May 6th, 2004, 12:54 pm
yes, I used this thing for credit VaR, but there's nothing really novel about that workit's been a couple of years, but what I recall is that the VaR most people use is too tail-y to get decent statistical power with this kind of thing... overall I don't really approve of the direction that VaR has been taken. It's a little like the way CAPM "evolved"... the theory is kind of nice but there are so many problems with practical implementation that the better practitioners just start over with their own ad-hoc methodology. Unfortunately, VaR is tied up with regulatory issues now, so you still have to jump through the hoop even if you think it's dumb.