Corporate Credit-Risk Dynamics

The lack of widely available and comprehensive data of default activity on the corporate level has encouraged the advance of theories of corporate credit risk far beyond related empirical work. This article addresses this deficit by examining a unique data set—corporate bond defaults and credit ratings compiled by Moody's Investors Service since 1920. The article investigates the time inhomogeneity of company default risk—that is, the pattern in which the risk of default for a company changes over time. The empirical analysis addresses company default risk within a regression framework of competing risks, or hazards. The analysis identifies (and quantifies) variations in default risk with macroeconomic conditions, with industries, and through time—an important characterization for many approaches to measuring and managing credit risk. The characterization, measurement, and management of credit risk are of increasing interest to academics, policy makers, and practitioners. After having focused resources on interest rate, exchange rate, and other financial risks, market participants and academicians alike have turned their attention to the treatment of credit risk. Critical to measuring and managing credit risk is a thorough understanding of credit-risk dynamics.In the absence of widely available and comprehensive databases of corporate bond prices and defaults, theoretical research into credit risk has progressed much more rapidly than empirical research. Existing models generate a diversity of term structures for credit risk—term structures that are in some cases dependent on macroeconomic variables—and these term structures imply a variety of patterns in the default dynamics of the underlying debtors. This article reports empirical research into historical credit-risk patterns.The unit of research in previous studies was either the individual bond or its par value. The unit of research in this article is the issuing company. The data came from Moody's Investors Service's unique data on corporate bond-rating histories and defaults during the 77-year period from 1920 through 1996. I used these data to estimate a model capable of incorporating the effects of changing business conditions and capable of generating a variety of patterns for company credit-risk dynamics.The hazard regression model used was flexible enough to incorporate the salient features of default data. It allowed for companies to exit the market without defaulting and for companies not to have exited the market at all as of the 1996 compilation date of the research database.The model revealed significant dependence of default risk on several macroeconomic variables. The effects of the macroeconomic variables have important consequences for the development of methods for both the measurement and management of credit risk. In addition, the model found that when macroeconomic effects were controlled for, default risk exhibited significant time inhomogeneity. That is, for U.S. nonfinancial corporations entering the public debt market for the first time with the Baa, Ba, and B ratings in the period studied, and conditional on prevailing business conditions, the risk of default had a hump-shaped pattern: It initially increased with time to some point, after which it decreased. In a hypothetical situation of constant business conditions (GDP growth, real and nominal interest rates, and expected future profits), the model predicted that the risk of default for newly issuing B rated manufacturing companies is initially about 3 percent a year. For nondefaulting companies, that risk climbs to about 8 percent a year within six years but after about 22 years, falls back down to nearly zero.