Yield curve data choice and potential moral hazard: An empirical exercise on pricing callable bonds

There are alternative providers of zero‐coupon yield curve datasets. This essential input for most financial purposes can also be estimated from cross‐sectional market price information. Although all these datasets are representations of the same reality, each dataset provides estimations from different baskets of assets and different fitting techniques. As the properties of the yield curves and their time series dynamics are different, the results of pricing interest rate derivatives may be sensitive to the choice among these datasets. We compare our own spot rates estimation from GovPX bond data and three popular available interest rate datasets. From an empirical exercise on pricing callable bonds, we observe relevant differences in economic terms when volatilities from different inputs are used. Therefore, financial institutions may have an incentive for choosing unduly one or another yield curve dataset in their self‐interest, raising a moral hazard problem.