Risk management seems to be a household topic these days. No longer an exclusive province of quants, it is discussed regularly by the popular press. From the shortcomings of value-at-risk to the latest stress tests imposed by the Fed on banks, risk management has gone mainstream. But why has it failed so dramatically in 1998, 2000, 2007, 2008 and again in 2010 with the European debt crisis? We will argue that too much focus has been placed on backward-looking risk measures and not enough on scenario analysis.

Risk Management is not Clairvoyance and Market Timing

But what is risk management anyway? While almost everyone uses the words ‘risk management’, the implied meaning seems to differ from person to person. Let’s start by discussing what risk management is not. Risk management is not about being a clairvoyant. It is not about forecasting specific future events. It is certainly not about timing those events i.e. ‘the big short’. Risk management is all about finding the suitable risk/return profile. Suitable, that is, to the holder of the portfolio. When your client has a suitable portfolio, he or she will not sell when the losses hit, because the possibility of those losses was discussed right from the beginning. The biggest problem for most investors is that they get in near the top and get out near the bottom. ‘Suitability’ ensures that this vicious circle is disrupted. If your clients don’t get out at the bottom, then they won’t have to jump back in at the top. But the only way that could happen if their risk/return tradeoff was chosen by considering a variety of scenarios: good, bad and ugly. Showing low risk simply because the realized volatility is low and VIX is heading toward single digits cannot really be called risk management. That is called rear view mirror driving.

Why Stress Testing?

Why do we need stress testing when we already have risk measures like Sharpe ratio, Beta, Value-at-Risk, Tracking Error, and  Expected Shortfall? The problem with these measures is that they are completely backward-looking. A tracking error or a Sharpe ratio today has no clue that interest rates can actually rise by more than trivial amounts, since nothing of that kind was observed in the data sample used to calculate those metrics.

Moreover, if you ask one of these measures about the possibility of a simultaneous rise in interest rates and a drop in equities, it will likely tell you that the probability of such a one-two punch is zero. When those measures will gain that knowledge, it will be too late to be useful, since the events will have already happened. Backward-looking risk measures still view long-dated Treasuries and any instruments with a great deal of interest rate risk as virtually riskless, due to the low volatility and safe haven appeal that we have observed over the recent decades. This will lead you to underestimate probabilities of events that are quite plausible in the next few years, such as inflation or stagflation.

Stress Testing Gaining Prominence

Have you noticed how Fed stopped talking about Value-at-Risk for banks? It practically disappeared from popular discourse starting in 2009. The Fed and the ECB now routinely talk about stress testing of banks. Of course, stress testing can also be misused and gamed by the banks, but at least they realized that estimating risk simply based on what happened over the past couple of years is a rather flawed strategy.

Shortcomings of Stress Testing

Every tool has limitations and has to be used properly.  A Maserati is not a bad SUV, it simply isn’t meant to go off-roading. Similarly, most of the arguments against stress testing are really arguments against a misplaced utilization  of that tool, not against the tool itself.

First « 1 2 3 » Next