VAR Backtesting multiple

scorpiomanoj

New Member
Hi David,

Page 62 (last para) of VaR (3rd edition) by Philippe Jorion says that "The general market-capital charge shall be set at the higher of [ the previous day's VaR] OR [ the average VaR over the last 60 business days times a multiplicative factor]. Any specific reason as to why the multiplicative factor is not applied to the previous day's VaR?

Thanks in anticipation,
Halan Manoj Kumar
 

David Harper CFA FRM

David Harper CFA FRM
Subscriber
Hi Manoj,

I *think* the previous day's VaR is meant to be give sort of override protection to the particular challenge of historical averaging; i.e., if volatility has recently spiked (eg, regime change), a long term average would underestimate current conditions. So, i view the k*60-day average as the "baseline" (as the backtest influences the k from 3 to 4) with the k=3 meant to translate an average historical VaR into a "stressed or adverse VaR" (even before BIS subsequently added an additional Stressed VaR overlay); that is, k=3 is something like, "let's plus up the average VaR to give cushion to adverse scenarios." Then the MAX( ,previous VaR) is, in my understanding, an override to cover the volatility regime change problem. In summary, I'm not sure why we'd expect a multiplier here; a single outlier observation could be 3x or more of the *average*. (I may be unawares of some specific BIS commentary on this). But keep in mind: it's not like the k=3 is necessarily itself justifiable. It's pretty mushy itself.

David
 
Top