An empirical analysis of structural models of corporate debt pricing

This paper tests empirically the performance of three structural models of corporate bond pricing, namely Merton (1974), Leland (1994) and Fan and Sundaresan (2000). While the first two models overestimate bond prices, the Fan and Sundaresan model reveals an extremely good performance. When considering the prediction of credit spreads, the three models under-estimate market spreads but, again, Fan and Sundaresan has a better performance. We find rating, maturity and asset volatility effects in the prediction power, as the models under-estimate less the spreads of riskier firms and of bonds with better rating quality and longer maturity. Moreover, our results reveal the existence of a new industry effect. Spread errors are systematically related to some bond- and firm-specific variables, as well as term structure variables.