Someone knows how KMV calculates their firm sigma's (std of asset value log-returns)?Because i had the possibility to have a look at their asset value time series, and the sigma'sobtained from them didn't matched the KMV sigma's (as they should?). In fact, since theseAsset value time series contain big changes in liabilities, volatility tends to be much higher thanthe volatility presented by KMV.Some interesting things about asset value correlations too, but nothing to tell here

I think there have been a few threads on this topic before. Basically, KMV does something like a simultaneous estimation of asset value and asset volatility, then they take a long-term (3 yr?) moving average of the volatility number from this estimation. But they are unwilling to divulge the details.

But imo, if i have asset value time series, the std of the logreturns (e.g for the last 3 yrs or for whatever period)should equal those sigma's, or shouldn't they?From what i saw KMV uses non-constant debt for their estimation of asset value time series. But if youdo this (from my own experience and checking the std of the asset value logreturns of KMV time series)you receive asset sigma's which are way to high (e.g. Allianz AG, Germany). While calculating asset value,i calculate the asset volatility iterative, and from my own experience in this field getting reasonable sigmasis only possible if you assume debt to be constant (at whatever values). Linear interpolation helps abitbut not really, especially not for financial institutions...Flex