January 16th, 2007, 11:04 pm
just my $.02: often when implied vols move, the whole surface will move in parallel. (a PCA will show you this). to reflect the parallel shocks best, it might be better to shock both the initial and long-run volatilty parameters by 1%. this will shock the forward ATM implied vol term structure roughly in parallel (the implied vols in the tails, on the other hand will not be shocked in parallel). But all the same, this is a better way of reflecting vega.