835966.
Fama, E. (1965), The Behaviour of Stock Market Prices, Journal of Business, Vol. 38,
No. 1 (January 1965), pp.34105.
Fama, E. (1970), Efficient Capital Markets: A Review of Theory and Empirical
Work, Journal of Finance, Vol. 25, No. 2 (May 1970), pp.383417.
Fama, E. and M. Blume (1966), Filter Rules and Stock Market Trading Profits,
Journal of Business, Vol. 30 (Special Supplement, January 1966), pp.226241.
Fama, E. and K.French (1988a), Permanent and Temporary Components of Stock
Prices, Journal of Political Economy, Vol. 96, pp.246273.
Fama, E. and K. French (1988b), Dividend Yields and Expected Stock Returns,
Journal of Financial Economics, Vol. 22, pp.325.
Financial Services Authority (2002), Consultation Paper 143.
Financial Services Authority (2003), Consultation Paper 195.
Finlaison, John (1829), Report of John Finlaison, Actuary of the National Debt, on the
Evidence and Elementary Facts on which the Tables of Life Annuities are founded.
Fisher, Irving (1896), Appreciation and Interest, Publications of the American
Economic Association (1896).
Fisher, Irving (1930), The Theory of Interest. Macmillan.
Fisher, Sir Ronald (1956), Statistical Methods and Scientific Inference. Oliver and
Boyd.
Ford, A. etal (1980), Report of the Maturity Guarantees Working Party, Journal of
Francis, John (1853), Annals, Anecdotes and Legends; A Chronicle of Life Assurance.
Longman, Brown, Green & Longmans.
Galloway, Thomas (1841), Tables of Mortality Deduced from the Experience of the
Amicable Society for a Perpetual Insurance Office.
Gauss, Carl Friedrich (1809), Theoria motus corporum celestium.
Geoghegan, T.J. etal (1992), Report on the Wilkie Stochastic Investment Model,
Journal of the Institute of Actuaries, Vol. 119, pp.173228.
Geske, R. (1979), The Valuation of Compound Options, Journal of Financial
Economics, Vol. 7, No. 1 (March 1979), pp.6381.
Gilley, D.F. (1972), The Dissolution of a Pension Fund, Journal of the Institute of
Gilley, D.F. and D.Funnell (1958), Valuation of Pension Fund Assets, Journal of the
Institute of Actuaries Students Society, Vol. 15, No. 1, pp.4368.
Gompertz, Benjamin (1825), On the Nature of the Function Expressive of the Law of
Human Mortality, and on a New Mode of Determining the Value of Life Contingencies.
Bibliography 329
Gompertz, Benjamin (1871), On one Uniform Law of Mortality from Birth to

extreme Old Age, and on the Law of Sickness, Journal of the Institute of Actuaries,
Vol. 16, No. 5 (October 1871), pp.329344.
Graunt, John (1662), Natural and Political Observations Mentioned in a Follow Index,
and Made Upon the Bills of Mortality by John Graunt, Citizen of London.
Greenwood, P.M. and T.W.Keogh (1997), Pension Funding and Expensing in the
Minimum Funding Requirement Environment, British Actuarial Journal, Vol. 3,
No. 3, pp.497582.
Hacking, Ian (1965), Logic of Statistical Inference. Cambridge University Press.
Hacking, Ian (1975), The Emergence of Probability. Cambridge University Press.
Hairs, C.J. etal (2002), Fair Valuation of Liabilities, British Actuarial Journal, Vol.
8, No. 2, pp.203340.
Halley, Edmond (1693), An Estimate of the Degrees of the Mortality of Mankind, drawn
from curious Tables of the Births and Funerals at the City of Breslau, with an Attempt
to ascertain the Price of Annuities upon Lives.
Hare, D.J.P. et al (2000), A Market-Based Approach to Pricing With-Profit
Guarantees, British Actuarial Journal, Vol. 6, I, pp.143213.
Harrison, M. and D.M.Kreps (1979), Martingales and the Arbitrage in Multiperiod
Securities Markets, Journal of Economic Theory, Vol. 20, pp.381408.
Harrison, M. and S.R. Pliska (1981), Martingales and Stochastic Integrals in the
Theory of Continuous Trading, Stochastic Processes and Their Applications, Vol. 11,
pp.215260.
Haynes, A.T. and R.J. Kirton (1952), The Financial Structure of a Life Office,
Transactions of the Faculty of Actuaries, Vol. 21, No. 178, pp.141218.
Heath, D., R.A.Jarrow and A.Morton (1992), Bond Pricing and the Term Structure
of Interest Rates: A New Methodology for Contingent Claims Valuation,
Econometrica, Vol. 60, No. 1 (January 1992), pp.77105.
Heywood, G. and M. Lander (1961), Pension Fund Valuations in Modern
Conditions, Journal of the Institute of Actuaries, Vol. 87, No. 3, pp.314370.
Hibbert, A.J. and C.J. Turnbull (2003), Measuring and Managing the Economic
Risks and Costs of With-Profit Business, British Actuarial Journal, Vol. 9, No. 4,
pp.725786.
Hicks, J. (1939), Value and Capital. Oxford University Press.
Hill, R.D. (1979), Profit Regulation in Property-Liability Insurance, Bell Journal of
Economics, Vol. 10, pp.172191.
Holbrook, J.P. (1977), Investment Performance of Pension Funds, Journal of the
Homans, Sheppard (1863), On the Equitable Distribution of Surplus, Journal of the
Institute of Actuaries, Vol. 11, No. 3 (October 1863), pp.121129.
Homer, Sidney and Richard Sylla (1996), A History of Interest Rates, Third Edition
Revised. Rutgers University Press.
Hooker, N.D. et al (1996), Risk-Based Capital in General Insurance, British
Actuarial Journal, Vol. 2, No. 2, pp.265323.
330Bibliography
Hull, J.C. and A.D. White (1990), Pricing Interest Rate Derivatives, Review of
Financial Studies, Vol. 3, No. 4, pp.573592.
Hume, David (1739), A Treatise of Human Nature, Being an Attempt to Introduce the
Experimental Method of Reasoning into Moral Subjects.
Huygens, C. (1657), Ratiociniis in aleae ludo.
Institute of Actuaries (1869), The Mortality Experience of Life Assurance Companies.
Charles and Edwin Layton.
Jellicoe, C. (1851), On the Determination and Division of Surplus, and on the
Modes of Returning it to the Contributors, The Assurance Magazine, Vol. 1, No.
2, pp.2228.
Jellicoe, C. (1852), On the Conditions which give rise to Surplus in Life Assurance
Companies, and on the Amount of the Return, or Bonus, which such Conditions
justify, The Assurance Magazine, Vol. 2, No. 4, pp.333341.
Jensen, M. (1978), Some Anomalous Evidence Regarding Market Efficiency, Journal
of Financial Economics, Vol. 6, No. 2 (1978), pp.95101.
Jensen, M. and W.H.Meckling (1976), Theory of the Firm: Managerial Behaviour,
Agency Costs and Ownership Structure, Journal of Financial Economics, Vol. 3,
No. 4, pp.305360.
Johnson, P.D. and G.B.Hey (1971), Statistical Studies in Motor Insurance, Journal
of the Institute of Actuaries, Vol. 97, No. 2/3, pp.199249.
Kendall, M. (1953) The Analysis of Economic Time-Series, Part I: Prices, Journal of
the Royal Statistical Society (Series A, General), Vol. 116, No. 1 (1953), pp.1134.
Kennedy, S.P.L. etal (1976), Bonus Distribution with High Equity Backing, Journal
of the Institute of Actuaries, Vol. 103, pp.1158.
Keynes, J.M. (1936), The General Theory of Employment, Interest and Money. Palgrave
Macmillan.
Keynes, J.M. (1983), edited by D.Moggridge. The collected writings of John Maynard
Keynes, Volume XII. Cambridge University Press.
King, G. (1905), On Staff Pension Funds, Journal of the Institute of Actuaries, Vol.
39, No. 2, pp.129206.
Knight, Roger (2013), Britain Against Napoleon. Allen Lane.
Laplace, Pierre Simon (1774), Memoire sur la probabilite des causes par les evenemens.
Laplace, Pierre Simon (1781), Memoire sur les probabilities.
Laplace, Pierre Simon (1810), Memoire sur les approximations des formules qui sont
fonctions de tres grand nombres et sur leur application aux probabilities.
Leeson, Francis (1968), A Guide to the Records of the British State Tontines and Life
Annuities of the 17th and 18th Centuries. Pinhorns.
Legendre, Adrien Marie (1805), Nouvelles methods pour la determination des orbites
des cometes.
Lintner, J.V. (1965), The Valuation of Risk Assets and the Selection of Risky
Investments in Stock Portfolios and Capital Budgets, Review of Economics and
Statistics, Vol. 47, No. 1 (February 1965), pp.1337.
Lochhead, R.K. (1932), Valuation and Surplus. Cambridge University Press.
Bibliography 331
Lundberg, F. (1909), Zur Theorie der Ruckversicherung, Transactions of the

International Congress of Actuaries.
Macaulay, Frederick R. (1938), The Movements of Interest Rates, Bond Yields and Stock
Prices in the United States Since 1856. Reprinted by Risk Books.
Macdonald, R.E. (1977), Presidential Address, Transactions of the Faculty of Actuaries,
Vol. 36, No. 253, pp.116.
Mackenzie, A.G. (1891), On the Practice and Powers of Assurance Companies in
regard to the Investment of their Life Assurance Funds, Journal of Institute of
Maclean, J.B. (1948), Some Recent Actuarial Developments in the United States of
America, Transactions of the Faculty of Actuaries, Vol. 18, No. 169, pp.281321.
Makeham, William Matthew (1859), On the Law of Mortality and the Construction
of Annuity Tables, Journal of the Institute of Actuaries, Vol. 8, pp.301310.
Makeham, William Matthew (1867), On the Law of Mortality, Journal of the
Institute of Actuaries, 13, pp.325358.
Makeham, William Matthew (1870), On the Objections to the Net-Premium Mode
of Valuation, Journal of the Institute of Actuaries, Vol. 15, No. 6 (July 1870),
pp.449452.
Makeham, William Matthew (1889), On the Further Development of Gompertzs
Law, Journal of the Institute of Actuaries, Vol. 28, No. 2 (October 1889),
pp.152159.
Manly, H.W. (1911), On Staff Pension Funds: The Progress of the Accumulation of
the Funds; The Identity of a Valuation with the Future Progress of a Fund; The
Manner of dealing with the Funds which are insolvent; and Sundry Observations,
Markowitz, Harry (1952), Portfolio Selection, Journal of Finance, Vol. 7, No. 1
(March 1952).
Marsh, T. and R.C.Merton (1986), Dividend Variability and Variance Bound Tests
for the Rationality of Stock Market Prices, American Economic Review, Vol. 76,
No. 3, pp.483398.
Maudling, R.G. (1929), On the Classification and Duration of Compensation
Claims in the Mining Industry, Vol. 60, No. 3, pp.251296.
May, G.E. (1912), The Investment of Life Assurance Funds, Journal of the Institute
of Actuaries, Vol. 46, No. 2 (April 1912), pp.134168.
McKelvey, K.M. (1957), Pension Fund Finance, Transactions of the Faculty of
McLeish, D.J.D. (1983), A Financial Framework for Pension Funds, Transactions of
the Faculty of Actuaries, Vol. 38, pp.267314.
McLeish, D.J.D. and C.M. Stewart (1987), Objectives and Methods of Funding
Defined Benefit Pension Schemes, Journal of the Institute of Actuaries, Vol. 114,
No. 2, pp.155225.
332Bibliography
Meech, Levi W. (1881), System and Tables of Life Insurance: A Treatise Developed from
the Experience and Records of Thirty American Life Offices Under the Direction of a
Committee of Actuaries.
Merton, R.C. (1973), Theory of Rational Option Pricing, Bell Journal of Economics
and Management, Vol. 4, No. 1 (Spring 1973), pp.141183.
Merton, R.C. (1974), On the Pricing of Corporate Debt: The Risk Structure of
Interest Rates, Journal of Finance, Vol. 29, No. 2 (May 1974), pp.449470.
Merton, R.C. (1990), Continuous-Time Finance. Blackwell.
Miller, M. (1977), Debt and Taxes, Journal of Finance, May 1977, pp.261275.
Miller, Merton and Franco Modigliani (1961) Dividend Policy, Growth, and the
Valuation of Shares, Journal of Business, Vol. 34, No. 4, pp.411433.
Miller, T. (1857), Some Suggestions respecting Fire Insurance Statistics, Journal of
Miller, T. (1880), Fire Insurance. A Theory of Statistics, Journal of the Institute of
Milne, Joshua (1815), A Treatise on the Valuation of Annuities and Assurances on Lives
and Survivorships. Longman, Hurst, Rees, Orme and Brown. Reprinted by Elibron
Classics.
Mitchell, B.R. (1988), British Historical Statistics. Cambridge University Press
Modigliani, Franco and Merton Miller (1958) The Cost of Capital, Corporation
Finance and the Theory of Investment, American Economic Review, Vol. 48, No.
3 (June 1958), pp.261297.
Modigliani, F. and R.Sutch (1966), Innovations in Interest Rate Policy, American
Economic Review, (May 1966), pp.178197.
Moody, P.E. (1964), Life Funds and Equity Investment, Journal of the Institute of
Actuaries, Vol. 90, No. 2 (September 1964), pp.175210.
Morgan, William (1779), The Principles and Doctrines of Assurances, Annuities on
Lives, and Contingent Reversions. [Note the first edition was published in 1779, the
second edition was published over forty years later, but the exact year of publica-
tion is unclear.]
Morgan, William (1829), A View of the Rise and Progress of the Equitable Society, and
of the Causes Which have Contributed to Its Success. Longman, Rees, Orme, Brown
and Green.
Murray, A.C. (1937), The Investment Policy of Life Assurance Offices. Transactions
of the Faculty of Actuaries, Vol. 16, No. 152, pp.247284.
Nicoll, J. (1902), The Actuarial Aspects of Recent Legislation, in the United
Kingdom and other Countries, on the subject of Compensation to Workmen for
Accidents, Journal of the Institute of Actuaries, Vol. 36, No. 5/6, pp.417564.
Ogborn, Maurice Edward (1962), Equitable Assurances. George Allen and Unwin.
Ogborn, M.E. (1953), On the Nature of the Function Expressive of the Law of
Human Mortality. Journal of the Institute of Actuaries, Vol. 79, pp.170212.
Osborne, M.F.M. (1959), Brownian Motion in the Stock Market, Operational
Research, Vol. 7, No. 2, pp.145173.
Bibliography 333
Parliamentary Papers (1825), Report of the Select Committee to Consider the Laws
Respecting the Friendly Societies.
Parliamentary Papers (1853), Report of the Select Committee on Assurance Associations.
Pegler, J.B. H (1948), The Actuarial Principles of Investment. Journal of the Institute
of Actuaries, Vol. 74, pp.179211.
Penman, William (1911), On the Valuation of the Liabilities of an Insurance
Company under its Employers Liability Contracts, Journal of the Institute of
Penman, William (1933), A Review of Investment Principles and Practices. Journal
Pentikainen, T. (1952), On the Net Retention and Solvency of Insurance compa-
nies, Scandinavian Actuarial Journal.
Pentikainen, T. and Rantala (1982), Solvency of Insurers and Equalisation Reserves.
Helsinki.
Plackett, R.L. (1971), Risk Theory, Transactions of the Faculty of Actuaries, Vol. 32,
No. 237, pp.337362.
Pollard, J.H. (1982), Outstanding Claims Provisions: A Distribution-Free Statistical
Approach, Journal of the Institute of Actuaries, Vol. 109, No. 3, pp.417433.
Porteous, D.A. (1936), Pension and Widows and Orphans Funds. Cambridge
University Press.
Price, Richard (1772), Observations on Reversionary Payments; on Schemes for Providing
Annuities for Widows, and for Persons in Old Age; To Which Are Added Four Essays.
T.Cadfll.
Puckridge, C.E. (1948), The Rate of Interest which Should be Employed in the
Valuation of Pension Fund and the Values which Should be Placed on Existing
Investments. Journal of Institute of Actuaries, Vol. 74, No. 1, pp.130.
Raynes, H.E. (1928), The Place of Ordinary Stocks and Shares (as distinct from
Fixed Interest bearing Securities) in the Investment of Life Assurance Funds.
Raynes, H.E. (1937), Equities and Fixed Interest Stocks During Twenty-Five Years.
Redington, F.M. (1952), Review of the Principles of Life-Office Valuations. Journal
Redington, F.M. (1981), The Flock and the Sheep and Other Essays. Journal of the
Institute of Actuaries, Vol. 108, pp.361404.
Redington, F.M. (1982), The Phase of Transition An Historical Essay. Journal of
Reid, D.H. (1978), Claim Reserving in General Insurance, Journal of the Institute of
Renshaw, A.E. (1989), Chain Ladder and Interactive Modelling (Claims Reserving
and GLIM), Journal of the Institute of Actuaries, Vol. 116, No. 3, pp.559587.
Roll, Richard (1988), R2, Journal of Finance, Vol. 43, No. 3 (July 1988), pp.541566.
334Bibliography
Ross, S.A. (1976), The Arbitrage Theory of Capital Asset Pricing, Journal of Economic
Theory, Vol. 13, No. 3, pp.341360.
Ross, S.A. (1977), The Determination of Financial Structure: The Incentive
Signalling Approach, Bell Journal of Economics, Spring 1977, pp.2340.
Rubinstein, M. (2006), A History of the Theory of Investments. Wiley.
Ryan, Gerald H. (1903), Methods of Valuation and Distribution of Profits in the
United Kingdom. Journal of the Institute of Actuaries, Vol. 38, No. 1, pp.6997.
Ryan, J.P. and K.P.W.Larner (1990), The Valuation of General Insurance Companies,
Ryder, J.M. (1976), Subjectivism A Reply in Defence of Classical Actuarial
Methods, Journal of the Institute of Actuaries, Vol. 103, No. 1, pp.59112.
Samuelson, P.A. (1960), The St. Petersburg Paradox as a Divergent Double Limit.
International Economic Review, Vol. 1, No. 1.
Samuelson, P.A. (1965), Rational Theory of Warrant Pricing, Industrial Management
Review, Vol. 6, pp.1339.
Samuelson, P.A. (1965), Proof That Properly Anticipated Prices Fluctuate Randomly,
Industrial Management Review, Vol. 6, No. 1 (Spring 1965), pp.4149.
Samuelson, P.A. and R.C.Merton (1969), A Complete Model of Warrant Pricing
that Maximises Utility, Sloan Management Review, Winter 1969, pp.1746.
Scott, Peter (2002), Towards the Cult of Equity? Insurance Companies and the Interwar
Capital Market.
Scott, P.G. etal (1996), An Alternative to the Net Premium Valuation Method for
Statutory Reporting, British Actuarial Journal, Vol. 2, No. 3, pp.527621.
Scott, W.F. (1977), A Reserve Basis for Maturity Guarantees in Unit-Linked Life
Assurance, Transactions of the Faculty of Actuaries, Vol. 35, pp.365415.
Seal, H. (1969), Simulation of the Ruin Potential of Non-Life Insurance Companies,
Transactions of the Society of Actuaries, Vol. 21.
Sharpe, W.F. (1963), A Simplified Model for Portfolio Analysis, Management Science,
Vol. 9, No. 2 (January 1963), pp.277293.
Sharpe, W.F. (1964), Capital Asset Prices: A Theory of Market Equilibrium under
Conditions of Risk, Journal of Finance, Vol. 19, No. 3 (September 1964),
pp.425442.
Sheldon, T.J. and A.D. Smith (2004), Market-Consistent Valuation of a Life
Assurance Business, British Actuarial Journal, Vol. 10, No. 3, pp.543626.
Shiller, R.J. (1981), Do Stock Prices Move Too Much to Be Justified by Subsequent
Changes in Dividends? American Economic Review, Vol. 71, No. 3 (June 1981),
pp.421436.
Skelman, R.S. (1968), The Assessment and Distribution of Profits from Life
Business, Journal of the Institute of Actuaries, Vol. 94, No. 1, pp.53100.
Skidelsky, Robert (2003), John Maynard Keynes 1883 1946: Economist, Philosopher,
Statesman. Penguin Books.
Smith, Edgar Lawrence (1924), Common Stocks as Long Term Investments. Macmillan.
Bibliography 335
Sprague, Thomas Bond (1863), On Certain Methods for the Valuation of Liabilities
of a Life Assurance Company. Journal of the Institute of Actuaries, Vol. 11, No. 2,
pp.90108.
Sprague, Thomas B. (1870), On the Proper Method of Estimating the Liability of a
Life Insurance Company under its Policies. Journal of the Institute of Actuaries,
Vol. 15, No. 6, pp.411432.
Stigler, Stephen M. (1986), The History of Statistics. Belknap Harvard.
Stiglitz, J.E. (1972), Some Aspects of the Pure Theory of Corporate Finance:
Bankruptcies and Takeovers, Bell Journal of Economics and Management Science,
Autumn 1972, pp.458482.
Supple, Barry (1970), The Royal Exchange Assurance: A History of British Insurance
17201970. Cambridge University Press.
Suttie, T.R. (1944), The Treatment of Appreciation or Depreciation in the Assets of
a Life Assurance Fund. Journal of the Institute of Actuaries, Vol. 72, pp.203228.
Taylor, G.C. (1983), An Invariance Principle for the Analysis of Non-Life Insurance
Claims, Journal of the Institute of Actuaries, Vol. 110, No. 1, pp.205242.
Taylor, G.C. (1984), Solvency Margin Funding For General Insurance Companies,
Thiele, T.N. (1871), En mathematisk Formel for Dodeligheden. Reitzel, Kjobenhaven.
Thornton, P.N. and A.F.Wilson (1992), A Realistic Approach to Pension Funding,
Tobin, J. (1958), Liquidity Preference as Behaviour Towards Risk, Review of
Economic Studies, Vol. 25, No. 2 (February 1958), pp.6586.
Todhunter, Isaac (1865), History of the Mathematical Theory of Probability From the
Time of Pascal to That of Laplace. London and Cambridge.
Trowbridge, J.R. (1977), Assessmentism An Alternative to Pensions Funding?
Turner et al (1984), Terminology of pension funding methods. Working Party
Report, Institute of Actuaries.
Vasicek, O. (1977), An Equilibrium Characterization of the Term Structure, Journal
of Financial Economics, Vol. 5, pp.177188.
Verrall, R.J. (1989), A State Space Representation of the Chain Ladder Linear
Model, Journal of the Institute of Actuaries, Vol. 116, No. 3, pp.589609.
Walford, Cornelius (1868), The Insurance Guide and Hand-Book. Charles and Edwin
Layton.
Walford, Cornelius (1878), On the Scientific Application of Data to the Purpose of
Deducing Rates of Premium for Fire Insurance, Journal of the Institute of Actuaries
and Assurance Magazine, Vol. 21, No. 1, pp.137.
Warner, S.G. (1918), Opening Address by the President, Journal of the Institute of
Actuaries, Vol. 51, No.1, pp.124.
Watson, Alfred W. (1903), An Account of an Investigation of the Sickness and Mortality
Experience of the I.O.O.F. Manchester Unity During the Five Years 18931897.
Cambridge University Press (re-printed 1953).
336Bibliography
Whittall, William (1882), On the Rates of Fatal Accidents in Various Occupations,

Journal of the Institute of Actuaries and Assurance Magazine, Vol. 23, No. 3,
pp.188220.
Whyte, Lewis G. (1947), Life Office Investments in a Planned Economy. Transactions
of the Faculty of Actuaries, Vol. 18, No. 168, pp.215259.
Wilkie, A.D. (1977), Maturity (and other) Guarantees Under Unit Linked Policies,
Transactions of the Faculty of Actuaries, Vol. 36, pp.2741.
Wilkie, A.D. (1986), A Stochastic Investment Model for Actuarial Use, Transactions
of the Faculty of Actuaries, Vol. 39, p.341403.
Wilkie, A.D. (1986; 2), Some Applications of Stochastic Investment Models,
Journal of the Institute of Actuaries Students Society, Vol. 29, pp.2551.
Wilkie, A.D. (1987), An Option Pricing Approach to Bonus Policy, Journal of the
Institute of Actuaries, Vol. 114, pp.2190.
Wilkie, A.D. (1995), More on a Stochastic Asset Model for Actuarial Use, British
Actuarial Journal, Vol. 1, p.777964.
Wilkie, A.D., H.R.Waters and S.Yang (2003), Reserving, Pricing and Hedging For
Policies with Guaranteed Annuity Options, British Actuarial Journal, Vol. 9,
pp.263391.
Williams, John Burr (1938), The Theory of Investment Value. BN Publishing.
Wise, A.J. (1985), The Matching of Assets to Liabilities, Transactions of the Faculty
of Actuaries, Vol. 40, pp.1881.
Woolhouse, W.S.B. (1839), Investigation of Mortality in the Indian Army.
Wright, P.W. et al (1998), A Review of the Statutory Valuation of Long-Term
Insurance Business in the United Kingdom, British Actuarial Journal, Vol. 4, No.
4, pp.803864 and British Actuarial Journal, Vol. 4, No. 5, pp.10291058.
Wright, T.S. (1990), A Stochastic Method for Claim Reserving in General Insurance,
Zillmer, A. (1863), Beitrage zur Theorie der Pramien-Reserve bei Lebens-
Versicherungs-Anstalten.
Index
A ASTIN. See Actuarial Studies in

Abbott, W.M., 304 Non-Life Insurance (ASTIN)
Accounting Standards Committee, 259 Australia, 211, 220, 282
Actuarial Studies in Non-Life Insurance
(ASTIN), 289, 290
Actuary of the National Debt, 77 B
Aggregate method (pension funding), Babbage, Charles, 81, 123
259 Bachelier, Louis, 15860, 179, 197
AIG, 319 Bacon, J., 239
Albert Insurance Company, 127 Baddeley, 278
Allen, R.G.D., 178 Bailey, A.H., 85, 98, 99, 102, 103,
Allin, S.J.H.W., 130 111, 114, 116, 121, 124, 127,
The Amicable Society for a Perpetual 187, 188, 281, 312
Assurance Office Annuities, 28 Basel Capital Accord, 219
Analysis of surplus, 279 Bayes, Thomas, 3846, 49, 57, 58, 64,
Analysis of variance, 302 298, 311
Anderson, J.L., 192, 193, 273, 313 Bayley, G.V., 109
Andras, H.W., 138, 139 Beard, Robert E., 28890, 313
Annuity Guarantees Working Party, Benjamin, Sidney, viii, 195200,
226 2024, 215, 217, 219, 288,
Ansell, C., 84n21, 84n22 293, 294, 300, 302, 304, 315,
Anti-selection, 13, 63, 78 321, 2955
Appraisal value, 307, 308 Benz, Norman, 190, 191
Arbitrage Pricing Theory, 158 Bernoulli, Daniel, 314, 215, 311
Arithmetic triangle, 8 Bernoulli, Jacob, 204, 32, 3840
Arthur, T.G., 2657, 269, 274 Bernoulli, Nicolas, 313
Assessmentism, 257, 258 Beta (asset pricing), 157, 158, 161, 163

DOI10.1007/978-3-319-33183-6
338Index
Beta probability distribution, 41, 42 Brown, L., 143n164

Binns, J.D., 104, 192, 193, 273, 313 Brown, Samuel, 58, 97, 98, 277, 283
Binomial probability distribution, 20,
21, 23
Binomial Theorem, 21 C
Binomial trees, 168 Cairns, A.J.G., 216
Black-Derman-Toy model, 176 Canada, 104, 119, 211, 220, 282
Black, Fisher, 158, 16174, 176, Canadian Institute of Actuaries, 304
177, 193, 199, 201, 215, Capital Asset Pricing Model, 1528,
217, 253, 314 177, 306
Black-Karasinski model, 176 Cardano, Gerolamo, 3, 5, 6, 8
Black-Scholes equation, 1637, 1724, Case-by-case estimation (general
217 insurance reserving), 277,
Blume, Marshall, 180 294, 296
Boden, D.E., 258 Casualty Actuarial Society, 275
Bolton, M.J., 226n75, 226n76, Census (1871), 281
227n77 Central Limit Theorem, 23, 469, 297,
(with-profit) Bonus 301
reversionary bonus, 68, 128, 129, Chain ladder method, 295, 297, 302,
136, 189, 192, 210, 212, 213 303
terminal bonus, 191, 192, 210, Champernowne, Professor, 178, 179
21214, 228 Chapman-Kolmogorov equation,
Bonus and Valuation Research Group, 159
216 Chi-Square significant test, 298
Bonus Distribution with High Equity Clarke, H.G., 304
Backing Working Party, 190 Clark, G.J., 268
Bonus reserve valuation, 129, 133, 135, Clarkson, Robert S., 214, 215, 217
141, 142, 212 Clerical Medical, 121
Boole, George, 43 Cochrane, John, 184
Bornhuetter-Ferguson method, 297 Colbran, R.B., 60, 259, 262
Bornhuetter, R.L., 297 Collins, T.P., 201, 202, 204, 210, 214
Bowie, David, 216 Common law, 281
Box, G.E.P., 298 Condorcet, Nicolas de, 9
Boyle, Phelim P., 201, 207, 208, 210, Consols, 96, 97, 102, 103, 205, 243
230 Contribution method (bonus), 138,
Brennan and Schwartz model, 175, 206 139, 143
Brennan, M.J., 175, 201, 206 Cooper, S.H., 193
Bride, M., 308, 309 Corby, F.B., 199202, 214
British Insurance Association, 290 Corley Committee of Inquiry regarding
Brown, Hugh, 285, 286 the Equitable Life Assurance
Brownian motion, 15863, 165, 167, Society, 228n78, 229, 229n79
168, 177, 179 Corley, R.D., 228n78, 229n79
Index 339
Cost of capital, 14850, 217, Duration (interest rate sensitivity), 96,

230, 3058 97, 1046, 10810, 128, 129,
Coutts, C.R.V., 103, 129, 130 170, 284
Cowles, Alfred, 181 Dynamic financial analysis, 304
Cox-Ingersoll-Ross model, 1746 Dynamic hedging/replication, 110,
Cox, J.C., 1668, 1746 111, 163, 168, 1704, 176,
Cox, P.R., 136n147 193, 2014, 313, 314
Crabbe, R.J.W., 233n1 Dynamic solvency testing (DST), 303,
Craighead, D.H., 297, 304 304
Cult of the equity, 248, 252 Dyson, A.C.L., 26770
D E
DAlembert, Jean le Rond, 33 Eagle Life, 97
Dalton, Hugh, 120, 188, 240, Eagles, Ian, 302
242, 319 The Economist, 111, 178, 183
Daston, L., 6n4 Edmonds, T.R., 82, 91, 123, 124
Davies, Griffith, 81, 82, 123 Edwards, John, 64, 65
Day, J.G., 246, 247, 249, 250, 260, Efficient frontier, 152, 154, 156
273 Efficient markets hypothesis, 181, 314
Daykin, C.D., 268, 300, 3057 Einstein, Albert, 159, 215
Debentures, 97, 99, 100, 119 Elderton, W.Palin, 94
De Jong, P., 298, 299, 3013 Emergence of surplus, 12830,
De Moivre, Abraham, 5, 227, 34, 38, 135, 294
49, 50, 54, 59, 90, 311, 622 Employers Liability Act, 1880, 281
De Morgan, Augustus, 33, 74, 88, 92, Employers liability insurance, 2806,
102, 123 311
Department of Trade and Industry, 226 Epps, G.S.W., 238
Derman, Emanuel, 176 Equitable life, 50, 55, 75, 81, 194,
De Witt, Johan, 915, 31, 311 228, 316
Diagonal Model (Sharpe), 156, 157 Equity & Law, 120
Discontinuance solvency assessment Equivalent martingale measure,
(pension funds), 267, 269 168, 177
Diversification (portfolio), 114, 153, Euler, Leonhard, 23, 45, 45n12
154, 288, 314 Excess-of-Loss reinsurance, 289
Diversification (time), 32, 114, 185 Exley, C.J., 216, 26770, 274, 309
Dividend discount model, 182 Expectations Hypothesis (term
Dodson, James, 4957, 60, 66, 67, 70, structure model), 169,
71, 74, 80, 121, 311, 321 170, 173
DST. See Dynamic solvency testing Experience rating/No Claims Discount,
(DST) 2901
Duffy, P.G., 309 Exposed to risk, 10, 73, 74,
Dullaway, D., 222 278, 284
340Index
F Geoghegan, T.J., 208n44

Faculty and Institute of Actuaries Life Germany, 9, 127, 320
Board, 218 Geske, Robert, 165
Fair value accounting, 220 Gilley, D.F., 2426, 255, 256, 259,
Fair Value Working Party, 218, 221 261
Fama, Eugene, 17981, 183, 184, 191, Global Financial Crisis, viii, 31821
198, 206, 315 Gompertz, Benjamin, 82, 84, 904
Fellows, D.E., 260 Goode Committee, 264
Ferguson, Duncan, 203, 209, 297 Graduation, 80, 8794, 311
Fermat, Pierre de, 59, 12, 13, 20, 31 Graunt, John, 915, 61
Financial economics, viii, 105, 14585, Great Fire of London, 1666, 277
187, 201, 204, 20622, 231, Green, S.J., 269
234, 266, 274, 277, 306, 308, Greenwood, P.M., 264n69
309, 31315, 317, 318 Gross premium valuation, 124, 130,
Financial Services Authority, 219 132, 318
Finlaison, John, viii, 7580, 89, 123, Guaranteed annuity option, 207, 216,
124 22231, 316, 319
Finland, 287 Guardian Assurance Company, 97
Fire insurance, 4, 30, 27682, 289 Gulland, C.M., 100n65
Fisher, Irving, 146, 147, 149, 150, 161, Gwilt, R.L. (Dick), 2235, 227, 317
169, 171
Fisher, Sir Ronald, 42, 43
Fishers Separation Theorem, 146, 150 H
Ford, A., 202n33, 203n34 Hacking, Ian, viii, 2n1, 58
France, 1, 9, 11, 23, 79 Hairs, C.J., 218
Francis, John, 18, 28, 78 Halley, Edmond, 5, 9, 1319, 247,
Frankland, R., 213 30, 31, 54, 59, 61, 215, 311
Freedom with publicity, 127, 292, Hansen, Peter, 184
300, 321 Hardy, G.F., 139
French, Kenneth, 183, 184, 191, Hardy, M., 230
198, 315 Hardy, Ralph, 235, 236
French Royal Academy of Sciences, 44 Hare, D.J.P., 216, 217, 220, 221
Funnell, D., 2426, 255 Harrison, Michael, 167, 168, 177
Hart, D.M., 300n48
Hawtrey, Ralph G., 120
G Hayek, Friedrich, 215
Galilei, Galileo, 3 Haynes, A.T., 105, 10811, 192, 193,
Galloway, Thomas, 83, 84, 89 224, 225, 252, 313
Galton, Francis, 88 Heath, D., 176
Gamma probability distribution, 287 Heath-Jarrow-Morton model, 176
Gauss, Carl, 479, 88 Hemsted, J.R., 249
General Insurance Study Group, 292, 299 Hey, G.B., 290, 291, 3057
General Records Office, 281, 282 Heywood, Geoffrey, 2446, 249, 273
Index 341
Hibbert, A.J., 2202 Kemp, M.H.D., 216

Hicks, John, 169 Kendall, Maurice, 17880, 183
Hidden reserve, 132, 133, 135, 141 Kennedy, S.P.L., 191n7, 192n9,
Holbrook, J.P., 252 209n45
Homans, Sheppard, 86, 137, 138 Keogh, T.W., 264n69
Hooker, N.D., 303n55, 304n56 Keynes, John Maynard, 113, 11520,
House of Lords, 229 151, 169, 215, 313, 185192
Hudde, Johannes, 13, 79 King, George, 235, 236, 238
Hull, J.C., 176 Kingston, T.D., 258
Hull-White model, 176 Kirton, R.J., 10810, 193, 225, 252,
Huygens, Christiaan, 6, 12, 13, 20, 31 273, 313
Kolmogorov, Andrey, 159
Kreps, D.M., 168n40
I
Illiquidity premium, 99, 100, 206
Immunization, 109, 202, 203 L
Incurred but not reported (IBNR) The Lancet, 82
claims or reserves, 276, 284 Lander, Maxwell, 2446, 249, 273
Ingersoll, J.E., 1746 Laplace, Pierre-Simon, 5, 23, 439, 88,
Institute of Mathematics and Its 311
Applications, 302 Larner, K.P.W., 307, 308
Insurable interest, 28, 64, 76 Laudable Society, 62
Insurance Companies Act of 1974, 192 (Williams) Law of Conservation of
International Accounting Standards Investment Value, 147, 148
Board, 217, 220 (Weak) Law of Large Numbers
International Association of Actuaries, (Bernoullis Theorem), 202,
219 153
International Association of Insurance Least Squares, 22, 469, 88, 291
Supervisors, 219 Legal & General, 82, 115, 119
International Congress of Actuaries, 119 Legendre, Adrien-Marie, 479, 88
Leibnitz, Gottfried Wilhelm, 2, 5
Life Assurance Companies Act 1870,
J 127, 292
Jarrow, R.A., 176 Life assurance contract
Jellicoe, Charles, 125, 136, 137 annuities, 4, 50, 73, 75, 76, 101,
Jenkins, G.M., 298 233
Jensen, Michael, 181, 182, 274 endowment, 116, 118, 140
Johnson, P.D., 290, 291 term assurance, 30, 31, 50, 54, 57
Jones, David, 123 whole-of-life, 37, 4955, 57, 6570,
116, 121, 140
Life Insurance Act 1774, 64
K Limited price indexation, 270
Kalman filter, 298, 303 Linear regression, 302
Karasinski, P., 176 Lintner, J.V., 155, 156
342Index
Liquidity Preference Hypothesis (term Maxwell, Robert, 264

structure model), 169, 170 May, George, 114, 121, 154
Lloyds of London, 275 McKean, Henry, 161
Lochhead, R.K., 132, 134, 140 McKelvey, K.J., 24651, 260, 273
Lognormal probability distribution, McKelvey, K.M., 248, 249
287, 302 McLeish, D.J.D., 2613
Lomax, M.W., 308, 309 Mean-reversion (in equity returns),
London Annuity Society, 62 1835, 202, 203, 266
London Assurance, 29, 55, 62, 98 Meckling, W.H., 150n10
London Bills of Mortality, 9, 10, 14, Mehta, S.J.B., 216, 270, 274, 309
51, 59, 60, 66, 67 Melville, H.E., 140
London Fire Engine Establishment, Mercers Company, 19
278 Merton (credit) model, 165
London Life, 97 Merton, Robert C., 162, 165, 182
Low, C.W.F., 260 Metropolitan Life, 97
Lumsden, I.C., 253 Miller, Merton, 151, 152
Lundberg, F., 287n17 Miller, Thomas, 279
Milne, Joshua, 73, 84, 123
Minimum Funding Requirement, 264,
M 316, 320
Macaulay, Frederick, 1046, 108, 119 Modern Portfolio Theory, 1528
Macaulay, T.B., 1046, 108, 119 Modigliani, Franco, 151
Macdonald, Angus S., 216 Modigliani-Miller Proposition I, 149
Macdonald, R.E., 251, 2568 Montmort, Pierre, 31
Mackenzie, A.G., 111 Moody, P.E., 190, 191
Maclean, J.B., 86n28, 224n71 Morgan, Arthur, 82, 85
Makeham, William, 924, 126 Morgan, William, 6470, 74, 75, 79,
Manches, J.L., 293 80, 82, 102, 123, 131, 135,
Manchester Unity sickness tables, 79 136, 276, 312, 316
Mandelbrot, Benoit, 215 Mortality tables
Manly, Henry, 2358 American experience table, 86
Marine insurance, 1, 4, 28, 29, 275 Amicable experience table, 84, 89
Market price of risk, 1724 Breslau table, 1316, 18, 61
Market Segmentation Hypothesis (term Carlisle mortality table, 73
structure model), 170 English Life Table, No. 3, 235
Markowitz, Harry, 145, 1526, Equitable experience table, 81, 82
160, 168 Northampton mortality table, 58
Marsh, Terry, 182 Seventeen Offices table, 846, 89
Maturity Guarantees Working Party, Thirty American Offices table, 86
204, 205, 210, 213, 215, 217, Twenty Offices table, 85
229, 230, 288 Mortgages, 69, 96, 97, 99, 100, 103,
Maud, J.M., 210 111, 312
Maudling, R.G., 239, 286 Morton, A., 176
Maximum Likelihood, 22, 43, 48, 302 Mossin, J., 155
Index 343
Motor insurance, 276, 290, 291, 293, Pegler, J.B. H., 121, 134, 135, 195
296, 316 Penman, William, 101, 105,
Motor Risk Statistics Bureau, 290, 291 2836, 295
Murray, A.C., 104, 105 Pension Protection Fund, 319
Mutual indemnity fund, 286 Pentikainen, T., 289
Mutual Life Insurance Company of Perks, Wilfred, 109
NewYork, 86 Pesonen, E., 289
Petty, William, 11
Pliska, S.R., 168, 177
N Plymen, 247, 250, 251
Napoleonic wars, 71143, 238 Pocock, John, 64, 65
Nash, W.O., 237 Poisson probability distribution, 3, 20,
National Bureau of Economic 21, 23, 38, 402, 458, 152,
Research, 104 196, 197, 2879, 298
National Mutual, 113, 115, 116, 120, Policyholders Reasonable Expectations,
129, 192 211, 212
Needleman, Peter, 222 Pollard, J.H., 297, 298, 301, 302
Neison, F.G.P., 281, 282 Post Office Directory, 278
Net premium valuation, 67, 12333, Poyser, C.A., 233n1
135, 21113, 219, 220, 227, Preferred Habitat Hypothesis (term
228, 318 structure model), 170, 173
Newton, Isaac, 2, 5, 215 Price, Richard, viii, 15, 19, 37, 38, 49,
Nicoll, John, 282, 283, 286 5764, 67, 72, 80, 102, 136,
Normal probability distribution, 23 183, 276, 311, 316, 322
Nowell, P.J., 201 Principle of Insufficient Reason,
42
Principles and Practices of Financial
O Management, 219, 221
Occupational Pensions Board, 259 Probability of ruin, 195200, 202,
Ogborn, M.E., viii, 194 209, 212, 217, 218, 230,
Option pricing theory, 8, 15869, 288, 289, 301
177, 201, 202, 204, 208, Problem of Points, 5, 8, 167
210, 211, 215 Projected unit credit method (pension
Orstein-Uhlenbeck process, 173 funding), 259
Osborne, M.F.M., 160, 178 Provincial Insurance, 115,
Outstanding claims reserves, 284, 296 116, 185
Prudential Assurance, 105
Prussia, 9
P Puckridge, C.E., 2404, 273
Paish, Professor, 178, 179
Pascal, BlaisePearl Assurance, 59, 12,
13, 20, 31, 167, 311 Q
Pearson frequency curves, 94 Quetelet, Adolphe, 9
344Index
R Sharpe, William F., 1557

Randall, P.A., 2657, 269, 274 Sheldon, T.J., 221, 222
Random walk hypothesis, 179 Shiller, Robert J., 182, 184
Raynes, H.E., 115, 119, 120, 192, 313 Short rate model, 173, 175, 176
Recknell, G.H., 192, 193 Sisson, Dr., 266, 267
Redington, F.M., viii, 10411, 126, Skerman, R.S., 190, 191, 195
168, 189, 191, 209, 253, 273, Small pox, 73
313, 322 Smith, Adam, 215
Reid, D.H., 296, 299, 301 Smith, Andrew D., 222, 269, 270, 274,
Renshaw, A.E., 302, 303 309, 313
Reverse yield gap, 189, 244 Smith, Edgar Lawrence, 115, 117, 119,
Risk-based capital, 218, 219, 303 151, 313
Risk-neutral valuation, 163, 164, Smith, P., 195n17
1668, 174, 201, 215 Solvency II, 217, 222, 320
Risk theory, 197, 28692, 296, Solvency margin, 294, 299301, 3036
299, 313 Solvency Working Group (general
Rock Life, 97 insurance), 301, 305
Roll, Richard, 183, 254, 259, 274, 315 South Sea bubble, 29
Ross, Stephen A., 158 Speed, Cliff, 216, 263, 298
Royal Exchange Assurance, 29, 30, 57 Sprague, Thomas, 125, 126, 128
Royal Society, 11, 14, 23, 34, 38, 50, Stagflation, 259, 261, 316
58, 82 Stalker, A.C., 191
Royal Statistical Society, 178 State-space model, 298
Rubinstein, Mark, 147, 155, 167, 168 Stewart, C.M., 2613
Run-off triangle, 295, 298 Stiglitz, J.E., 150n9
Ryan, J.P., 297, 3079 Storr-Best, R.H., 136
Ryder, J.M., 292 Strong, W.R., 295
Sun Life Assurance Society, 73
Sun Life of Canada, 104, 119
S Surrender value, 67, 101, 124, 214
Saint Petersburg Paradox, 313 Sutch, R., 170
Samuelson, Paul, 159, 160, 178 Suttie, T.R., 133, 1402
Saunders, Adrian, 216n58 Sweden, 287
Savage, Jim, 159
Scholes, Myron, 171
Schwartz, E.S., 175, 201, 206 T
Scott, F.C., 116, 185 Taylor, G.C., 107, 298, 299, 301, 305,
Scottish Ministers Fund, 233 306, 308
Scottish Widows, 120, 192, 223 Technical provisions, 299301
Scott, P.G., 21113, 218 Term premium, 170, 207
Scott, W.F., 197200, 202 Thiele, T.N., 94
Scurfield, H.H., 295n34 Thornton, Paul, 260, 262, 263,
Seal, Hilary, 287 266, 271
Index 345
Tinner, Thomas, 237 Waters, Howard, 229

Tobin, James, 1546 Watson, Alfred W., 79n12
Tobin Separation Theorem, 154 White, A.D., 176
Todhunter, Isaac, 7, 23 Whittall, William, 281, 282
Tomlinson, L.P., 272 Whyte, Lewis G., 120, 134, 135
Toy, W., 176 Wilkie, A.D., 197, 199, 200, 2028,
Treen, W.R., 304 210, 21418, 221, 229,
Treynor, J.L., 155 230, 258, 266, 269,
Trowbridge, J.R., 257, 258 301, 315
Turnbull, C.J., 2202 Wilkie model, 2048, 230, 301,
315, 318
Williams, John Burr, 147
U Wilson, A.F., 263, 266
Ulpian, 16 Wilson, D.C.E., 84n20
Ultimate loss ratio, 302 Wise, A.J., 2524, 271
Unexpired premium reserve, 283 Woolhouse, W.S.B., 85, 89, 235
Unexpired risk reserve, 283 Workmens Compensation Acts (1897,
Uniform (reversionary) bonus rate, 139 1900, 1906, 1925, 1926),
United Kingdom Consumer Price 282, 285
Index, 237 World War One, 100, 103, 111, 112,
United States, 86, 142 114, 115, 140, 141, 237, 238,
Unit of exposure, 276 240, 248, 312, 313
University Life Assurance Society, 138 World War Two, 71143, 187231,
Urwin, R.C., 268 235, 240, 248, 280, 286, 287,
Utility, 32, 33, 174, 214 289, 313, 319
Wright, P.W., 212, 213, 218
Wright, T.S., 303
V
Value-at-risk, 263
Variational calculus, 298 Y
Vasicek, Oldrich, 145, 1717, 206, 207 Yang, S., 229, 230
Vehicle and General, 293 Yield curve modelling, 108, 164,
Verrall, R.J., 302, 303 16977, 207
W Z
Walford, Cornelius, 17, 280 Zehnwirth, B., 298, 299, 3013
Warner, Samuel George, 71 Zillmer, August, 127, 128

Craig Turnbull (Auth.) - A History of British Actuarial Thought-Palgrave Macmillan (2017)

Uploaded by

Copyright:

Available Formats

Craig Turnbull (Auth.) - A History of British Actuarial Thought-Palgrave Macmillan (2017)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Craig Turnbull (Auth.) - A History of British Actuarial Thought-Palgrave Macmillan (2017)

Uploaded by

Copyright:

Available Formats

Craig Turnbull

A History of British Actuarial Thought

ISBN 978-3-319-33182-9ISBN 978-3-319-33183-6(eBook)

Library of Congress Control Number: 2016955056

The Editor(s) (if applicable) and The Author(s) 2017

Printed on acid-free paper

This Palgrave Macmillan imprint is published by Springer Nature

To paraphrase Winston Churchill, the title of this work doubtlessly requires

1 Probability andLife Contingencies, 16501750:

2 Revolutionary Developments Between 1750 and181037

3 Life fromtheNapoleonic Wars totheSecond World War71

4 A Brief History ofFinancial Economics forActuaries145

5 Life Offices After theSecond World War:

6 British Actuarial Thought inDefined Benefit

7 British Actuarial Thought inGeneral Insurance (18511994)275

Conclusions: Looking Back, Looking Forward311

Craig Turnbullis an Edinburgh-based Fellow of the Institute and Faculty of

The Author(s) 2017 1

 robability andLife Contingencies

We have to go forward another 2,000 years to the thinkers of the Scientific

As noted above, marine insurance and, to a lesser degree, fire insurance

Pascal andFermat (1654)

Fermats approach is unexciting. Furthermore, the process of tabulating each

Chapter II, Todhunter (1865).

Mathematical expectation as the probability-weighted sum of uncertain

Cox etal. (1979).

Together, these ideas and insights represent a fundamental breakthrough

J ohn Graunt, Johan de Witt andTheir

information on the distribution of the population by age (so there was no

As a graduate of law and mathematics, de Witt was well equipped to apply

128/30=6.7 %; and 1126/128=1.6 %.

Edmond Halleys Breslau Table (1693)

of life annuities, presented a paper to the Royal Society in 1693 on mortal-

Hacking (1975), p.113.

However, his paper is silent on the quantitative steps he took to transform

Figure 1.1 compares the three seventeenth-century mortality tables dis-

0.9 Graunt (1662)

0.7 Halley (1693)

Fig. 1.1 Three seventeenth-century mortality tables

L ife Contingencies at theEnd oftheSeventeenth

Fig. 1.2 Government annuity prices and Halley

Francis (1853), p.59.

The Mercers Company, established in London, was one such example. It

J acob Bernoulli andtheFirst Law ofLarge

where p, m and n are defined as above, and is an arbitrarily small number.

Bernoullis Theorem fundamentally requires the population probability, p,

 e Moivres Contributions toProbability

achieved particular renown in astronomy and complex analysis. He is perhaps

Consulting Dr Halleys Table of Observations, I found that the Decrements of

Stigler (1986), p.84.

where r is the annually compounded rate of interest, n is the number of years

Cumulative Decremnents of Life

Fig. 1.4 A comparison of de Moivres annuity price approximation with exact

The remainder of Annuities on Lives is dedicated to the pricing of annuities

L ife Contingencies intheFirst Half

Chapter IX, Francis (1853).

be materially improved by extending their charter to include fire insurance.

 aniel Bernoulli: TheBeginnings ofUtility Theory

The ultimate modern solution to the paradox, as argued by Samuelson,40

2 Revolutionary Developments Between 1750 and181037

3 Life fromtheNapoleonic Wars totheSecond World War71

4 A Brief History ofFinancial Economics forActuaries145

5 Life Offices After theSecond World War:

6 British Actuarial Thought inDefined Benefit

7 British Actuarial Thought inGeneral Insurance (18511994)275

Conclusions: Looking Back, Looking Forward311

robability andLife Contingencies

Pascal andFermat (1654)

J ohn Graunt, Johan de Witt andTheir

Edmond Halleys Breslau Table (1693)

L ife Contingencies at theEnd oftheSeventeenth

J acob Bernoulli andtheFirst Law ofLarge

e Moivres Contributions toProbability

L ife Contingencies intheFirst Half

aniel Bernoulli: TheBeginnings ofUtility Theory

From Probability toStatistics, 17641810

Laplaces Bayesian Analysis (17741781)

L east Squares, Errors andtheCentral Limit Theorem

The Emergence ofwith-Profits, 17561782

James Dodson andWhole-of-Life Assurance (17561772)

The Beginnings oftheEquitable (17571772)

Richard Price andHis Observations (1772)

illiam Morgan andtheActuarial Management ofwith-

Developments inMortality Modelling, 18081881