V O LU M E 2 0 | N U M B E R 4 | FAL L 2 0 0 8

Journal of
APPLIED CORPORATE FINANCE
A MO RG A N S TA N L E Y P U B L I C AT I O N

In This Issue: Honoring Stewart Myers

The Contributions of Stewart Myers to the Theory 8 Franklin Allen, University of Pennsylvania, Sudipto
and Practice of Corporate Finance Bhattacharya, London School of Economics, Raghuram Rajan,
University of Chicago, and Antoinette Schoar, MIT

MIT Roundtable on Corporate Risk Management 20 Panelists: Judy Lewent, Merck; Donald Lessard and
Andrew Lo, MIT; and Lakshmi Shyam-Sunder,
International Finance Corporation.
Moderated by Robert Merton, Harvard Business School.

Risk Management Failures: What Are They and When Do They Happen? 39 René Stulz, Ohio State University

Brealey, Myers, and Allen on Valuation, Capital Structure, and Agency Issues 49 Richard A. Brealey, London Business School,
Stewart C. Myers, MIT, and Franklin Allen,
University of Pennsylvania

Brealey, Myers, and Allen on Real Options 58 Richard A. Brealey, London Business School,
Stewart C. Myers, MIT, and Franklin Allen,
University of Pennsylvania

Equity Issues and the Disappearing Rights Offer Phenomenon 72 B. Espen Eckbo, Dartmouth College

Can Companies Use Hedging Programs to Profit from the Market? 86 Tim R. Adam, Humboldt University, and Chitru S. Fernando,
Evidence from Gold Producers University of Oklahoma

Corporate Leverage and Specialized Investments by Customers and Suppliers 98 Jayant R. Kale, Georgia State University, and
Husayn Shahrur, Bentley College

Estimating Risk-Adjusted Costs of Financial Distress 105 Heitor Almeida, University of Illinois at Urbana-Champaign,
and Thomas Philippon, New York University
 17456622, 2008, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1745-6622.2008.00202.x by University Of South Africa, Wiley Online Library on [30/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Risk Management Failures:
What Are They and When Do They Happen?

by René M. Stulz, Ohio State University*

BI
n media accounts and popular commentaries on job of anticipating the likely consequences of low-probability
the current financial crisis, a constant refrain is outcomes and developing effective responses to them.
that the risk management function in many of
the world’s largest financial institutions has failed Was the Collapse of LTCM a Risk Management Failure?
to carry out its responsibilities. To cite just one example, an We now know a good deal about the rise and fall of Long-
article in the Financial Times declares “it is obvious that there Term Capital Management, or LTCM.2 In 1994, ex-Salomon
has been a massive failure of risk management across most Brothers traders and two future Nobel Prize winners started a
of Wall Street.”1 hedge fund called the Long-Term Capital Fund. LTCM was
In this article, I want to challenge, or at least qualify, the company that managed the fund. The fund performed
this assertion by examining what it means for risk manage- superbly for most of its life, with investors earning 20% for
ment to fail. My main aim in these pages is to show that ten months in 1994, 43% in 1995, 41% in 1996, and 17%
the fact that an institution suffers an extremely large loss in 1997. But 1998 turned out very differently. In August and
does not necessarily imply that risk management failed, or September of 1998, following the default of Russia on its
that the institution made a mistake. In making my case, I ruble-denominated debt, world capital markets were in crisis
will not spend much time on the current crisis, but instead and LTCM lost most of its capital. Before its collapse, LTCM
begin with the collapse of Long-Term Capital Management had capital close to $5 billion, assets in excess of $100 billion,
in 1998. Having accumulated a decade of hindsight on this and derivatives with a notional amount in excess of $1 tril-
case, we now have enough information to at least ask the right lion. By mid-September, LTCM’s capital had fallen by more
questions to determine whether risk management failed and than $3.5 billion and the Federal Reserve Bank of New York
to ensure that, if risk management is blamed, it is blamed for coordinated a rescue by private financial institutions that
the right reasons. injected $3.65 billion into the fund.
Getting the diagnosis right is important because the Does a loss of more than 70% of capital, and a rescue by
changes in risk management that take place in response to banks involving an injection of $3.65 billion of new capital,
the crisis could be the wrong ones. Most troubling, top execu- necessarily represent a failure of risk management? It turns
tives and investors could continue to expect more from risk out that this is not an easy question to answer. To define a risk
management than it can deliver. With this goal in mind, I management failure, one first must have a clear understand-
show when bad outcomes can be blamed on risk management ing of the role and the limitations of risk management.
and when they cannot. And in so doing, I offer what amounts In a typical company, the role of risk management is to
to a taxonomy of risk management failures. identify and evaluate the risks faced by the firm, to communi-
I also address the question of whether and how risk cate these risks to senior management (and possibly the board
management failures can be used to improve the practice of risk of directors), and to monitor and manage those risks in a way
management. I conclude that the probabilities of large losses that ensures the firm bears only the risks its management
are measured very imprecisely and that, as a consequence, and board want exposure to. To guide them in monitoring
companies should rely less on estimates of such probabilities and managing risk, most companies specify one (or more)
and pay more attention to the implications of large losses for measures of overall risk, perhaps along with other metrics or
their survival. Among other suggestions, I propose that greater indicators. When a risk measure exceeds the company’s stated
use of scenario planning could allow institutions to do a better tolerance for risk, risk is reduced. But when the risk measure

* I am grateful for assistance from Jérôme Taillard and Mike Anderson, and for com- 2. The best public source of data on LTCM is the collection of four Harvard Business
ments by Rich Apostolik, Don Chew, Cliff Smith, Adrian Tschoegl, and Peter Tufano. School case studies by André Perold published in 1999. See “Long-Term Capital Man-
1. “Wall Street Dispatch: Imagination and Common Sense Brew a Safer Culture,” by agement (A) – (D),” available from Harvard Business School Publishing. Many books
David Wighton, Nov. 26, 2007, FT.com. have been written on LTCM. Some of the numbers used in this article come from Roger
Lowenstein, When Genius Failed: The Rise and Fall of Long-Term Capital Management
(New York: Random House, 2000).

Journal of Applied Corporate Finance • Volume 20 Number 4 A Morgan Stanley Publication • Fall 2008 39
 17456622, 2008, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1745-6622.2008.00202.x by University Of South Africa, Wiley Online Library on [30/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
falls below the firm’s targeted risk position, the firm will likely But suppose the bad outcome—the one-in-a-hundred
choose to increase risk. event—materializes, and the fund makes headlines for having
For financial institutions, one commonly used risk lost $3.5 billion. The natural inclination, of course, is to inter-
measure is Value-at-Risk, or VaR. As a measure of downside pret a 70% loss as clear evidence of a risk management failure.
risk, VaR is the maximum expected loss at a given confidence But in the example I’ve just recounted, the job done by the
level over a given period of time. For example, in the case of a firm’s risk managers could not have been improved on.
bank whose senior management specifies a 99% confidence As stated earlier, then, the main responsibility of risk
level, a one-day VaR of $150 million would mean that the management is to assess the firm’s risks and then commu-
firm has a 1% chance of making a loss in excess of $150 nicate that assessment to top management. And the final
million—provided, of course, the VaR has been correctly decision to take known risks rests not with the risk manager,
estimated. Although financial institutions generally report but with top management. This decision depends on the
daily VaR measures, VaRs can also be estimated for longer risk appetite or tolerance of an institution, and defining
periods of time. that appetite is one of the most important responsibilities
Given this definition of the role of risk management, of management and the board. It is at the heart of the firm’s
the actual returns of LTCM tell us little about whether its strategy and how it creates value for its shareholders.
risk management failed. To understand why, it is helpful to When determining the risk appetite of an organization,
consider a very simple hypothetical example. Suppose that you the general approach is to weigh the benefits of increased
stood in the shoes of the managers of LTCM in January 1998 risk-taking against the costs, and to aim for the point where
and had the opportunity to invest in trades that, viewed as a the marginal benefits equal the marginal costs. In the case of
portfolio, had a 99% chance of producing a return for the fund LTCM, it could be argued that the costs to the organization
before fees of 25% over the coming year, and a 1% chance from losing $3.5 billion for its investors was limited to just
of a loss of 70%. The expected return on the fund would be those losses—that is, there were no additional costs beyond
estimated to be 24.05% [(25% x .99) + (-70% x .01)]. Such the direct monetary loss. For many companies and finan-
an expected return would, of course, have been very attractive cial institutions, however, large losses can have significant
for a hedge fund, or any investor, again provided the assigned “deadweight” costs. For example, after a large loss, a finan-
probabilities and other underlying assumptions were reason- cial institution may be forced to scale back its investments
ably consistent with economic reality. and sell assets in unfavorable markets, face increased scrutiny
Though hypothetical, these numbers are roughly consis- from regulators, and lose valuable employees and custom-
tent with the actual returns of LTCM before 1998 and the ers concerned about the institution’s staying power. In any
expectations LTCM was holding out to its investors at the institution, management and the board must take account
time. For one thing, in its two best years, the fund earned of such deadweight costs when making decisions that create
more than 50% before fees, so that a return of 25% does the risk of large losses.4
not sound implausible. Second, LTCM told its investors to So, again, it is top management’s job to determine a
expect losses in excess of 20% in one year out of 50—which, company’s risk appetite. And taking that appetite as a given,
again, is roughly comparable to an expected loss of 70% in risk managers can help assess the expected profitability of
one year out of 100.3 a proposed investment by evaluating how much capital is
Now, given this information, let’s assume that whether required to support it.5
the fund had the high return or not depended on the flip of a Also worth noting, an investment that is not value-adding
“heavily loaded” coin—one whose odds of turning up heads for a given level of risk could become so if the firm’s risk
(a 25% return) was 99%. Had the fund been given the chance appetite increases—because less capital is then required to
to carry out this investment strategy for 100 years, it would support it. But whether taking large risks is worthwhile for
have earned 25% in 99 of them. an institution ultimately depends on its strategy, which in
In this hypothetical example, we are assuming that when turn depends heavily on its risk appetite. And risk managers,
the partners of LTCM made their investment decisions, those as already suggested, do not set either; both strategy and risk
decisions were based on the best available assessment of the appetite are the prerogatives of top management.
expected distribution of possible outcomes of the fund. To To illustrate this point, suppose that a company follows
the extent this assumption is correct, the fund’s risk managers the common practice of setting its risk tolerance by choos-
would have provided a sound basis for decision-making and ing a targeted credit rating. Once the target rating is chosen,
should have earned a gold medal for their work. there are multiple combinations of risk and capital that can

3. See Lowenstein, p. 63. 5. My article with Brian Nocco, Enterprise Risk Management: Theory and Practice,
4. Such costs are at the foundation of financial theories of how risk management cre- Journal of Applied Corporate Finance, Fall 2006, v18(8), 8-20, describes the key prin-
ates shareholder value. See René M. Stulz, Risk Management and Derivatives, Thomp- ciples of enterprise risk management, issues that arise in its implementation, and the role
son Publishing, 2003. of capital allocation.

40 Journal of Applied Corporate Finance • Volume 20 Number 4 A Morgan Stanley Publication • Fall 2008
 17456622, 2008, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1745-6622.2008.00202.x by University Of South Africa, Wiley Online Library on [30/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
be expected to achieve the target rating. For a given choice of firm’s risk appetite. The job of risk management is to ensure
leverage, the firm does not have much flexibility in choosing that top management knows and understands the probabili-
its risk level if it wants to achieve its target rating. If faced with ties associated with possible outcomes of the firm’s strategy
promising investment opportunities that would increase risk, before they make decisions to commit the firm’s capital.
the firm has two main choices: it can choose to reduce lever-
age (and financial risk) while taking on more business risk A Taxonomy of Risk Management Failures
to keep its rating; or it can maintain its leverage and accept a Having described the role of risk management, let’s now
lower credit rating. And there is also, of course, a third choice: consider what can go wrong.
maintain both the credit rating and existing leverage, but at The first step in risk management is to identify and
the cost of passing up the investment opportunity. measure risks. Risk measurement takes place in the context
LTCM provides a good example of such tradeoffs. In the of risk metrics like VaR that aggregate various types of risks
fall of 1997, the managers of LTCM concluded that they did to help top management understand the risk position of
not want to manage a business earning just 17% for its inves- the firm. The choice of risk metrics is the cornerstone of
tors, which is what their investors had earned for the year. risk management because it determines what top manage-
Instead, they wanted the higher returns achieved in 1995 and ment learns from risk managers about the firm’s overall risk
1996. At the end of 1997, LTCM had capital of $7.4 billion position. If the risk managers assess risk using measures that
but decided to return 36% of the capital to its investors. With are ill-suited to a firm’s strategy, risk management can fail
less capital, LTCM could still execute the same trades; but before the computers are ever turned on.
to fund them it had to borrow more and hence increase its Once a risk measure is chosen, there are two basic kinds
leverage. By increasing its leverage, it could boost the return of mistakes that can be made in measuring risk: (1) known
to its shareholders if things went well—but only by raising risks can be mismeasured and (2) important risks can be
the expected losses if things went poorly. ignored, either because they are undetected or wrongly viewed
Was increasing leverage a bad risk management decision? as immaterial.
In my example, I assume that the partners of LTCM knew Once the risks are identified and measured, they must be
the risks and the rewards from higher leverage. In the well- communicated to the firm’s leadership. A failure in commu-
worn language of financial economics, increasing leverage nicating risk to management is also a risk management
appeared to be a positive NPV decision when it was made. failure.
But ex post, or after the fact, it proved to be a costly decision After risks have been measured and communicated, top
since it meant that when assets fell in value, the value of the management decides how much and what kinds of risks to
fund’s equity fell much faster. take. At this point, it is the responsibility of risk management
There has been much discussion of top management to make sure that the firm actually takes these risks and not
incentives during the credit crisis, with many observers point- others. A company’s risk managers must keep track of and
ing to stock options as an inducement to take excessive risk. manage the firm’s risks to ensure they remain within the
Before reaching this conclusion, however, it’s useful to keep in established guidelines, a task that could involve hedging risks
mind that financial economists have argued for decades that and rejecting proposed trades or projects.
management’s incentives become better aligned with those of As this discussion suggests, then, there are five types of
their shareholders when management has a large stake in the risk management failures:
firm’s equity. Top management owned hundreds of millions 1) Failure to use appropriate risk metrics.
of dollars of equity in both Bear Stearns and Lehman at the 2) Mismeasurement of known risks.
peak valuation of these firms. And the partners of LTCM 3) Failure to take known risks into account.
collectively had almost $2 billion invested in the fund at 4) Failure in communicating risks to top management.
the beginning of 1998. If such equity stakes do not provide 5) Failure in monitoring and managing risks.
managers with strong incentives to make the right decisions I now discuss each of these failures in turn.
for their shareholders, what will?
In sum, effective risk management does not provide a Risk Metrics and Risk Management Failures
guarantee against failure. Even in companies with the best The risk metric that is most closely associated with modern
risk management people and systems, large losses can and risk management is VaR. I use this metric to show how risk
will occur as long as taking the risk of large losses increases management failures can result when a risk metric does
expected profits sufficiently for top management to be willing not provide the right information to top management not
to take that risk. With good risk management, such losses will because it is calculated incorrectly, but because it answers
be attributable to an unlucky “draw,” to a one-in-a-hundred the wrong question. A useful analogy is a weather forecast. If
event. Ultimately, how likely such large losses are will depend you are deciding to go on a three-day hike in the wilderness,
on choices made by those entrusted with determining the a weather forecast for the first day is not really helpful when

Journal of Applied Corporate Finance • Volume 20 Number 4 A Morgan Stanley Publication • Fall 2008 41
 17456622, 2008, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1745-6622.2008.00202.x by University Of South Africa, Wiley Online Library on [30/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
making the decision to go or not—yes, you can still turn of risk management performance based on VaR exceedances
back in the middle of the first day, but you suffer the brunt would indicate that the firm has excellent risk management.
of the bad weather if it occurs in the second day. Choosing In other words, VaR does not capture catastrophic losses that
the right risk metrics is similar to choosing the right weather have a small probability of occurring.
forecast horizon. I have not seen monthly VaR estimates from LTCM.
The daily VaR measure is widely used in financial institu- What we do know is that, from March 1994 to December
tions for assessing the risk of trading activities. Though VaR 1997, LTCM had only eight months with losses and the worst
has proved to be an extremely useful risk measure, it is useful was 2.9%. In contrast, it had 37 months with gains.6 As a
only as long as the question it answers is well understood. result, one would have a hard time using historical monthly
VaR is the largest loss the firm expects to incur at a given returns to conclude that its risk management was flawed. Yet,
confidence level, which means that VaR tells us nothing about in August and September 1998, LTCM experienced losses
the distribution of the losses that exceed VaR. This limita- that were not predictable from its historical record. It was
tion of VaR has nothing to do with how institutions estimate as if an earthquake struck. The best daily or monthly VaR
their VaR. Rather it is simply a matter of how VaR is defined: measure would not have shown the possibility of such an
VaR is an estimate of the minimum worst loss expected, as event because the probability of an earthquake in any day or
opposed to the expected worst loss. Though appropriate for month is always extremely low. With our hypothetical LTCM
some purposes, VaR is likely to be useless for others. example, a one-year VaR at the 99% confidence level would
Large banks usually disclose data on that measure have been -70%. But the one-day or one-month VaR at the
quarterly, and will generally report the number of times in a 99% confidence level could have been much lower because
quarter the P&L had a loss that exceeded the daily VaR. For the possible loss of -70% would have been reflected in VaR
instance, in its annual report for 2006, UBS reported that it only at a much higher confidence level.
never had a loss that exceeded its daily VaR. But in 2007, it Daily VaR measures also implicitly assume that assets can
reported that it exceeded its daily VaR 29 times. The change be sold quickly or hedged, so that a firm can limit its losses
from 2006 to 2007 may well indicate not an increase in the essentially within a day. But both in 1998 and in the past year,
bank’s risk position, but fundamental changes taking place in we have seen that markets can become suddenly less liquid, so
the economy that made it difficult for risk managers to track that daily VaR measures lose their meaning. If a firm sits on a
risk on a daily basis. Moreover, such a large number of VaR portfolio that cannot be traded, a daily VaR measure is not a
exceedances provides little or no information about the impli- measure of the risk of the portfolio because the firm is stuck
cation of these exceedances for the financial health of UBS. with the portfolio for a much longer period of time.
The larger number of exceedances could have been attribut-
able mainly to rapid increases in volatility—increases that were Mismeasurement of Known Risks
anticipated by the traders and resulted in large trading gains. Now let’s turn to the case where management has chosen
Or, there could have been some very large losses but few large the right metrics, but the risks have been measured incor-
gains. In the former case, the firm could be ahead at the end rectly, for a variety of reasons discussed below. In the LTCM
of the year; in the latter it could be in serious trouble. example presented earlier, the mismeasurement of risk could
Thus, though calculation of the daily market VaR can have taken a number of different forms.
be intellectually satisfying for risk managers (since it involves When measuring risk, risk managers attempt to under-
up-to-date quantitative techniques), an exclusive focus on stand the distribution of possible returns. In our simple
it could lead risk managers to ignore critical dimensions example, we chose to use a binomial distribution—the kind of
of risk. The daily market VaR should generally not be the distribution that would result from the toss of an appropriately
primary focus of top management, which must pay attention loaded coin. Risk managers could make a mistake in assessing
to longer-run indicators and implications of risk. As I show the probability of a large loss, or they could be wrong about
with an example below, short-run VaR measures can continue the size of the loss given that it takes place. Or they could
to mislead management by suggesting low risk until a series use the wrong distribution altogether. Further, in the case
of events produces a huge loss and the firm fails. of a financial institution with many positions, although the
Consider a firm that has a one-day VaR of $100 million for distribution associated with each position may be estimated
its trading book at the 1% probability level. This means that the properly, the correlation among the different positions may
firm has a one percent chance of losing more than $100 million. be mismeasured. Such correlations are extremely important
If this firm exceeded its VaR once in 100 trading days but lost in risk management because, as correlations increase, so does
$10 billion on that one occasion, all existing statistical tests the vulnerability of a portfolio to low-probability events.

6. These monthly returns are for Long-Term Capital Management, L.P. (B), prepared
by André Perold, Harvard Business School, 1999.

42 Journal of Applied Corporate Finance • Volume 20 Number 4 A Morgan Stanley Publication • Fall 2008
 17456622, 2008, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1745-6622.2008.00202.x by University Of South Africa, Wiley Online Library on [30/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
In the LTCM example, the true probability of a loss of may well have been one of them—risk managers could not
70% may have been higher than 1%, perhaps as high as have been expected to anticipate such an abrupt increase in
25%. In that event, the expected return of LTCM in my correlations. But, once such shifts in correlations are recog-
hypothetical example would have been a paltry 1.25%. At the nized, risk managers’ assessments of the risk of a portfolio or
time, investors could have earned a higher expected return by company must take account of them.
investing in T-bills. And such a risk management mistake— The Limits of Risk Measurement. To sum up, then, statisti-
assessing the probability of the bad outcome at 1% instead cal techniques are generally used to estimate the distribution
of 25% —would have led the fund managers to make trades of known risks. Such approaches work well when there is a
that, on an ex ante or probabilistic basis, were expected to lot of data and when it’s reasonable to expect future returns
reduce value. to have the same distribution as past returns. For instance,
But if LTCM made such a mistake, how would we ever suppose that a risk manager wants to estimate the volatility
know? The answer is that we cannot know based only on the of the return of a liquid stock. She will have hundreds of data
information that it made a loss of 70%. We cannot identify points to fit a model of the volatility of the stock; and in most
such a mistake after the fact because LTCM lost 70% on only cases that model will perform reasonably well.
one occasion. Given the assumed distribution of possible losses, But, in other cases, historical data will be of little
LTCM was capable of losing 70%, whether the true probabil- use—say, because a risk has not manifested itself in the past.
ity of that loss was 1% or 25%. And given the hypothetical For instance, prior to the subprime crisis, there had been no
conditions of this example, we can conclude nothing about the experience of a downturn in the real estate market when large
effectiveness of LTCM’s risk management from its 70% loss. amounts of securitized subprime mortgages were outstanding.
Another possible failure of risk measurement would be In such a situation, it was not possible to obtain a distribution
to mistake some other probability distribution for a standard of losses associated with a sharp downturn in real estate by
binomial one. For instance, in our hypothetical example, in using only historical data. To evaluate the risks of under-
addition to a 1% chance of a 70% loss, there may have been a writing and purchasing mortgage-backed securities whose
9% chance of a 100% loss. In this case, although the expected realization could not be observed in the past, a risk manager
return would have been a respectable-sounding 12.8%, the would have needed to understand both the likelihood of a
expectation of a 9% chance of a total wipeout might have decrease in real estate prices, and the expected effect of such
caused LTCM’s partners to invest differently. Nevertheless, in a decrease on the prices of those securities. Such an exercise
January 1998, the probability of such a loss given the portfolio would have been complex and worth undertaking only if the
of trades they had assembled (or were planning to assem- likelihood of a large decrease in real estate prices was recog-
ble) could well have been very small—a once-in-a 100-year nized as material from the start. But without the historical
event—or it might have been as high as 25% (though this data to evaluate such risks, risk managers cannot model them
seems unlikely). But, unless we can evaluate those probabili- using conventional quantitative approaches.
ties in a way that doesn’t depend mainly on hindsight, we In such cases, if probability assessments are made by risk
cannot pass judgment on risk management. managers, they are bound to have significant elements of
As already noted, when an institution has many positions subjectivity. Different risk managers can reach very different
or projects, the risk of the institution depends on how the conclusions. When that happens, statistical risk measure-
risks of the different positions or projects are related. If the ment reaches its limits and risk management changes from
correlation between the positions or projects is high, it is more science to art. And once risk management moves away from
likely that all the firm’s activities will perform poorly at the established quantitative models, it becomes easily embroiled
same time, which leads to a higher probability of a large loss. in intra-firm politics. At that point, the outcome depends
These correlations can be difficult to assess—and they can much more on the firm’s risk appetite and culture than its
change over time, at times abruptly. One of the partners of risk management models.
LTCM attributed much of the fund’s losses in August and
September of ’98 to sudden sharp increases in correlations Mismeasurement Stemming from Overlooked Risks
that were previously thought to be extremely small. Now let’s turn to a different kind of measurement failure—
Partly as a result of the experience in 1998 of funds like failure to take account of risks. In my taxonomy, such
LTCM, it is now well-known that correlations increase in “ignored” risks can take two different forms that have differ-
periods of crisis. And to the extent risk managers under- ent implications for a company. First, a firm’s risk managers
estimate the correlations between asset classes, they may may ignore a known risk, perhaps because of a mistaken
understate the risk of a portfolio or a company. In most cases, assumption that it is immaterial, or because of the difficulty
however, the problem of mismeasurement of correlations is of incorporating it in the risk models. Second is the case of
more subtle, since correlations evolve randomly over time and risks that are truly unknown, or at least completely unan-
sometimes jump unexpectedly. In such cases—and LTCM ticipated.

Journal of Applied Corporate Finance • Volume 20 Number 4 A Morgan Stanley Publication • Fall 2008 43
 17456622, 2008, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1745-6622.2008.00202.x by University Of South Africa, Wiley Online Library on [30/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Ignored Known Risks firm’s risk. Complicating matters, such risks could turn out
Let’s now consider the case of risk management failure in to be highly correlated with both credit and market risks. For
which material risks are known (to at least some people instance, for many banks the loss of income from securitiza-
inside the firm), but not reflected in the firm’s risk models. tion was the realization of a business risk that turned out to
In a well-functioning, truly enterprise-wide risk management be correlated with not only market risk—namely, the loss in
system, all major risks would be identified, monitored, and value of the securities issued through securitizations—but
managed on a continuous basis. In practice, of course, many also credit risk—the inability to use securitization to lay off
corporate risks tend to be managed in decentralized “silos,” the risks associated with loans.
which increases the possibility that known material risks are Thus, accounting for all the risks in a risk measurement
excluded from the central risk modeling process. When this system is a difficult and costly undertaking. But a failure or
happens, risks that are not captured by the system are not refusal to attempt it means that the firm’s top executives are
adequately monitored. And given the tendency of unmoni- managing the company with blinders on—they see only part
tored risks to expand within large organizations, such risks of the big picture they need to manage effectively.
can become material if they weren’t before. There are well-known examples of large losses from risks
For example, consider a trader whose risks are only partly missed due to an incomplete inventorying of the firm’s risks.
measured and monitored. Most traders have a compensation One of the best involved the former Union Bank of Switzer-
formula that involves an option-like payoff: they receive a land. In the second half of the 1990s, the bank was putting
significant share of the profits they generate, but do not have together risk management systems that would aggregate
to pay back the losses. Such compensation creates incentives risks within its trading operations. One group of traders that
for traders to take risks. If only some of the risks of traders are focused on equity derivatives was extremely successful. But
monitored, they can increase their expected compensation by because this group was using different computers from the rest
increasing the risks that are not monitored. Another example of the bank, integrating their systems with the bank’s would
is the situation where new financial instruments are not yet have required the group to make changes in computers that
incorporated in risk models. Some firms prohibit trading of would have involved some costs and downtime. Eventually,
financial instruments unless they are properly modeled—but the bank’s top management decided that it was more impor-
such a choice has costs since it is often extremely valuable for tant to allow the traders to continue making money than
traders to be active at the beginning of a new market. However, disrupt their operations. Soon after this decision, this group
if financial instruments are not included in the models, traders of traders suffered a large loss—one that was partly responsible
are tempted to build up positions in these instruments, creating for the bank’s forced merger with another Swiss bank.7
the possibility of large losses.
In financial institutions, risks are commonly divided into Unknown Risks
three categories: market, credit, and operational risks. These Most unknown risks do not create risk management prob-
distinctions are partly artificial in the sense that they are lems. To see this, let’s go back to the statistical model of
driven by regulatory considerations. Although the trading risk measurement for an individual stock (as opposed to a
books of financial firms are typically marked-to-market, their portfolio of stocks). Suppose a risk manager has modeled the
credit books use accrual accounting. But securities held in return of a stock using the normal distribution and has no
trading books can have credit risk as well as market risk. For reason to believe that future returns will come from a differ-
instance, a bank might have credit default swaps in its trading ent distribution than the one that held in the past. According
book. In some credit default swaps, the bank is the purchaser to this model, each period’s stock return will be a random
of protection. But whether the bank actually receives the outcome—like the outcome of the toss of a coin. Since it will
expected compensation in the event of a default depends on come from a known distribution, the risk manager does not
the ability of the protection seller to pay that compensation. need to know why the return of the stock was 10% in one
And because of this somewhat artificial distinction between period and –15% in another. He has captured the relevant
credit and market risk, a bank may fail to adequately take risk characteristics of the stock through his estimation of the
into account the counterparty risk of the credit default swaps statistical distribution of the returns of the stock. His model
it holds on its books. tells him that the volatility (expressed as a standard deviation)
Similarly, normal business risks—which can be quite of the stock’s return is 20% per year, and that there is a 5%
different from those covered under the Basle II rules’ narrow chance of a loss of, say, 30% or more over that period. He
definition of operational risk—are often of critical importance does not need to understand or explain why a loss took place.
and have to be carefully assessed as part of the evaluation of a If the stock drops by 15% because of the manifestation of

7. See Dirk Schütz, La Chute de l’UBS: Les raisons du declin de l’Union de Banques
Suisses, Bilan, 1998.

44 Journal of Applied Corporate Finance • Volume 20 Number 4 A Morgan Stanley Publication • Fall 2008
 17456622, 2008, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1745-6622.2008.00202.x by University Of South Africa, Wiley Online Library on [30/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
some specific risk, the fact that the risk manager did not know retaining or laying off risks. But for managements and boards
about this specific risk is generally irrelevant in risk manage- to understand such risks and their consequences for the firm’s
ment as long as the risk manager has a good understanding operations, risk managers must communicate effectively. Even
of the distribution of the stock’s return. For instance, this risk if a firm has the best possible risk systems, if the risk manager
manager could not be faulted for having failed to anticipate is unable to make the top management understand their
the possibility of a fire destroying a factory and resulting in output, the systems may do more harm than good. If viewed
a 15% loss in that stock. as a “black box,” such a system could either lead to distrust
Unknown risks may not matter simply because they have and excessive conservatism on the part of top management—
a very low probability. For example, there is some probability or excessive risk-taking stemming from an exaggerated sense
that a building will be hit by an asteroid. But such a risk of the protection risk management can provide.
does not affect managerial decisions. The unknown risks Communication failures appear to have played a role in
that matter for our purposes are those that, had top manage- the recent crisis. For example, UBS published a report for its
ment been aware of them, would have resulted in different shareholders in which it discusses the causes of its subprime-
actions. For example, the risk of a drop in housing prices of related write-downs. In that report, it notes that
30% from peak to trough could well have been viewed as an
unknown risk in the past. But now, of course, that possibility  a number of attempts were made to present subprime or
…
has materialized in a number of regions. Had people been housing related exposures. The reports did not, however,
aware of that risk a few years back, many decisions would communicate an effective message for a number of reasons,
likely have been different. in particular because the reports were overly complex,
In sum, part of the risk manager’s job is to think hard presented outdated data or were not made available to the
about all major sources of uncertainty, perhaps using scenario right audience.” (p. 39).
planning (a suggestion I come back to later) as a methodi-
cal way of identifying and anticipating unknown risks. But And as stated in a report by the Senior Supervisors Group
having said that, some risks will inevitably remain unknown (which includes top regulators from the U.S., England, and
until they actually materialize. After all, how many corporate Germany), “In some cases, hierarchical structures tended to
risk managers anticipated the events of 9/11? And acknowl- serve as filters when information was sent up the management
edging this possibility of unknown and unknowable risks, chain, leading to delays or distortions in sharing important
risk managers at some point must concede the limitations data with senior management.”8 Finally, in the words of an
of their models—and perhaps recommend that additional industry commission report on the crisis, “risk monitoring and
capital be set aside for this possibility. management reduces to the basis of getting the right infor-
mation, at the right time, to the right people, such that those
Communication Failures people can make the most informed judgments possible.”9
Risk management is not an activity undertaken by risk
managers for the benefit of risk managers. Nor should it ever Failures in Monitoring and Managing Risks
be viewed as an operation or profit center unto itself. Instead, Risk management is responsible for making sure that the firm
it should be designed to help the firm maximize value by takes the risks that it wants to take and not others. As a result,
“protecting” its ability to make the optimal investment and risk managers must constantly monitor the risks the firm
operating decisions. is taking. Further, they have to hedge or otherwise manage
But, as noted earlier, it is not the job of risk manage- known risks to meet the objectives of top management.
ment to determine the firm’s overall target level of risk or We have already discussed the problem that a firm may be
the kinds of risks it takes. Those decisions are the purview taking risks it does not know about. When we discussed that
of top management and the board—of people who, when problem, we focused on it as an “inventory” issue—that is,
setting the firm’s strategy, should begin by identifying the as a matter of identifying and measuring risks that are fairly
firm’s competitive advantages and the risks its investors are stable. Such an approach is likely to be well suited to industrial
“paying” the firm to take. (For example, should the airlines companies, where risks tend to change slowly. But it may have
really be in the business of taking oil price risk, or should that to be modified for financial firms, where risks can change
be laid off to commodity traders?) abruptly even if the firm does not take new positions.
The role of risk management in such strategic decision- The problem arises from the fact that financial firms
making is to provide timely information to the board and top have many derivatives positions and positions with embed-
management that allows them to assess the consequences of ded derivatives. The risk properties of portfolios of derivatives

8. Senior Supervisors Group, “Observations on Risk Management Practices during the 9. “Containing systemic risk: The road to reform,” The Report of the CRMPG III, Au-
Recent Market Turbulence,” March 6, 2008, p. 9. gust 6, 2008.

Journal of Applied Corporate Finance • Volume 20 Number 4 A Morgan Stanley Publication • Fall 2008 45
 17456622, 2008, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1745-6622.2008.00202.x by University Of South Africa, Wiley Online Library on [30/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Figure 1 Evolution of ABX AAA Tranches Based on Subprime
Residential Mortgage-backed Securities (RMBS)

120

AAA 2006-1 AAA 2007-1

100

80
Price

60

40

20

0
1/19/06 4/19/06 7/19/06 10/19/06 1/19/07 4/19/07 7/19/07 10/19/07 1/19/08 4/19/08
Timeline

can change very rapidly with no trading whatsoever. This When the risk characteristics of securities can change
is because complex derivatives often have exposures to risk this rapidly, it is challenging for risk monitors to capture
factors that are extremely sensitive to market conditions. For these changes and risk managers to adjust their hedges.
example, we now have products that can start the day with a This challenge is especially great when risk can change
significant positive exposure to an interest rate increase but, as dramatically in response to small changes in the determi-
a result of a small change in rates, end the day with a signifi- nants of security prices. As a result, risk managers may fail
cant negative exposure. For such products, hedges adjusted to adequately monitor or hedge risks simply because the risk
daily could end up creating large losses because the hedge characteristics of securities may change too quickly to allow
that was optimal at the start of the day could be amplifying the managers to assess them and put on effective hedges.
risk at the end of the day. Especially in these circumstances, an important compo-
One of the most obvious demonstrations of how risk nent of risk management is to identify possible solutions that
exposures can change is the pricing of subprime derivatives. can be implemented quickly if a firm has to reduce its risk
The ABX indices, which are indices based on a basket of credit- over a short period of time. Contingency hedging plans are
default swaps written on subprime securitization tranches, therefore critical for responding to unexpected difficulties.
have been the most readily available source of data on the Moreover, it’s important to recognize that when liquidity
value of securities issued against subprime mortgage collateral. dries up in the markets, many risk-reducing options that are
New indices have been created every six months that reflect effective in normal periods can no longer be used.
new securitizations. Initially, the AAA indices, which represent Paradoxically, the introduction of mark-to-market
the pricing of credit default swaps on AAA-rated tranches, accounting has made it even harder for risk managers to
showed almost no variation. And so reasonable assessments estimate risk and put on effective hedges.10 In many ways,
of the risk of the AAA-rated tranches using historical data mark-to-market has introduced the Heisenberg Principle
would have indicated little risk. But, as can be seen in Figure into financial markets: For large organizations, observing the
1, in the second-half of 2007, the value of these securities fell value of a complex security affects the value of that security.
off a cliff. Holders of AAA-rated tranches of subprime securi- The reason for this is straightforward: When mark-to-market
ties reported sudden large losses if they chose to use the ABX losses become known, they set off a chain reaction of adjust-
indices as proxies for the value of their holdings. ments at other institutions and affect the prices of possible

10. For a discussion of some of the issues concerning mark-to-market accounting that Song Shin, Marking to Market: Panacea or Pandora’s Box?, 2008, Journal of Accounting
accounts for possible feedback effects, see Guillaume Plantin, Haresh Sapra, and Hyun Research 46, 435-460.

46 Journal of Applied Corporate Finance • Volume 20 Number 4 A Morgan Stanley Publication • Fall 2008
 17456622, 2008, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1745-6622.2008.00202.x by University Of South Africa, Wiley Online Library on [30/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
trades as the market better understands the capital positions crisis and to have a strategy for responding to it.
of different institutions. Risk models, as we have already seen, are generally not
In large complex organizations, it is also possible for designed to capture risks associated with crises and help
traders to take risks that remain hidden for a while. Such companies manage them. The models use historical data
positions could be constructed with securities whose risks and, particularly when using risk measures such as VaR, are
escape detection, at least for a time, by even the best designed most precise for horizons that are numbered in days; and
risk management models and systems. How can this happen? when using such short horizons, crises appear to be highly
All organizations are forced to make tradeoffs, and risk improbable events. But, when the horizon expands to years,
management is no exception. A company’s risk management the probability of a crisis becomes material, something clearly
function could, at least in theory, be designed to know every- worth management’s attention.
thing at all times. But if it were organized that way, the risk And at least for purposes of evaluating the consequences
management function, besides being hugely costly, would of financial crises and planning for them, an extension of the
likely stifle innovation and reduce the competitiveness of the horizon of risk management models is likely to be valuable
firm. In fast-moving markets, employees need flexibility. for at least two reasons: First, as illustrated dramatically by
But, as suggested, that same flexibility makes it possible events since the summer of 2007, financial crises can involve
for unobserved pockets of risk to emerge. And when such the abrupt withdrawal of liquidity from the markets. And
risks manifest themselves, it is not clear that they represent a the absence of liquidity means that firms are stuck with
risk management failure. Although risk management could positions they did not expect to hold. Positions whose risk
have ensured that these risks were not taken, the firm and its was evaluated over a single day suddenly became positions
shareholders would have been worse off. And even ignoring that had to be held for weeks or months. Second, during
its effects on flexibility and innovation, improvements in risk crisis periods, companies will repeatedly experience losses that
monitoring are expensive enough that, at some point, they exceed their daily VaRs, substantially weakening their own
cease to be cost-effective. capital positions. For these reasons, even firms whose trading
One final point here: The effectiveness of risk monitor- is guided mainly by daily VaR should consider complement-
ing and control depends crucially on an institution’s culture ing their use with longer-term measures.
and incentives. If risk is everybody’s business, it is harder for As we have also seen in the case of LTCM, financial crises
major risks to go undetected and unmanaged. Moreover, if involve complicated linkages and interactions among risks
employees’ compensation is affected by risks, they will be and institutions. Statistical risk models typically take returns
more judicious in taking them. The best risk models in a firm to be “exogenous” to the firm and ignore risk concentrations
with limited incentives are likely to be much less effective across institutions. But if such an approach is appropriate
than in a firm where the incentives of employees are better for many institutions, it will not work for others that, because
aligned with the risk-taking objectives of the firm. of their size or degree of connectedness, can affect prices and
volumes in certain markets. For instance, it is well-known
Lessons from Failure (or How to Prepare that LTCM had extremely large short positions in the index
for the Next Crisis) option market. During the crisis, it had little ability to
To come up with useful assessments of risk, risk managers adjust these positions because they were such a large part of
must look at longer horizons and take a comprehensive view the market.
of their risks. For example, one-year horizons are widely used Further, large institutions facing distress can be exposed
as measures of firm-wide risk in enterprise-wide risk manage- to “predatory” trading—that is, trades made by others that
ment (ERM) programs. Most financial institutions that focus have the effect of compounding the losses from such large,
on one-year measures of firm-wide risk aim for credit ratings illiquid positions. An example is a situation where traders
that imply an extremely small yearly probability of default— from other institutions benefit from selling and pushing
say, 0.03%. Such approaches are useful not only in assessing prices down with the aim of bringing about a fire sale (and
a firm’s risk, but also in estimating the optimal amount and buying back their positions). As one firm experiences large
cost of capital both for the entire firm and its various projects losses, it may drag down prices for other institutions and
and lines of business. make funding more costly for all of them. Typical risk
But such approaches are not sufficient. A high target credit management models do not account for such possibilities,
rating effectively means that the firm tries to avoid default in leading to a significant understatement of the risk of positions
all but the most extreme circumstances. When a firm aims in the event of a crisis.
for an AA credit rating, it effectively chooses a probability of But, as already suggested, there is little hope for statisti-
defaulting of roughly one in 1,000 in the coming year. But cal risk models that rely on historical data to capture such
since crises are more likely than that, management needs to complicated effects. And rather than attempting to introduce
understand the implications of its decisions in the event of a greater complexity and realism into their models, risk manag-

Journal of Applied Corporate Finance • Volume 20 Number 4 A Morgan Stanley Publication • Fall 2008 47
 17456622, 2008, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1745-6622.2008.00202.x by University Of South Africa, Wiley Online Library on [30/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ers should consider complementing their reliance on models effective guidance to top management and boards. And once
with tools such as scenario analysis to investigate how crises top management has used that information to determine the
unfold, how the firm will be affected by them, and how it firm’s risk appetite and strategy, risk management can also
should best respond to them. With the help of such scenarios, fail to monitor risks appropriately and maintain the firm’s
top managements can anticipate the threat posed by extreme targeted risk positions.
events to the franchise value of their institutions and develop But if risk management has been mistakenly identified
strategies for limiting and responding to such events. as the culprit in many cases, there is no question that risk
But contrary to some current practice, such a scenario management practice can be improved by taking into account
approach must be informed by economic and financial the lessons from financial crises past and present. Such crises
analysis. It cannot be done by risk management departments have occurred with enough frequency that crisis conditions
populated only by physicists and mathematicians. Such an can be modeled, at least to some extent. And when models
approach also cannot succeed unless top management believes reach their limits of usefulness, companies should consider
that the scenarios considered represent serious threats to the using scenario planning that assesses the implications of crises
institution—serious enough to affect the formulation of the for their financial health and survival. Rather than relying
firm’s strategy and to require a plan to deal with it. on past data, scenario planning must use economic analysis
to evaluate the expected impact of sudden illiquidity and
Conclusion the associated feedback effects that are common in financial
The difficulties of the past year have convinced some observ- crises. But, to serve as an effective part of a firm’s strategy,
ers that there are important, and perhaps irremediable, flaws scenario planning and the analysis that comes out of it must
in risk management, and that such flaws were major contrib- be deeply rooted in a firm’s culture as well as the strategic
utors to the current financial crisis. In this paper, I show thinking of top management.
the need to distinguish between flawed assessments by risk
managers and corporate risk-taking decisions that, although
resulting in losses, were fundamentally reasonable at the time rené m. stulz holds the Reese Chair of Banking and Monetary Economics
they were made. To help in making this distinction, the paper at Ohio State’s Fisher College of Business. He is also a research associate
also identifies a number of different ways that risk manage- at the National Bureau of Economic Research and a fellow at the European
ment can fail. In addition to choosing the wrong risk metrics Corporate Governance Institute. He belongs to the Board of Trustees of
and otherwise mismeasuring risks, risk managers can fail to the Global Association of Risk Professionals and its executive committee.
communicate their risk assessments and otherwise provide

48 Journal of Applied Corporate Finance • Volume 20 Number 4 A Morgan Stanley Publication • Fall 2008
 17456622, 2008, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1745-6622.2008.00202.x by University Of South Africa, Wiley Online Library on [30/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Applied Corporate Finance (ISSN 1078-1196 [print], ISSN Journal of Applied Corporate Finance is available online through Synergy,
1745-6622 [online]) is published quarterly, on behalf of Morgan Stanley by Blackwell’s online journal service, which allows you to:
Blackwell Publishing, with offices at 350 Main Street, Malden, MA 02148, • Browse tables of contents and abstracts from over 290 professional,
USA, and PO Box 1354, 9600 Garsington Road, Oxford OX4 2XG, UK. Call science, social science, and medical journals
US: (800) 835-6770, UK: +44 1865 778315; fax US: (781) 388-8232, • Create your own Personal Homepage from which you can access your
UK: +44 1865 471775. personal subscriptions, set up e-mail table of contents alerts, and run
saved searches
Information for Subscribers For new orders, renewals, sample copy requests, • Perform detailed searches across our database of titles and save the
claims, changes of address, and all other subscription correspondence, search criteria for future use
please contact the Customer Service Department at your nearest Blackwell • Link to and from bibliographic databases such as ISI.
office (see above) or e-mail [email protected]. Sign up for free today at http://www.blackwell-synergy.com.

Subscription Rates for Volume 19 (four issues) Institutional Premium Rate* Disclaimer The Publisher, Morgan Stanley, its affiliates, and the Editor cannot
The Americas† $377, Rest of World £231; Commercial Company Premium be held responsible for errors or any consequences arising from the use of in-
Rate, The Americas $504, Rest of World £307; Individual Rate, The Ameri- formation contained in this journal. The views and opinions expressed in this
cas $100, Rest of World £56, €84‡; Students** The Americas $35, Rest of journal do not necessarily represent those of the Publisher, Morgan Stanley,
World £20, €30. its affiliates, and Editor, neither does the publication of advertisements con-
stitute any endorsement by the Publisher, Morgan Stanley, its affiliates, and
*The Premium institutional price includes online access to current content Editor of the products advertised. No person should purchase or sell any
and all online back files to January 1st 1997, where available. security or asset in reliance on any information in this journal.

†
Customers in Canada should add 6% GST or provide evidence of entitlement Morgan Stanley is a full service financial services company active in the
to exemption. securities, investment management, and credit services businesses. Morgan
Stanley may have and may seek to have business relationships with any
‡
Customers in the UK should add VAT at 6%; customers in the EU should person or company named in this journal.
also add VAT at 6%, or provide a VAT registration number or evidence of
entitlement to exemption. Copyright © 2008 Morgan Stanley. All rights reserved. No part of this publi-
cation may be reproduced, stored, or transmitted in whole or part in any form
**Students must present a copy of their student id card to receive this rate. or by any means without the prior permission in writing from the copyright
holder. Authorization to photocopy items for internal or personal use or for
For more information about Blackwell Publishing journals, including online the internal or personal use of specific clients is granted by the copyright
access information, terms and conditions, and other pricing options, please holder for libraries and other users of the Copyright Clearance Center (CCC),
visit www.blackwellpublishing.com or contact your nearest Customer Service 222 Rosewood Drive, Danvers, MA 01923, USA (www.copyright.com), pro-
Department. vided the appropriate fee is paid directly to the CCC. This consent does not
extend to other kinds of copying, such as copying for general distribution
Back Issues Back issues are available from the publisher at the current single- for advertising or promotional purposes, for creating new collective works,
issue rate. or for resale. Institutions with a paid subscription to this journal may make
photocopies for teaching purposes and academic course-packs free of charge
Mailing Journal of Applied Corporate Finance is mailed Standard Rate. Mail- provided such copies are not resold. Special requests should be addressed to
ing to rest of world by DHL Smart & Global Mail. Canadian mail is sent Blackwell Publishing at: [email protected].
by Canadian publications mail agreement number 40573520. Postmaster
Send all address changes to Journal of Applied Corporate Finance, Blackwell
Publishing Inc., Journals Subscription Department, 350 Main St., Malden,
MA 02148-5020.