Red-Blooded Risk: The Secret History of Wall Street

Beyond Profit and Loss

In all my examples, I've discussed daily profit and loss as something to predict. That's due to my background in managing financial businesses in which most of the risk was in strategies liquid enough to be priced every day, and not so high-frequency that a day was not a meaningful reporting interval. But a middle-office risk manager can use a VaR for any variable related to firm performance that can be measured objectively, and measured over a time scale that allows for statistical verification. It's more important that the variable be objective and measured frequently enough than that it be the most relevant measure of the firm's business.

For example, a middle-office risk manager of a hospital might want to develop a VaR on how many patients in the hospital at midnight die before the following midnight. This is not something the hospital wants to minimize; the way to do that is not to admit any sick people. It's also not the best measure of the hospital's performance. A bad hospital might not get any seriously ill patients and a great one might get more than its share. Moreover, there are bad things a hospital can do that don't kill anyone, and good things it can do that don't save lives. But deaths are related to the hospital's business, and they are objective, and they can be measured daily.

If you tried to implement this VaR, even at a well-run hospital, my prediction is you would find all kinds of system barriers to compiling the information you need ...