The high returns attained by well known hedge fund managers such
as George Soros have attracted the attention of investors enamored
by the 20%, 30%, 50% returns, as well as, the ever so often reported
100% returns attained by "macro" hedge fund managers. What is even
more disconcerting, is a 1998 study by Montgomery Securities that
revealed investors expect macro managers to return over 30% annually.
As astonishing as these expectations may be, these fund managers
and their returns are not indicative of the hedge fund industry
as a whole. The question remains "can the risks in hedge funds be
quantified given their unique nature?" Value-at-Risk (VAR) is widely
accepted by the financial markets as a risk measurement standard.
It is a model acclaimed for its ability to measure risk across different
financial instruments. After careful examination of the underlying
assumptions used to calculate VAR and the characteristics of hedge
funds, we have concluded that VAR may not be the appropriate
risk measurement model to employ for all hedge funds. This report
will highlight some of the pitfalls of VAR when used to measure
hedge fund risk. Value-at-Risk was employed by Long-Term Capital
Management (LTCM); clearly it did not work…but why?
Hedge funds are private investment partnerships within the "alternative
investment" asset class. The investment manager typically has market
skills, career experience and a track record that is exemplary and
often unique. Hedge funds are similar to mutual funds in that they
are actively managed investment "pools" of capital holding positions
in publicly traded securities. They differ from mutual funds in
their investment and trading strategies, use of a broad range of
securities, high performance criteria, and risk/return objectives.
Hedge funds often seek to invest in more complex markets, riskier
securities and pursue more aggressive investment or trading strategies.
Because of this flexibility to invest in a variety of markets, hedge
fund managers may not seek to outperform a benchmark, such as the
S&P 500 Index, but rather they seek to exploit mis-pricing in the
global financial markets.
Originally hedge funds were established as investment vehicles
that attempted to hedge market risk in the portfolio. A common strategy
would be to buy long positions in undervalued securities and simultaneously
sell short positions in overvalued securities, thus creating a "hedged"
position. The result is an effective "bet" that the long position
will appreciate while the short position did nothing in bull markets;
but more than offset losses in the long position in bear markets.
This long and short stock selection, when executed properly,
hedges out market risk.
Hedge funds tend to employ dynamic trading strategies. A dynamic
trading strategy may utilize any combination of the following: leverage;
concentration; market timing and market trending strategies; short
selling of securities; event driven; and convergence or divergence
between two or more securities.
Modern portfolio theory indicates that: (a) investors should
be compensated for risk and; (b) asset diversification reduces portfolio
volatility or risk.1 Hedge
funds offer both of these advantages to investors. Investors have
the ability to add riskier, and therefore higher returning assets
to a portfolio while, at the same time, reducing the volatility
or risk of their portfolio.
Over the last several years risk management has become the focus
of financial markets participants. Risk management is the process
of identifying and managing potential risks an entity faces during
the course of business. Although the main impetus of risk management
began with banks looking at risk in their own portfolios, risk management
has expanded to other parts of the financial industry. Now investment
managers are being required to explain investment strategies, demystify
their models, and become more open about their portfolio risks.
There are several motivations behind the push for transparency
of risk exposures. First, are the incidences of billion dollar losses
stemming from risk management negligence by several financial entities.
The most notable is the $ 1.5 billion dollar loss sustained by the
Orange County Investment Pool in December 1994.2
In 1998, Long-Term Capital Management [LTCM] lost almost
$3 billion of investor capital. Public attention and debates of
these incidences have made individual investors more sophisticated
about the numerous risks inherent in financial markets.3
Second, regulators of financial markets have addressed the issue
of risk measurement and have made full disclosure mandatory in some
public sectors. Third, and perhaps most important, the flow of institutional
money into hedge funds is growing at a steady pace. Pension funds
are allocating more money to hedge funds. They are also required
by law to have sufficient knowledge of the risk exposure incurred
by their portfolios. Hedge funds have been scrutinized by Congress.
Disclosure by hedge funds to their creditors and partners was at
the heart of the 1998 Congressional Hearings regarding LTCM and
the hedge fund industry.
The Securities and Exchange Commission (SEC), Bank for International
Settlements, and the Global Derivatives Study Group are among financial
market regulators and study groups which have developed and recommended
risk management standards and policies, as well as, procedures for
risk measurement. One quantitative model that has received numerous
endorsements from the financial markets industry to address risk
measurement and support risk management mandates is Value-at-Risk.
Value-at-Risk (VAR) is a method of measuring the financial risk
of a portfolio over some specified time period. It provides a consistent
approach to valuing asset exposures using probability analysis based
upon a consistent confidence level and time horizon4.
VAR is a rather simple model to manipulate but is extremely dependent
on the assumptions and methodology used. It offers an adequate approximation
of how much a portfolio or firm could lose from changes in the prices
of assets it holds5.
Numerous industry participants have found VAR extremely useful in
analyzing their business and market risks. Mutual fund managers
have incorporated VAR into their risk management process and disclose
this information to investors. An in depth study of the practicality
of hedge fund managers using VAR to measure their market risk exposures
is beyond the scope of this paper. Instead, we will examine the
usefulness of a VAR calculation to an investor who is evaluating
and comparing hedge funds. This paper will also evaluate the accuracy
of this risk approximation when combined with a traditional portfolio
of stocks and bonds.
VAR is an estimate of the expected maximum loss over a target time
horizon within a given confidence level under assumed market conditions.
Introduced in 1993, among the reasons VAR has gained popularity
is its ability in measure risk across different financial instruments.
In general, it can be used to calculate risks implicit in price
movements for equities, commodities, bonds and currency exposures
provided there is sufficient data on assets and risk factors from
which to calculate the probabilities and correlations of future
price changes. VAR can capture the combined effect of a portfolio's
underlying volatility and its exposure to market risks. VAR is a
forward looking market risk indicator.
A VAR calculation is dependent upon the selection of two quantitative
factors: the length of the risk horizon and the confidence level
desired. The target risk horizon is the length of time a manager
needs to evaluate risk exposure. For instance, a manager can calculate
risk exposure over a daily, monthly, or quarterly time interval.
The proper selection of a risk horizon varies among industry users
and is related to performance evaluation, trading strategies, and
the liquidity of securities within a portfolio. The confidence level
is the degree of confidence that one can place that a loss will
not exceed the VAR indication. For instance, at a 95% confidence
level, a manager can expect (based on historical probabilities)
to lose no more than the VAR measure 95% of the time over the risk
Once the risk horizon and confidence level are selected an investment
manager must generate a probability distribution of possible changes
in the portfolio value over the risk horizon. Price and return distributions
for each security is generated. These distributions represent the
price or returns for each security over the risk horizon. Therefore,
the "value at risk" of a portfolio is the dollar loss corresponding
to a pre-defined probability or "confidence level." Here in lies
the "achilles heel" of value-at-risk modeling.
ASSUMPTIONS USED BY VAR
In general, some VAR models are based upon several key assumptions6:
(a) returns are, in statistical terms, normally distributed; (b)
markets are liquid; (c) historical and current market price data
are available; (d) correlation data are either available or easily
calculated from historical or current market data; (e) securities
and portfolios are static over the risk horizon. Based upon these
assumptions, VAR is not an appropriate risk measurement model for
most hedge fund managers to employ.
VAR calculations assume that portfolio assets and risk/return
profiles are static over the time horizon. Volatilities in illiquid
and inefficient markets, such as emerging markets, may prove difficult
to model using VAR. There may be very little historical data and
risk profiles may change over time making price returns difficult
to forecast. Losses in these markets can be steep and sudden. The
Asian crisis is a perfect example. Inaccurate information
about debt exposures companies had on their balance sheets led to
inaccurate market valuations. Asian stock markets lost over 50 percent
of their original value and currencies were devalued by more than
50 percent by the end of 1997. Asia's risk profile and volatility
changed literally overnight. "One off" events, such as the Russian
debacle in August 1998 can also position the VAR approach ineffective.
Some hedge funds employ market or event driven strategies. These
strategies perform well when the market is adversely affected, as
they did in the stock market crash of 1987. VAR calculates the expected
maximum loss over a target time horizon within a given confidence
level under "assumed" market conditions. Turning the market upside
down we can assert that a hedge fund that follows a market event
driven strategy is safe in adverse market conditions. A traditional
mutual fund performs well during normal market conditions and most
event driven strategies perform well during abnormal market conditions.
Different assumptions must be made to properly calculate a VAR for
both of these strategies. However, once the hedge fund's risk measure
is calculated: (a) how comparable is this number to other VARs;
and (b) how can an investor incorporate this measure into their
traditional portfolio VAR? Can these numbers be combined to formulate
a total VAR for an investor's portfolio?
To complete the story on risk measurement, there are a number
of risks in the financial markets that can not be captured in VAR
or any quantitative model. Most notable for hedge funds are liquidity
risk, political risk, model risk, and technology risk. Long-Term
Capital VAR models were ineffective because of extremes that occurred
in each of these areas, particularly August 1998.
Risk management is a process meant to engage investment managers
and investors in the constant evaluation of risk/reward performance.
A risk management model is only one aspect of this process. Value-at-Risk
is a summary statistic that quantifies the exposure of an asset
(or portfolio of assets) to risk; or the risk that a position declines
in value with adverse market price changes. It is a powerful technique
that has improved the evaluation and understanding of market risk,
but it is an expectation of outcomes based on specific assumptions.
VAR, depending on its assumptions, may be difficult to implement
on complex asset return characteristics and dynamic investment strategies
employed by hedge fund managers, especially macro managers and managers
using illiquid securities and illiquid derivatives. VAR is more
accurate in traditional liquid markets, however, hedge funds tend
to invest in illiquid and complex markets. Conclusively, VAR did
not work for Long-Term Capital Management nor Orange County. It
did nothing to prevent massive portfolio losses during the Asian
crisis and the Russian debacle. At best, VAR is a rear-view mirror
approach future risks.
Hedge funds are a unique asset class that has received attention
for their impressive returns. The risks, however, can be equally
striking. These returns must not enamor investors. Investors should
seek hedge funds that provide the necessary risk and return objectives
that complement their traditional assets. Great care must be taken
to fully explore and quantify risk. To complement any risk measurement
model, it is strongly suggested that investors perform in-depth
due diligence on a case-by-case basis. Individual investors may
also want to seek hedge fund advisors who have experience in analyzing
hedge fund risk/reward components.
Asset diversification reduces risk because all securities and markets
are not correlated; that is, different markets and security classes
react differently to changes in market conditions. Securities can
either move in opposite directions or at a different pace to be
not correlated to each other.
For a detailed discussion of Orange County Investment Pool loss
see for example, Philippe Jorion, "Lessons From the Orange
County Bankruptcy", Journal of Derivatives Vol. 4, No. 4
(Summer 1997): pp 61-66.
There are a number of articles and books that address and fully
explain the derivative losses of 1993 - 1995 (Proctor & Gamble,
Barings PLC, and Orange County).
Although there is quite a bit of literature on VAR for a general
description, see Philippe Jorion, Value at Risk (Chicago:
Irwin Professional Publishing, 1997)
We are using a very simple explanation of VAR for the purposes of
this paper. There are three different VAR models that use different
assumptions and methods for calculating risk measures. For a more
in depth explanation see for example Tanya Styblo Beder,
"Report Card on Value at Risk: High Potential but Slow Starter",
Bank Accounting & Finance.
As previously stated for more a more in-depth discussion of VAR
models and assumptions see for example Tanya Styblo Beder,
"VAR: Seductive but Dangerous" Financial Analysts Journal:
September/October 1995: 12-24.