David Einhorn, founder of Greenlight Capital, was quoted in the New York Times as saying that value-at-risk (VaR) “creates a false sense of security…like an air bag that works all the time except when you have an accident.” Is that accurate? I agree with my colleague Laurent Birade when he says that implying that VaR creates a false sense of security is like blaming a high-speed car accident on the speedometer. VaR and other risk measurement models are gauges that tell us when we reach or exceed pre-defined limits. When observed, they can assist us in making proper decisions. But, it is still the human behind the wheel determining the speed at which to travel.
VaR, popularized after the stock market crash of 1987, is used primarily to define risk as mark-to-market loss on a fixed portfolio over a fixed time horizon – assuming normal markets. Prior to the 1987 crash, standard models failed to consider recurring (one or two per decade) crises. Financial crises affect many types of uncorrelated markets at once and with typically no discernable warning and models at that time overwhelmed the statistical assumptions embedded in them. VaR is a way to look at aggregated firm-wide risks and hypothesize about extreme or “fat tail” events without crashing the model with thousands of scenarios and assumptions.
VaR was never purported to be the mother-of-all-risk-measures. The prudent risk manager therefore uses it with other modeling techniques – such as exposure valuations, credit spreads, event/default/risk migration correlations, recovery uncertainty, loss distributions (in addition to VaR), capital allocation, risk-adjusted return on capital (RAROC) and portfolio optimization – plus empirical judgment.
Consequently, it may not be able to accurately predict extreme events, but as Phillipe Jorion says, “the greatest benefit of VaR lies in…critically thinking about risk. Institutions [using] VaR are forced to confront their exposure to financial risks and to set up a proper risk management function.” VaR forces management to confront, rather than ignore, risk.
An entire science is built around how human beings react to risk and risky behaviors. With time and temptation, we have the propensity to ignore long-term costs. Some people are thrilled with driving at high speeds; others are put off by it. With risky behaviors comes a cost. For speeding drivers, the costs include higher insurance premiums, punishment for traffic violations and at worst, an increase in traffic fatalities. In today’s economy, that cost is the latest financial crisis and the domino effect it has had on the job market, housing values and bank closings.
It is not that the models are broken; it is that people’s responses to what the models tell us are either built on faulty assumptions, uninformed gut reactions, or ignorance of what they tell us. Rather than throw out the models, filter them with experience and limits. Read more about VaR and how it can be helpful when combined with stress tests.