Estimating the cross-sectional market response to an endogenous event: Naked vs. underwritten calls of convertible bonds

The literature on conditional event studies advocates the use of endogenous switching models to analyze cross-sectional variation in the stock market's response to corporate announcements of endogenous events. This paper proposes the use of a flexible Bayesian Markov chain Monte Carlo (MCMC) approach for estimating such models. The Bayesian MCMC approach offers several advantages over the Heckman–Lee two-stage estimation approach. In particular, analysis of the “treatment effect” (the difference between the observed and counter-factual outcomes) controls for the endogeneity of the firm's choice. As an application, the paper examines the market's response to naked and underwritten calls of convertible bonds. The paper reports evidence that the market's response to calls of convertible bonds is correlated with the private information partially revealed by the firm's choice of call type. However, although the average treatment effect associated with an underwritten call is negative, it is not significantly different from zero.