This paper examines alternative methods for making inferences about the value and dynamics of (unobserved) credit quality from market prices. Using data on Brady bonds issued by Mexico, Venezuela, and Costa Rica, we show that estimates of credit quality from of a simple model (often used by practitioners) with a constant conditional probability of default are dynamically inconsistent; the market distinguishes between current and future credit quality. We then examine two models with this feature. The first assumes that credit quality follows a diffusion process while the second assumes mean reverting dynamics for credit quality. Although those models have very different implications for the asymptotic probability of default, we show that it is impossible to distinguish between them on the basis of the few years of high frequency data in our sample. We then examine the possible consequences for the pricing of new issues that can arise from misspecification of the asymtotic probabilities.
Download is not available
