Sunday, March 1, 2009
Value at Risk- Thank you Quants
Greed was the driving force behind the credit collapse but Greed couldn't have done much without a whole Batman's utility belt full of tools. One of the best was the Value at Risk model. VaR was developed at J.P. Morgan in the mid 90's to meet a demand by portfolio managers. More and more banks were deeper into trading and more and more trading involved hard to price derivitives (it didn't help that another quants invention, Black-Scholes, was being used to price these). VaR puts a dollar amount on the risk of any portfolio, no matter how hard to price. As it turns out that dollar amount is as useful as Rain Man's estimates on what things cost. (remember- "about a $100") It became industry norm to quote the 1% VaR number which is supposed to be the most a portfolio can lose 99% of the time. Compensation became tied to this number. As a trader or portfolio manager you aimed to have high profit numbers relative to your VaR. In theory this compared the risk you were taking to the profits you were making. The problem was the system was easily gamed. If a trade (a bet really) has very little likelihood of losing money but has a less than 1% chance of exploding in your face then it shows up as 0 risk on 1% VaR. It didn't take long for traders looking to make big bonuses to figure this out and start getting short wings in trader parlance. Those wings became cheap-- way cheaper than they would have been if VaR did not exist. If something went wrong they would quickly increase in price as one short after another tried to cover his position. Guess what? Something went wrong.