From the answer to R37, Q13B: “For more complex portfolios containing options and time-sensitive bonds, the historical method might be more appropriate” [than the variance-covariance method]. I get why analytical VAR wouldn’t be appropriate for those asset classes, but historical VAR for time-sensitive bonds? I figured that’d be the least attractive choice given that bond sensitivities change over time (e.g. duration), etc. Any thoughts?
Duration decreases with time and the negative skew of bonds due to credit events make historical more appropriate.
good – thx. makes sense…decreasing duration will make historical VAR more conservative with respect to interest rate risk. as for the negative skew i suppose if the historical series is long enough to incorporate at least a full credit cycle i could go with historical… yup, i’m convinced. on to the next one…
even if you include full cycle delta of options will increase/decrease of callable instruments.