Red-Blooded Risk: The Secret History of Wall Street

Trans-VaR Scenarios

The next middle-office tool for exploring the trans-VaR region is scenario analysis. It bears some similarities to stress testing, but instead of a sudden shock you imagine a plausible bad sequence of events. This is a test of the firm's active defenses. The point is to make decisions ahead of time, when everyone is calm and has plenty of time, rather than in the midst of panic. Equally important is to make decisions by consensus. In an emergency, decisions are often made by the most confident or ambitious person, who may be the least qualified to make them. Sometimes it comes down to the person with the loudest voice. Even if that's not true, key people to provide input and make decisions may be unavailable when the crisis hits. So it's worth getting their thoughts ahead of time.

Scenario analyses are always run to the death. That is, you don't test up to any specified point; you just keep piling on losses until further decisions are pointless because either the firm no longer exists, or the firm has cut all possible risks to zero, or the scenario has become so implausible that people no longer know what reactions are appropriate. As a quant middle-office risk manager, I always want quantitative rules that I can program into a computer for automated warning. For example, one scenario analysis everyone runs is trading losses at the firm. You might start with a $10 million loss the first day, and double the loss each day as long as no action is taken. The executive ...