The calculation of a bond's duration based on historical data. Empirical duration is estimated statistically using historical market-based bond prices and historical market-based Treasury yields. When the historical yields change, the historical bond prices will change accordingly, which forms that basis for empirical duration. Regression analysis is the statistical process used to estimate empirical duration. 

Duration is the percentage change in a bond's price with a 100-basis-point change in yield. 



|||The calculation of empirical duration has some advantages and disadvantages over other duration calculations, such as effective duration or modified duration. The advantages of using empirical duration include that the estimate does not rely on theoretical formulas and analytic assumptions, and the only inputs needed are a reliable series of bond prices and a reliable series of Treasury yields.

Some disadvantages are that a reliable series of a bond's price may not be available, and the series of prices that is available might not be market based, but rather modeled or matrix priced (the price is based on a similar security).