Files
Abstract
In recent years, the reduced-form approach to valuation has become widely used in asset pricing. Unlike earlier structural models, reduced-form models do not require that data on the underlying asset be available, which makes them a convenient tool for empirical applications. I develop a simple reduced-form model of mortgage pricing where both default and prepayment are exogenous, and empirically estimate it using historical data on about one million fixed-rate residential mortgages. The empirical analysis proceeds in several stages. First, individual mortgage histories are used to estimate effects of exogenous variables, those being loan-specific characteristics, on hazards of termination. In the second stage, particle filtering is employed to estimate stochastic hazard processes. Once the parameters of hazard processes, as well as those of the conventional risk-free terms structure, are obtained, I use calibration to convert physical processes into risk-neutral ones. After that, tests of pricing performance are conducted.