VaR...oom

Thomas Cooley thinks we got complacent:

The last really sharp recession in the United States was from 1981 to 1982, when real output fell by more than 4% below trend, and the unemployment rate rose to over 10%. ....

Following that recession, something remarkable happened. The volatility in the U.S. economy declined sharply. Even though we have had two recessions in the ensuing years--in 1991 and 2002--both were relatively mild and short-lived.

Surprisingly, the U.S. economy remained dramatically more stable in spite of some major disruptions in financial markets in the U.S. and abroad over the same period.

....How calm was it? One measure that people have used is the cyclical volatility of the gross national product--it declined by one-half in the period from 1983 to 2006 compared to the period from 1955 to 1983.

The variability of consumption, investment and employment also fell by roughly similar amounts. The U.S. economy was less volatile and hence seemed less risky. A similar decline in volatility was documented in other countries as well. This widespread decline in volatility was dubbed the Great Moderation.

....There is [a] possible link between the Great Moderation and the financial crisis that is worth thinking about, because it may help to inform the financial regulation of the future. The idea is simply that the decline in volatility led financial institutions to underestimate the amount of risk they faced and overestimate the amount of leverage they could handle, thus essentially (though unintentionally) reintroducing a large measure of volatility into the market.

Financial institutions typically manage their risk using what they call value at risk or VaR. Without getting into the technicalities of VaR (and there is a very long story to be told about the misuse of these methods), it is highly likely that the Great Moderation led many risk managers to drastically underestimate the aggregate risk in the economy. A 50% decline in aggregate risk is huge, and after 20 years, people come to count on things being the same.

Risk managers are supposed to address these problems with stress testing--computing their value at risk assuming extreme events--but they often don't. The result was that firms vastly overestimated the amount of leverage they could assume, and put themselves at great risk.