Dynamic Rating with Feedback Effects

Abstract

In a dynamic signaling game, a rating agency observes a firm’s cash flow blurred by a persistent measurement error. Higher measurement errors imply earlier default. The firm’s owner can signal quality by cash injection in distress. With Bayesian directional learning, the rating agency eliminates high measurement errors, resulting in a lower cost of capital for the same level of observed cash flow. Thus, the rating feeds back into the owner’s cash injection decision and vice versa. The model provides a novel explanation for rating inflation and delayed default at lower asset values, despite the rating agency aiming for unbiased ratings.

Keywords: Rating, Asymmetric Information, Learning, Feedback Effect

JEL Classification: G24, G33, D83