Beruflich Dokumente
Kultur Dokumente
Investors take risks to deliver excess returns over their benchmarks. The most
recent crisis in the financial markets has re-emphasized the need for practical
principles to evaluate what risks are acceptable for long-term performance
and which ones are better hedged out. This paper summarizes ten operational
rules for tail-risk management of investment portfolios that investors ought to
consider when constructing portfolios, especially those that have potential for
impact from adverse credit events.
1 INTRODUCTION
No risk management term has entered the vernacular of investors as rapidly as tail-
risk management has in the last three years. Market participants have not only had
to deal with a substantial loss to their wealth, but also to their confidence, and most
of all to many core principles of investing that they have held dear almost forever.
What has the crisis taught us about investment risk management?
The most important lesson that we have learnt is that, although the details of this
crisis are different, in many ways the current episode has striking similarities to past
crisis events. Among these similarities is the central role played by deleveraging,
illiquidity and flight to quality, with a spike in the degree of comovement of assets
and the inability of investors to anticipate the degree of damage to portfolios.
This crisis has also taught us that well-positioned portfolios are not only those
that outperform others in periods of normalcy, but those that can weather storms.
Whether one can anticipate the exact nature of these rare events is doubtful; but
just as homes in areas that are prone to natural disasters are designed to withstand
significant stresses, portfolios in increasingly accident-prone markets are designed
with structural risk control to survive severe market catastrophes. While the nature
of these catastrophes changes with time, there are usually sufficient opportunities to
fortify portfolios against the most perverse crises at a reasonable and relatively small
I would like to thank PIMCO colleagues and clients who saw the need for tail-risk hedging when
others did not, and encouraged the development and implementation of our thoughts ahead of the
subprime crisis of 2007.
115
i i
i i
i i
116 V. Bhansali
i i
i i
i i
also rise in their absolute value. This provides a free lunch, in that a completely
disconnected macro market in normal times becomes a good hedge against the tails
in distressed times. It is more efficient to sacrifice basis risk and close matching of
hedges in order to obtain efficiency of hedges. For example, due to the beta of credit
to other markets, credit market tail-risk events are frequently hedged better with
equity options and currency options than with customized credit hedge instruments.
Indeed, this blurs the distinction between an active position and a hedge, but when
the dust settles the classification of a position into ex post categories is less important
than the economic benefits it confers.
Credit market hedges were extremely cheap in 2007, because of the incessant sell-
ing of structured products and the demand driven by excess liquidity. These same
hedges are not as attractively priced today (though there are others that are). Thus
hedging credit with credit market instruments might no longer be an attractive solu-
tion. The estimation of option prices and sensitivities in general is based on the
academic foundations of continuous trading and risk-neutral investors. As I have
written previously (Bhansali (2007b)), these ideal conditions are rarely met in tur-
bulent markets. Hence using what are now traditional models of option pricing are
flawed by construction when applied to tail hedges. A viable alternative approach
is to run scenarios that investors are averse to and evaluate the expected value of
portfolios under those scenarios with and without insurance assets. The difference
in these two portfolio values is the proper price for the insurance from the investors
perspective and might be superior to the market price of the insurance. We all
know that catastrophe insurance can trade too cheaply in the natural insurance mar-
kets; it is possible for this to also happen in the world of finance, especially in a
world of innovative financial engineering that ports risks from one type of market
to another.
Why is it that cheap tail hedges can be found in almost all market environments?
There are many possible reasons. For example, speculative demand for particular
types and classes of assets may drive the price of these assets very high. In periods
of low returns, as observed until the middle of 2007, yield hoggers increased
options sales as a source of carry. The belief of mean-reversion participants is that
out-of-the-money options will never be exercised. This is generally true, except that
as the leverage in the market place increases due to everyone simultaneously doing
levered option sales, the notional size has to increase to generate the same carry.
At some point this type of system becomes unstable to small noise and creates a
domino effect of hedgers all trying to cover their deltas at the same time (the other
choice is to risk certain ruin).
i i
i i
i i
118 V. Bhansali
i i
i i
i i
FIGURE 1 Correlation of typical hedge fund strategies versus the VIX index for different
periods.
1.0
Positive correlation
(lower volatility, higher return)
0.5
0.5
1.0
Composite
hedge fund
Convertible
arbitrage
Dedicated
short bias
Emerging
markets
Equity market
neutral
Event
driven
Distressed
Risk
arbitrage
Fixed-income
arbitrage
Global
macro
Long/short
equity
Managed
futures
Multi-
strategy
Source: PIMCO.
i i
i i
i i
120 V. Bhansali
shocks. When everyone else is trying to put on the same hedges, there is no guarantee
that an investor will be able to do so before others. In the approach we are describ-
ing the portfolio is subjected to reasonable but rare super-shocks and insurance is
purchased using one of the four techniques previously listed. So there is an explicit
price to be paid for the hedge. The role of the portfolio manager is to reduce the
cost of these hedges over the hedge horizon. Especially when the horizon is long
enough, hedges can usually be bought very cheaply. For instance, long-dated options
on foreign exchange such as the US dollaryen exchange rate recently had negligible
cost over a one-year horizon due to the roll-up of forward rates from the interest
rate differential between US dollar rates and yen rates (see Bhansali (2007c)). In
periods of stress one common outcome is deleveraging and flight to low-yielding
currencies such as the yen. This directionality of the foreign exchange movement
and the associated increase in volatility makes the real-world expected value of the
hedging package much higher than the theoretical, risk-neutral value. So it is the
possibility of asymmetric payouts in particular states of the world that makes the
hedges worth more to the investor than their stand-alone theoretical value.
i i
i i
i i
(a)
This table shows the shock to the 7100 tranche of the IG8 CDX index and the resulting performance in points. In
(b) we shock the spreads of the index (displayed in first column). The second, third and fourth columns show the
performance of the hedge under different base correlation assumptions using a Gaussian copula model. Part (c)
repeats the exercise for a one-year horizon shock, so it incorporates time decay in the hedge.
perform better for increased correlation in periods of stress. The one-year horizon
shock accurately captures the cost or time decay as the tranche (which is nothing
but an option on losses) shortens in maturity.
i i
i i
i i
122 V. Bhansali
250
200
Spread (bps)
50 Idiosyncratic risk
0
Mar 04 Mar 05 Mar 06 Mar 07 Mar 08 Mar 09
The three lines are the idiosyncratic component, the economy-wide risk and the sectoral risk, as indicated. The
economy-wide risk is at highly elevated levels. This figure is from Bhansali et al (2008).
In recent work (Bhansali et al (2008)) we break down the index spreads for various
indexes using a three-jump approach first used by Longstaff and Rajan. The third
component of the spread, which corresponds to systemic risks, has clearly become
elevated in the recent crisis and may only revert to normal levels with a long lag
and after substantial liquidity is provided to alleviate insolvency fears. Regression
of other markets, such as municipals, asset-backed securities, etc, on the systemic
component of the spread shows significant dependence of prospective returns on
how the systemic risk factor changes. In other words, many asset classes carry large
amounts of systemic illiquidity risk. No wonder that, at the time of writing, municipal
bonds, especially general obligation bonds of natural AAA-rated states, are trading
at even higher yield levels than treasuries, despite the fact that they are tax exempt.
The impact of government action on systemic risks is obvious when we compare the
systemic and idiosyncratic components in Figure 2. Prior to the Bear Stearns event in
spring 2008, the systemic component had already started to rise and reached the same
order of magnitude as the idiosyncratic spread component. When the government
provided unprecedented liquidity this component of spreads started to narrow. With
the mid September failure of Lehman Brothers the authorities had to make a choice,
in which they sacrificed the interests of common equity holders relative to senior
debt holders. We see that the tranche markets immediately responded to this choice
with the idiosyncratic component blowing out. This, as is discussed further below,
coincided with the meltdown of the equity market that saw a decline of almost half
in its capitalization.
i i
i i
i i
i i
i i
i i
124 V. Bhansali
assets are being supported by a diminishing equity base, each unit of falling equity
prices very drastically magnifies the economy-wide leverage. If we are on a path
of deleveraging that is not at its final resting place, then this denominator effect
(where equity is the denominator and the net assets are the numerator) will require
a further downward adjustment of the numerator, ie, of asset valuation broadly.
To see the impact of this simple approach in more detail, let us go back to asset
pricing done the arbitrage pricing theory way. By definition, the beta of the stock
market to itself is 1. The beta of any other security is the correlation of its returns
times the ratio of the volatility of the asset to the volatility of the stock market. So if
a security such as a treasury bond shows a beta of 0:10, that means that its returns
are negatively correlated to the stock market. On the other hand, high-yield bonds
show an equity beta of almost 1, ie, not only are they positively correlated to the
stock market but they have volatility of the same order of magnitude as the equity
market. One would only want to buy such a security if the compensation for holding
the default risk exceeded the risk from the market volatility of the equity market
by a wide margin. Many hedge funds and other absolute return strategies turned
out to be absolute negative strategies in 2008 for similar reasons: they were short
hidden equity, or what amounts to the same thing, liquidity risk. Higher equity beta
should, in efficient markets, compensate with higher return, or so goes the capital
asset pricing model, but higher beta also means higher drawdown and tail risk.
i i
i i
i i
than knowing that hedges exist that can make the difference between survival and
almost certain ruin.
12 CONCLUSION
In this paper I have offered some perspectives on tail-risk hedging from an investors
perspective. The essential insight is that tail risks for typical diversified portfolios
are a result of systemic risks and rising correlations. Thus a proper tail-risk hedging
program takes into account the relative pricing of broad macro markets and strate-
gies and evaluates the best alternatives from combinations of these alternatives to
immunize the portfolio against improbable but not impossible shocks.
REFERENCES
Bhansali, V. (2007a). Markowitz bites back: the failure of CAPM, compression of risky
asset spreads and the path back to normalcy. PIMCO Viewpoints, Spring 2007.
i i
i i
i i
126 V. Bhansali
Bhansali, V. (2007b). Putting economics (back) into quantitative models. Journal of Port-
folio Management 33(3), 6376.
Bhansali, V. (2007c). Volatility and the carry trade. Journal of Fixed Income 17(3), 7284.
Bhansali, V. (2008). Correlation risk what the market is telling us and does it make sense?
In Credit Risk: Models, Development and Management, Wagner, N. (ed.). Chapman and
Hall, Boca Raton, FL.
Bhansali, V., and Wise, M. B. (2001). Forecasting portfolio risk in normal and stressed
markets. The Journal of Risk 4(1), 91106.
Bhansali, V., Gingrich, R., and Longstaff, F. A. (2008). Systemic risk, what is the market
telling us? Financial Analysts Journal 64(2), 1624.
Fung, W., and Hsieh, D. A. (2001). The risk of hedge fund strategies: theory and evidence
from trend followers. Review of Financial Studies 14(2), 313341.
i i
i i
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.