Red-Blooded Risk: The Secret History of Wall Street

Validation

One thing that VaR does is subject key risk information to validation. Some middle-office risk managers prefer other measures or metrics, but all competent ones believe in constant, rigorous, objective validation. Otherwise they're just pushing more bad numbers around and not doing any good for anyone. From this perspective, it's not surprising that VaR calculations turned out to contain lots of useful information. Just as the one-eyed man is king in the valley of the blind, a validated number, even if it is of little direct interest, is king in the valley of imaginary numbers.

Middle-office risk managers compute VaR, or other statistical predictions, on things other than profit and loss. Some of them invented key risk indicators (KRIs) and key performance indicators (KPIs). The idea is that whenever you measure something for risk purposes, people distort it. Even though you tell people you're not trying to minimize or maximize the measure, once people know you're looking at it, they change it. One of the reasons we selected profit and loss for VaR was that people were already trying to distort them, and financial controllers were on the job to fight the distorters. The controllers weren't always successful, but at least there were two sides in action. For KRI and KPI, you select an average of things that may be of only marginal interest ... and you don't tell anyone what those things are! People don't know what to distort.

For example, you might average employees out ...