Estimating Structural Bond Pricing Models

Abstract:     
A difficulty which arises when implementing structural bond pricing models is the estimation of the value and risk of the firm's assets - neither of which is directly observable. We perform a simulation experiment in order to evaluate a maximum likelihood method applicable to this problem. The properties of the bond price estimators are examined using four theoretical bond pricing models: the Black & Scholes (1973) / Merton (1974) model, the Leland & Toft (1996) model, the Briys & de Varenne (1997) model, as well as the Ericsson & Reneby (2001) model. We contrast the performance of the maximum likelihood estimators to that of estimators traditionally used in academia and industry. The results are strongly supportive of the maximum likelihood approach. In fact, the inefficiency of the traditional estimator may explain the failure of past attempts to implement structural bond pricing models.

Keywords: Credit Risk, Maximum Likelihood, Corporate Bonds

JEL Classifications: G12, G13

Working Paper Series