An Empirical Examination of the Classical Theory of Corporate Security Valuation

Abstract

This paper tests the main implications of the classical Black-Scholes-Merton theory of corporate security valuation. Data on the debt of companies with a single outstanding bullet bond is used to focus on the cleanest possible application of the theory. The basic Merton model is tested first. Then we offer the first systematic tests of whether observed pricing errors can be mitigated by more complicated models that admit default before maturity, stochastic interest rates and differential recovery rates. The models are tested under a variety of assumed conditions to examine the effects of possible mis-measurement of hard-to-measure parameters. The relatively successful configurations of the Merton model produce mean absolute errors between 80 and 90 basis points. Errors are related to coupon and time to maturity but are not related to credit rating. Evidence on the effect of firm size is equivocal. The Longstaff-Schwartz model's ability to accommodate early default and stochastic risk-free interest rates does not improve prediction accuracy in our sample. Efforts to improve accuracy through use of each firm's industry-average recovery rate were similarly unavailing.