Key to the analysis is the finding that liquidity risk - blamed for much of what happened - is essentially "hidden" market risk. Contrary to suggestions that this liquidity risk is unforeseeable, in fact this hidden market risk is quantifiable, as previous Riskdata research, undertaken before and after the crisis, has demonstrated.
"Hidden" market risk means simply that an asset can potentially lose much more than can be observed from its track record. It is always the accumulation of hidden market risks which eventually ends up with a major liquidity and credit crisis. Unfortunately, the common extrapolation of past return fluctuations crucially misses three key sources of hidden risk:
*Assets exposed to risk factors which have been very quiet for some years eg. sub-prime flourished during a long period of low credit spreads and volatility;
*An asset can have a negative convexity versus a risk factor, so that while quite de-correlated in normal times, it becomes very correlated when markets collapse;
*If the asset is not properly "marked to market", then the investor is unable to anticipate potential losses based on its P&L experience. This can include fraud, but even without fraud, returns can be smoothed.
Fundamentally, traditional asset allocation and return-based risk methods miss hidden risks, now making the structural asset class risk premium an illusion. In business-as-usual conditions, traditional asset allocation can exhibit attractive patterns, with diversification mitigating day to day volatility and risk premiums (say on equities) leading to out-performance of scheme liabilities. But steady excess performance does not equal true alpha, as correlation changes dramatically in crisis periods, and whatever the asset class, there is a potential time bomb which can damage returns. High risk is no longer rewarded by higher return: the return of the S&P500 since 1987 is similar to that on government bonds, although the equity risk is twice the fixed income one.
The Liability-Driven Investment approach also relies on an implied assumption which may not work in a systemic crisis: the fact that a "safe" hedge exists, meaning a counterparty which cannot default. Both approaches rely on a rigid portfolio structure, which leads mechanically to unexpected concentration of risks, due to the changing market dynamic.
Riskdata's research across over 3,000 funds shows that in Autumn 2008, 64% of fund of funds exhibited hidden risk (meaning losses exceeding twice the past maximum drawdown in the fund history), along with 52% of relative value funds, 42% of event driven funds, 26% of equity hedge funds and 9% of macro funds.
Shifting from the asset allocation paradigm to "factor" risk allocation can help detect hidden risk, with techniques such as the bias ratio and stress Value-at-Risk - which combines stress tests and classical VaR - demonstrating the ability to foresee risk in advance, by capturing the sources of "non-linear" correlations between assets. Riskdata's analysis of fund selection by Stress VaR shows that in autumn 2008 those funds with lower stress VaR (under the past maximum drawdown) had very substantially lower levels of materialized hidden risk.
For investors - whether long-only or hedge funds - the best way to "risk budget" the next crisis is to set up extreme risk budgets by risk types, based on a long term view, and rebalance in a regular and disciplined way. The critical question investors should ask themselves is "how much can I afford to lose in such types of market scenario". Investors can "pay" for this extreme risk budgeting by actually relaxing business-as-usual risk constraints, which for most investors don't in fact protect in a crisis.
Raphael Douady, Research Director of Riskdata and author of the research commented: "Traditional diversification strategies don't protect against market dislocation, because they rely on a static view of asset class behaviour. They implicitly assume that liquidity and correlation will be the same in a crisis as in business-as-usual times, and that's no longer true, if it ever was."
Olivier Le Marois, Chairman of Riskdata added: "Using a "Factor" risk budget is a good way for an investment committee to discipline itself, since it helps resist the temptation to bet too much on the current source of easy returns, which will at some point turn out to be the next source of market dislocation. And it works - as long as you are using a risk model capable of detecting the hidden sources of risk, the sort that creates bubbles. Of course, to profit from extreme risk budgeting, you need to communicate it clearly as an agreed investment rule with the external and internal clients, otherwise the temptation to seek more superficially attractive ways of investing money may prove too tempting."