1

Ragnar Arnason

Productivity and Productivity Growth

in the
Icelandic Fisheries

A paper submitted at the conference

Competitiveness within the Global Fisheries
Akureyri
April 6-7 2000

DRAFT
Not to be quoted without consulting the author

* Department of Economics
University of Iceland
 2

[email protected]
 - Abstract -

This paper attempts to measure productivity growth in the Icelandic fisheries during
the period 1975 to 1995. The standard theory of total factor productivity (TFP) is
extended to accommodate the special case of the fisheries, where the size of the fish
stocks represents a major input into the production process. Utilizing aggregative time
series data on the Icelandic fisheries from 1974-1995, a Törnquist approximation to
the appropriate Divisia index is employed to obtain estimates of changes in total
factor productivity in the Icelandic fisheries. According to these measurements the
average annual growth in total factor productivity has been quite high during this
period compared to that of other major industries in Iceland and abroad.. Moreover,
there are no signs that this growth in total factor productivity has abated over time.
Indeed, it seems to have increased during the latter half of the period. It is tempting to
associate this experience with the impact of the ITQ fisheries management which
became the dominant form of fisheries management in Iceland during this period.
 2

0. Introduction

This paper attempts to estimate productivity and productivity growth in Icelandic

fisheries during the period from 1974 to 1995. This interval of time was imposed by
the availability of the necessary data at the time the basic study commenced (1998).
Since then, one year of additional data, i.e. 1996, has become available. Some data
relevant to productivity measurements before 1974 are available but they are
somewhat spotty and most likely inconsistent with the time series that have been
systematically collected since then.

Several studies of productivity in Icelandic fisheries have previously been

done. Most, however, are quite simplistic and their results, consequently, not very
reliable. Recently, however, there have been four more noteworthy studies in the area.
Three of them (Hagfræðistofnun 1997, Valsson and Klemensson 1998 and the study
by the National Economic Institute published in Committee on Natural Resources,
1999) employ the classical methods of total factor productivity (TFP) which I in this
study have chosen to refer to as two-factor productivity or 2FP. The fourth, published
by the Ministry of Fisheries 1999, is more advanced and employs inter alia the
method of three factor productivity or 3FP. The current paper is extensively based on
this last study.1

This paper has two basic aims. First, to explain the deficiency of the
traditional 2FP or TFP to measure productivity in industries where unpriced natural
resources constitute an important part of the inputs and to develop the more
appropriate 3FP measure for these cases. The second aim is to present reasonably
reliable estimates of productivity and productivity growth in the Icelandic fisheries.
These results are interesting in themselves not least in relation to the rather dramatic
changes in the institutional framework of the Icelandic fisheries in the past two-three
decades.

As it turns out, productivity growth in the Icelandic fisheries since 1974 has
been very high compared to what is usually observed in mature industries. This, of
course, suggests the need for explanations in terms of the causal relationship from
exogenous changes, institutional and otherwise, to this unusually high rate of
productivity growth. For reasons of space, however, I will refrain from providing such
explanations in this paper.

The paper is organized broadly as follows. The next section reviews the basic
theory of total factor productivity (TFP) and its measurements. This is followed by a
section on the special problems for productivity measurements posed by resource
based industries such as fisheries and the development of the 3FP measure to account
for the impact of the natural resource. The third section describes the data on which
the productivity measurements are based. In the fourth section, actual productivity
measurements for the Icelandic fisheries are presented and compared to corresponding
productivity measurements for other industries. The final section of the paper,
provides a brief discussion of some of the results.

1. Basic theory
1
My main co-workers in that study were Sveinn Agnarsson, Benedikt Valsson and Jon Oskar
Þorsteinsson. Their contribution to the current paper is hereby gratefully acknowledged.
 3

Productivity refers to the quantity of outputs obtained from a given quantity of inputs.
Productivity change refers to a shift in this relationship. This shift can obviously be
either positive or negative, i.e. it can either increase or decrease the volume of outputs
obtained from a given level of inputs.

Let us for illustrative purposes assume the existence of a single valued

production function2

(1) y = Y(x,t),

where y represents the quantity of output, the vector x the quantity of inputs and t
time. More generally we may regard (1) as the (upper) boundary of a production
possibility set.

Factor productivity is defined as the quantity of output obtained from the use of
a given quantity of input. More precisely:

(2) F(xi; y, x, t)  y/xi,

where F(xi; y, x, t) is read as the ‘productivity of factor xi in producing y’.

In the case of production processes with more than one input, factor productivity
is obviously a poor measure of productivity. As indicated in expression (2), factor
productivity in this case depends in general on the use of the other inputs.
Consequently, any given factor productivity may be altered by varying the other
inputs. This deficiency has motivated the development of the concept of total factor
productivity (TFP) to indicate the overall productivity of all inputs (factors) (Coelli et
al. 1998, Grosskopf, 1993). TFP may be defined as:

(3) F(X;y, x, t)  y/X,

where X represents the appropriate index for all inputs.

The rate of change of total factor productivity may be written as:

(4) dlnF(X;y, x, t)/dt   ( X ; y , x, t ) /

F F(X;y, x, t) = y / y  X / X ,

where y / y  dlny/dt and X / X  dlnX/dt represent proportional changes in outputs

and inputs, respectively.

TFP change can be usefully decomposed into (a) shifts in the production
possibility frontier and/or (b) movement by the producers toward or away from the
production possibility frontier. The former is called technical change. The latter is
referred to as a change in (production) efficiency (Grosskopf, 1993).

Figure 1 illustrates these two components of TFP change. The shift in the
production possibility frontier (production function) represents technical change. The
2
The extension of the following theory to account for many outputs is straight-forward (Coelli
et al. 1998.
 4

shift in production from point A to point B (with the production possibility frontier
unchanged), represents an improvement in production efficiency.

Figure 1
Technical and Efficiency Change
(A= initial production point, B= final production point)

Output

Technical
change

Efficiency
change
A

Inputs

To formally establish this decomposition of TFP change into technical change

and efficiency change , it is convenient to write the production fucntion as:

(5) y = Y(x,t)g(t),

where g(t)[0,1] is the efficiency parameter measuring the (relative) distance of

actual production from the production frontier. Obviously, g(t)=1 indicates full
efficiency and g(t)<1 less than full efficiency.

Differentiating (5) in a logrithmic form yields:

(6) dlny/dt = Y/xi(xi/y) x i /xi + Yt/Y + gt/g.

Profit maximization (i.e. efficiency) under perfect competition implies that Y/ xi=
wi/p, all i, where p is the market price of output. Hence we may rewrite (6) as:

(7) dlny/dt  si x i /xi = Yt/Y + gt/g,

where si(wixi/ py) is the expenditure on input i as a fraction of total revenues. It

should be noted that the expression si x i /xi represents the change in a Divisia index
 5

for the use of inputs (Deaton and Muellbauer 1980, Chambers 1988). Divisia indices,
as is well known, are actually in many respects ideal indices (Diewert 1976).

Now, according to our definition of TFP above3, the LHS of (7) is the rate of
change of TFP where the use of inputs are measured by the appropriate Divisia index.
The first term on the RHS of (7), Yt/Y, is the rate of change in the production
possibility frontier. It represents, in other words, technical change. The second term,
gt/g, is the change in production efficiency. This establishes the proposition that TFP
change can be decomposed into (a) technical change and (b) efficiency change. For
later reference we summarize our results so far in the expression:

(8) dlnF(X;y, x, t)/dt = dlny/dt  si x i /xi = Yt/Y + gt/g,

where, as before, F(X;y, x, t) is TFP.

In interpreting any observed TFP change it is helpful to recognize that its two
components, technical change and efficiency change, are qualitatively different.
Technical change refers to shift or alterations in the production possibilities. It is
purely a technological matter. Efficiency, on the other hand, has nothing to do with
production possibilities. It is a measure of how well firms are run, i.e., how close they
are to the technically attainable upper boundary of production.

The first equality in expression (8), is the fundamental equation for the
measurement of TFP change. It has been widely used to estimate changes in TFP in
firms, industries and the macro-economy (Solow, 1957, Jorgenson and Griliches
1967, Denison 1972). More recently it has been used in the so-called growth
accounting (Laitner 1993, Herbertsson, 1999).

The great advantage of (8) is that it doesn’t require knowledge of the production
function or its parameters. It only requires observations on the quantity of outputs,
inputs and input cost shares.

For observations in discrete time, (8) may be approximated by the Törnquist

(1936) expression:

I
(9) lnF(X;y,x,t)  lnF(X;y,x,t-1)=(lnyt - lnyt-1)  0.5  ( sit  sit 1 ) (lnxit. lnxit-1)
i 1

Finally, before proceeding further, it may be helpful to draw the reader’s

attention to the following:

 In many productivity studies gross factor income (or value-added) is employed

instead of gross production, y. This has inter alia the advantage that the factor
costs shares, si sum to unity.
 In most cases it is inappropriate to talk about total factor productivity, TFP. There
are almost always factors that are not accounted for. Consequently, the
terminology of multi-factor productivity and more specifically one-factor
3
See equations (3) and (4).
 6

productivity, two-factor productivity etc. is more accurate and, therefore, probably

preferable.
 It is important to realize that, at least in principle, factor productivity and in
particular productivity growth should be largely independent of the level of the
inputs used. The reason is that the productivity measurements constitute an
attempt to focus on shifts in the production possibility frontier and/or production
efficiency by eliminating, to the extent possible, the impact of the input level. In
the case of separable technological shifts and efficiency as e.g. given by the
production function y = A(t)Y(x)g(t), where A(t)and g(t) represent the level of
technology and efficiency, respectively, this independence would be perfect.

2. Productivity in Fisheries

All industries use natural resources as inputs. In some circumstances this complicates
the measurement of productivity and its development over time. If the natural
resource inputs are bought in the market place like the other inputs there is no
problem. Equation (8) for total factor productivity still applies and can be measured
by its empirical counterpart, equation (9). Many natural resources used in production,
however, are common property and obtainable free of charge. For these ‘non-market
natural resources’ the corresponding cost share is, consequently, zero and equations
(8) and (9) no longer apply.

Fisheries are to a very great extent based on the use of non-market natural
resources, namely the fish stocks.4 Empirical studies of fisheries production functions
show that the size of the fish stocks is a major determinant of the volume of landings
(see e.g. Helgason and Kenward 1985, Bjorndal, 1987, Arnason, 1990). It follows that
for the estimation of productivity change in fisheries it is necessary to include the size
of the fish stocks.

The following very simple fisheries model may help to clarify these ideas.
Figure 2 illustrates
Figure 2 the conventional sustainable fisheries model. Measured along the
horizontal axis are the
The Sustainable usual economic
Fisheries Model inputs, capital and labour, here subsumed
under the heading fishing effort. The volume of production, i.e. harvest, is measured
up along the vertical axis and stock size as biomass down along the vertical axis in the
Harvest, y
lower half of the diagram. Sustainable
y1 yield

y2

e1 e2 Effort, e

4
It is interesting that
z2 this may be changing with the advent of individual transferable quota systems
that generate an implicit price for fish stock utilization.

z1 Sustainable
biomass
Biomass, z
 7

The curve in the upper half of the diagram in Figure 2 represents the
sustainable yield function which traces out the sustainable relationship between
sustainable effort and the harvest. The line in the lower half of the diagram is the
sustainable biomass curve which traces out the relationship between sustainable
fishing effort and biomass. As indicated in the diagram, sustainable biomass falls with
increased fishing effort and vice versa.

Now, the effort level e2 corresponds to output y2 and biomass z2. Reducing
effort to e1 will lead to an increase in sustainable harvest to y1. Hence, this appears as
an increase in productivity. This, however, is not correct as the other input, biomass,
has increased to z1. In fact, there has been no shift in the production function and both
production points (y1,e1,z1) and (y2,e2,z2) lie on the production possibility frontier.
Hence there has been no change in productivity.

With fish stocks as a production fact, our basic expression for change in total
factor productivity, TFP, equation (8) must be modified as follows:

(10) dlnF(X;y, x, t)/dt = dlny/dt  si x i /xi   i z i /zi,

where zi denotes the volume of fish stock i or some appropriate aggregation of several
fish stocks and i represents the elasticity of production (harvest) with respect to stock
i. The corresponding discrete time Törnquist approximation is:

(11) lnF(X;y,x,t)  lnF(X;y,x,t-1)=(lnyt - lnyt-1)  0.5  ( sit  sit 1 ) (lnxit. lnxit-1)

i 1
J
0.5  ( it   it 1 ) (lnzit. lnzit-1).
i 1
 8

It should be noticed that this analysis has clear implications for the impact of
improved fisheries management on productivity. The long term aim of fisheries
management is to move the fishery toward a more efficient point as for instance
described by the movement in fishing effort from e2 to e1 in Figure 2. This, however,
is merely a movement along the sustainable yield function. There is no increase in
productivity. Thus, improved fisheries management in the sense of adjusting fishing
effort to a more appropriate level and rebuilding the fish stocks does not imply an
increase in productivity as this concept is normally defined.

Nevertheless, it may be the case, that improved fisheries management

indirectly leads to increased productivity. Thus, for instance, a movement from a
common property fishery toward a private property one, often not only reduces
fishing effort and rebuilds fish stocks but also leads to an increase in economic
efficiency as more efficient fishing firms and more appropriate fishing methods
replace previous ones.

3. Data

According to equation (11), calculation of total factor (or rather three-factor)

productivity requires data on:

(1) Production quantity or gross factor income

(2) Labour use,
(3) Physical capital use,
(4) Natural capital (i.e. fish stocks).

The available data consist of annual time series observations on these four sets of
variables. These data are obtained from public sources; the National Economic
Institute and the Marine Resource Institute, and cover the period from 1974-1995. The
data are listed in Appendix 1.

The volume of production is estimated as gross factor income5 at constant

prices as calculated by the National Economic Institute.6 It is important to realize that
this series is deflated on the basis of its own separate price index. Hence, the quite
considerable real increase in the price of fish landings since the early 1980s7 do not
show up in this series. Consequently, they do not affect the productivity
measurements either. The development of this variable is illustrated in Figure 3.

5
Gross factor income is defined as total fishing revenues less all costs except those associated
with labour (labour remuneration and all associated charges) and capital (depreciation). This is
essentially value added in the production process.
6
For the basic methodology employed to generate this series see National Economic Institute
1994.
7
Thus, between 1987 and 1995, the years for which data are readily available, the real price of
landed cod increased by almost 18% or just over 2% per year.
 9

Figure 3
Fisheries Gross Factor Income 1974-1995
(B.ISK, Constant 1995 prices)

B.ISK
40

35

30

25

20

15

10
1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994
Source: National Economic Institute

Labour use is estimated as the total number of man-years in the harvesting

sector of the fishing industry. It is of some importance to realize that in addition to
vessel crew this includes a good deal of land-based labour such as vessel and gear
maintenance staff, office workers, procurement and marketing people, mangers and so
on. The evolution of this variable since 1974 is illustrated in Figure 4.
 10

Figure 4
Fisheries Use of Labout 1974-1995
(Man-years)

7000

6500

6000

5500

5000

4500

4000
1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994

Source: National Economic Institute

Physical capital is measured as the total value of the fishing fleet according to
a series maintained by the National Economic Institute (National Economic Institute
1994 and 1999). Basically this series is calculated as the accumulation over time of
the value of annual investments in vessels and equipment at procurement prices
deflated by the appropriate price index and depreciated annually according to
predetermined depreciation rates for the various components (vessel hulls, engines
and equipment) of the investments. The path of this series is illustrated in Figure 5.
 11

Figure 5
Fisheries Physical Capital
(B.ISK, 1995 prices)

90
85
80
75
70
65
60
55
50
45
40
1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994

Source: National Economic Institute

Finally, the biological capital variable was obtained aggregating the biomass
of all major stocks (Marine Research Institute 1998) multiplied by their market value
in 1995 (Fisheries Association of Iceland 1996), the basic price level of the
calculations. More precisely:

Figure 6 I
Z=  pi zi ,
Fish stocks:
i 1
Aggregate measure
(B.ISK)
where Z represents the aggregate stock, xi the stock size of species i, pi the
corresponding unit landings price in 1995 and I the number of species involved.8
Since, the elasticity of catch with respect to biomass differs greatly between demersal
240
and pelagic species, two separate biological capital variables one for each of these
group of species were actually generated The evolution of the aggregate biological
220
capital according to this statistic is illustrated in Figure 6.
200

180

160

8 140Actually, the calculation involves 9 species (cod, haddock, saithe, redfish, Greenland halibut,
shrimp, Norway lobster, herring and capelin) which account for over 90% of the total value of
120
landings.

100
1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994

Sources: Marine Research Institute, Fisheries Association of Iceland

 12

4. Measurement

We are now in a position to turn our attention to the actual measurement of

productivity and productivity growth in the Icelandic fisheries. Our basic equation for
measuring TFP-change, or more properly 3FP in the Icelandic fisheries is the
Törnquist approximation as given in equation (11) in section 2.
I
(11) lnF(X;y,x,t)  lnF(X;y,x,t-1)=(lnyt - lnyt-1)  0.5  ( sit  sit 1 ) (lnxit. lnxit-1)
i 1
J

 0.5  ( it   it 1 ) (lnzit. lnzit-1),

i 1

where, it will be recalled, the output measure, y, is gross factor income, the economic
market inputs, xi, are the use of physical capital and labour respectively and the non-
market input, zi, represents the size of the fish stocks. These variables were described
in section 2 and listed in the appendix.

The cost (or factor) shares, si, employed are the actual factor shares as
calculated by the National Economic Institute every year. The fish stocks were
divided into two substocks; the pelagic stocks and the other stocks (of which
demersals dominate). The reason for this is that studies of harvesting production
functions have generally found the elasticity of output (harvest) with respect to these
 13

two species of fish to be very different. More preciesly, it is generally found that the
elasticity of harvest with respects to stocks is quite low for pelagic (and schooling)
species, but approaches unity for demersal species. Following the estimates of
Helgason and Kenward (1985), Björndal (1987) and Arnason (1990), we set in this
study these elasticities to 0.1 for the pelagics and 0.85 for the other stocks. These
elasticities are assumed constant over the calculation period.

The resulting calculated path of 3FP is illustrated in Figure 7. The

corresponding numerical results are listed in Appendix 2.

Figure 7
Three factor productivity for the Icelandic fisheries

200

180

160
Index

140

120

100

80
1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994

Figure 7 indicates a substantial growth in productivity in the Icelandic

fisheries as measured by the 3FP measure. Over the data period as a whole (22 years),
factor productivity has increased by over 91%. The average rate of productivity
growth has been about 3.1%.9 This rate of productivity growth over such a long period
is much higher than in other major Icelandic industries and indeed most major
industries abroad.

The Institute of Economic Studies (Hagfræðistofnun) at the University of

Iceland has recently published a report on the trend in TFP for various Icelandic
industries (Hagfræðistofnun, 1997). These results in comparison with the current
results for the Icelandic fisheries are listed in Table 1

9
I.e. the compound rate of growth calculated as ln[FP(1995)/3FP(1974)]/21, where FP(t) is
measured productivity at time t.
 14

Table 1
Trend* in total factor productivity in Icelandic industries
(1974-1995; 3FP for fisheries 2FP for the others industries)

Fisheries 3,49
Fish processing 1,25
Agriculture 0,8
Manufacturing industry 1,02
All industries 1,13
* Calculated as the coefficient b in the regression Z=a+btime, where Z represents the estimated
productivity.

Table 1 shows that productivity growth in the Icelandic fisheries has been
much higher (about 3 times higher) than in other major Icelandic industries.

Comparison with foreign industries produces a similar result. Table 2 lists

average annual growth10 in factor productivity in the Icelandic fisheries and several
industries in Denmark and the USA.

Table 2
Average productivity growth* in industries in Iceland, Denmark and USA
(Denmark and USA 1973-93 and 2FP; Iceland 1974-95 and 3FP)
Annual average
Country Industry productivity growth
USA Agriculture/fisheries -1.1%
Manufacturing industry 0.5%
Services 1.0%
Denmark Agriculture/fisheries -0.9%
Manufacturing industry 1.3%
Services 0.2%
Iceland Fisheries 4.4%
* Calculated as [FP(T)-FP(0)]/(T-1), FP(t) is measured productivity at time t and T is the length of the
data period.

The international comparison tells very much the same story as the domestic
one. Productivity growth in the Icelandic fisheries has been much faster than in the
listed industries abroad. This result seems sufficiently robust and the magnitude great
enough to be regarded as highly remarkable. As such it suggests the need for an
explanation in terms of special factors.

One explanation that could be forwarded is that the 3FP is responsible for this
result. Afer all this novel productivity measure is used to estimate the productivity of
the Icelandic fisheries and not the other industries. This, however, cannot be the case
for the simple reason that the more standard 2FP measure yields very much the same
average productivity growth for the Icelandic fisheries for the period as a whole. This
is illustrated in Figure 8 where both the 3FP measure and the 2FP one are drawn.

10
Defined simply as the total growth divided by the number of years-1.
 15

Figure 8
Two and three factor productivity for the Icelandic fisheries

200

180

160
Indices

140

120

100

80
1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994

2FP 3FP

As indicated in Figure 8, total productivity growth over the period as a whole

is almost identical on both the 3FP measure and 2FP measures. Hence, the
comparatively high rate of measured productivity growth in the Icelandic fisheries
cannot be a spurious outcome of the 3FP measure.

In spite of yielding about the same overall productivity growth, the periodic
difference between the 2FP measure and the 3FP one illustrated in Figure 8 may
appear curious. There is, however, a ready explanation. The 2FP does not correct for
the size of the fish stocks. Hence, the variability of the fish stocks during the period
(illustrated in Figure 6 above) is bound to show up in the 2FP measure. Indeed, it
turns out that there is a substantial positive correlation between the size of the fish
stocks and the 2FP measure. The correlation coefficient is r=0.54 and is highly
significant.

The 3FP measure, on the other hand, in order to focus on true factor
productivity, attempts to correct for the impact of the fish stocks. In this, it seems to
have been highly successful because it turns out that the correlation between the
calculated 3FP and the size of the fish stocks is virtually zero (r=-0.03). Herein lies
the great advantage of the 3FP measure for the current work and probably fisheries
productivity measurements in general. It manages to eliminate almost all the impact
of the fish stocks on output and hence allows us to see true productivity changes more
clearly.
 16

Given all this, it is not surprising that almost all the observed difference
between the evolution of 2FP and 3FP can be explained in terms of variations in the
fish stocks. The following figure, Figure 9, traces out the path of the fish stocks as
given in Figure 6 and the annual difference between the calculated 2FP and 3FP.

Figure 9
Fish stocks and 2FP-3FP

250

200

150

100

50

0
1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994
-50

2FP-3FP Fish stocks

As indicated in Figure 9, the graph for the difference between the two
productivity measures, namely (2FP-3FP), and the graph for the fish stocks are
almost identical. Indeed, the correlation between the two is r=0.99. This, again
verifies the appropriateness of the 3FP as a measure of productivity.

By correcting for fish stocks, the 3FP measure manages to substantially

smooth out the evolution of calculated productivity growth over the period from
1974-1995 compared to the 3FP measure. This is, of course, as it should be. Shifts in
production frontiers and efficiency are more likely to exhibit a fairly stable trend than
erratic fluctuations. Nevertheless, in Figure 7 we see clear indications of certain
variability in productivity growth. For instance there is high growth in productivity
from 1974 to about 1980. This is followed by a large drop in productivity from 1981
to 1984 with the result that overall productivity growth from 1974 to 1984 is
relatively low or just above 1% per annum. From 1984, on the other hand,
productivity growth in the fisheries has been quite high and stable and shows no signs
of abating. This raises the question of whether the productivity series shows some
signs of a structural change in the statistical sense. Are there indications of a shift in
the underlying data generating process during the period, or can we conclude that
what seems to be a systematic pattern is just a random fluctuation?
 17

To investigate these questions, linear splines have been fitted to the 3FP
series: The splines take the following general form:

3FP(t) = a + bt,

where t refers to time (here 1974-1995) and a and b are coefficients to be estimated.
These splines were fitted to the 3FP series over various intervals of time. The
question of structural break in the series can then be tested by standard statistical tests
e.g. the Chow test (Chow 1960) or a simple F test on the estimated slope coefficients,
i.e. b.

These investigations suggest a clear and highly significant structural break in

1983/4. This can be interpreted as evidence that productivity evolves differently over
time in the pre-1984 period than in the post-1984 period. There is also some evidence,
although less significant, of a similar structural break in 1990/91. The corresponding
trends in productivity (the b coefficient) over the sub-periods are reported in Table 3.

Table 3
Growth in Productivity (3FP) in subperiods
Periods Growth trend, b
1974-1983/4 3.6
1983/4-1995 6.0
1983/4-1990/1 5.6
1990/91-1995 7.4

According to Table 3, the growth in 3FP productivity is substantially and

significantly higher during the period, from 1983/4, than in the preceding period.
Similarly it seems that productivity growth has increased after 1990/1 compared to
1983/4-1990/1. This increase, however, is not statistically significant at the 5% level
although it is close to being so.

5 Discussion

The estimated productivity growth in the Icelandic fisheries discussed above is quite
high and much higher than in other major Iceland industries as well as comparable
industries abroad. It follows that there must be some special factors operating in
fisheries that account for this difference. Several possibilities jump to mind. Among
them the following may be mentioned: (1) The extension of the fisheries jurisdiction
to 200 miles. (2) Technological advance in fishing methods. (3) Improved
management of the fishing companies encouraged partly by the greatly increased
public ownership of these companies. (4) Improvements in the fisheries management
system, especially the introduction of the property rights based ITQ system that took
place in a stepwise fashion during the period. (5) The development of new fisheries.
(6) The liberalization of the Icelandic economy.
 18

Of these hypotheses, the last one should apply to all the industries in the
Icelandic economy. It cannot, therefore, explain the difference between the
productivity growth in the Icelandic fisheries and Iceland’s other industries.

Of the remaining five possible explanations, (1), (2) and (5) should be mainly
felt as a shift in the production possibility frontier. The rest should be primarily felt as
an increase in production efficiency. In particular, most of the impact of (3), the
improved management of fishing companies, and (4), the improved fisheries
management system, should precisely have this effect, i.e. enable the fishing
companies to get closer to the technological production possibility boundary. Indeed
there is a great deal of circumstantial evidence from the industry such as improved
quality of landings, greatly increased specialization in fisheries, better co-ordination
of harvesting to demand and etc., that indicates that this is precisely what has taken
place.

If this is true, then it may be expected that a good deal of the productivity
increase in fisheries since 1974, actually consists of increases in efficiency rather than
technological progress. Another study currently in progress (Agnarsson, 1999)
employing more varied and in some respects more flexible productivity measures, has
uncovered evidence that this may in fact be the case.

The role of the new fisheries management system, the ITQ system, in the
productivity growth of the Icelandic fisheries is an inviting area for further research. It
is easy to conjure stories to explain precisely how the ITQ system can have this effect.
In fact, some of the other explanations such as improved management of the fishing
companies and the development of new -fisheries can be traced back to the impact of
the ITQ system. Also, in many respects, the movement of productivity over time fits
nicely with major advances in the ITQ system. However, these are just possibilities
that hardly amount to formal arguments and certainly not proofs. An interesting and
relatively easy test of this hypothesis would be to compare fisheries productivity and
productivity growth across a number of fisheries that have become subject to property
rights based fisheries management systems at different times or not at all.
 19

References

Agnarsson, S. 1999. Efficiency and productivity of the Icelandic fisheries: A DEA

approach. Unpublished manuscript.
Arnason, Ragnar. 1990. A Numerical Model of The Icelandic Demersal Fisheries. In
G. Rodrigues (ed.) Operations Research and Management in Fishing. Kluwer.
Björndal, T. 1987. Production Economics and Optimal Stock Size in a North Atlantic
Fishery. Scandinavian Journal of Economics. 89
Chambers, R.G. (1988). Applied production analysis: A dual approach. Cambridge;
New York and Melbourne: Cambridge University Press
Chow, G.C. 1960. Tests of the Equality between Sets of Coefficients in Two Linear
Regressions. Econometrica 28.
Coelli, T., P.D.S. Rao og G.E. Battese. 1998. Introduction to Efficiency and
Productivity Analysis. Kluwer Academic Publishers.
Committee on Natural Resources. 1999. Progress Report with Attachments (in
Icelandic). Committee on Natural Resources, Reykjavik.
Deaton, A. og Muellbauer, J. 1980. Economics and Consumer Behaviour. Cambridge
University Press.
Denison, E. F. 1967. Classification of Sources of Growth. Review of Income and
Wealth, 18.
Diewert, W.E. 1976. Exact and Superlative Index Numbers. Journal of Econometrics
4.
Fisheries Association of Iceland. 1996. The Fishing Industry 1995 (in Icelandic).
Reykjavík.
Grosskopf, S. 1993. Efficiency and Productivity. Í H. O. Fried, C. A. K. Lovell og S.
S. Schmidt (ritst.): The Measurement of Productive Efficiency. Oxford
University Press.
Hagfræðistofnun Háskóla Íslands (1997). Productivity in Icelnadic Industries:
Comparison with Denmark and the United States. University of Iceland.
Reykjavik.
Helgason, Th. and Kenward M.G. 1985. Estimation of fishing power with relation to
exploited biomass. ICES C.M. 1985/D7.
Herbertsson, T. 1999. Sources of Economic Growth. University of Iceland Press.
Jorgenson, D. W. og Z. Griliches. 1967 The Explanation of Productivity Change.
Review of Economic Studies, 34.
Laitner, J. 1993. Long Run Growth and Human Capital. Canadian Journal of
Economics 26.
 20

Marine Research Institute. 1998. Utlilized Fish Stocks 1997 (in Icelandic). Reykjavík.
Marine Research Institute. Unpublished stock data.
Ministry of Fisheries. 1999. Productivity in Icelandic Fisheries. Working Group
Report. Ministry of Fisheries. Reykjavik.
National Economic Institute. 1994. National Accounts 1945-1992 (in Icelandic).
National Economic Institute, Reykjavik.
National Economic Institute. 1999. Óútgefin gögn um framleiðslureikninga
sundurliðað á einstaka báta og togarflokka.
National Economic Institute. 1999. Report on Industries 1966 (in Icelandic). National
Economic Institute, Reykjavik.
Solow, R.A. 1957. Technical Change and the Aggregate Production Function. Review
of Economcis and Statistics, 39.
Törnqvist, Leo. 1936. The Bank og Finland's Comsumption Price Index. Bank of
Finland Monthly Bulletin, nr. 10.
Valsson, Benedikt og Ólafur Klemenson. 1998. Productivity and Profitability of the Icelandic
Fisheries. X EAFE Conference.
 21

Appendix 1
Basic data

Gross Number Value of Value

factor of fishing of
income Man-years capital fish stocks
(B.ISK) (B.ISK) (B.ISK)
1974 14,53 5160 44,8 158,0
1975 15,63 5110 47,2 156,1
1976 15,62 5197 46,6 163,2
1977 19,22 5189 50,6 183,4
1978 22,17 5339 52,4 188,3
1979 26,60 5207 54,4 205,6
1980 29,57 5506 55,7 233,8
1981 30,61 5541 57,8 213,5
1982 27,04 5691 59,1 195,7
1983 21,25 5751 59,7 179,0
1984 21,04 5498 59,7 185,6
1985 24,57 5970 59,2 189,6
1986 30,14 6047 62,6 189,2
1987 35,00 6511 68,6 208,6
1988 36,46 6271 75,9 221,5
1989 35,06 6425 76,5 213,8
1990 37,40 6761 75,0 197,1
1991 34,14 6726 73,6 180,6
1992 34,09 6247 75,9 166,5
1993 35,94 6820 73,6 169,3
1994 34,53 6480 72,6 164,5
1995 34,35 6372 69,7 154,4
 22

Appendix 2
Calculated Productivity

Year 2FP 3FP

1974 100,0 100,0
1975 107,4 108,7
1976 106,1 103,6
1977 128,2 114,2
1978 143,5 125,3
1979 173,7 140,4
1980 184,0 132,4
1981 188,4 146,5
1982 162,1 134,5
1983 126,1 113,2
1984 129,6 114,4
1985 142,1 124,0
1986 170,2 148,9
1987 182,8 146,7
1988 190,5 145,4
1989 179,6 140,6
1990 186,1 156,6
1991 171,5 154,8
1992 178,4 175,6
1993 178,9 174,8
1994 178,9 178,4
1995 182,3 191,5