Testing Simple Markov Structures for Credit Rating Transitions

Models abound that analyze changes in credit quality. These models are designed to determine the reserves and capital needed to support the risks of individual credits as well as portfolios of credit instruments. Historical information on the transition of credit exposures from one quality level, or rating, to another is often used to estimate models that describe the probabilistic evolution of credit quality. A popular specification is the simple, timehomogeneous Markov model. While the Markov specification cannot describe credit processes in the long run, it may be useful for describing short-run changes in portfolio risk. In this convenient specification, the entire stochastic process can be characterized in terms of estimated transition probabilities. However, the simple homogeneous Markovian transition framework is restrictive. We propose a simple test of the null hypotheses of timehomogeneity that can be performed on the sorts of data often reported. The test is applied to data sets on municipal bonds, commercial paper, and sovereign debt. We find that municipal bond ratings transitions are adequately described by the Markov model for up to five years, that commercial paper on a 30-day transition scale seems Markovian up to six months (the extent of the available data), and that the transitions of sovereign debt ratings are adequately described by the Markov model (a result that may derive from the limited data of small sample sizes). The views expressed in this paper are those of the authors alone, and are not necessarily those of the Office of the Comptroller of the Currency or the Department of the Treasury. Nicholas M. Kiefer is a Professor in the Departments of Economics and Statistical Sciences, Cornell University, Uris Hall, Ithaca, NY 14853. C. Erik Larson is a Senior Financial Economist in the Office of the Comptroller of the Currency’s Risk Analysis Division, Washington, DC 20219.