Serving the Quantitative Finance Community
MBS IO/PO Emprical Durations
I am trying to estimate MBS IO/PO durations using the price data. Because the high convexity of IO/POs - the relationship between between price and rates in very unlinear. Does it make sense to use linear regression (e.g. regresssing price differences on swap rate differencies) to estimate empirical greeks for these products? Or some other techniques should be used?Thanks, Michael