Testing Homogeneity of Time-Continuous Rating Transitions
by Rafael Weißbach of Dortmund University of Technology,
August 23, 2005
Abstract: When modelling rating transitions as continuous-time Markov processes, in practice, time-homogeneity is a common assumption, yet restrictive, in order to reduce the complexity of the model. This paper investigates whether rating transition probabilities change after the origination of debt. Accordingly, we develop a likelihood-ratio test for the hypothesis of time-homogeneity. The alternative is a step function of transition intensities. The test rejects time-homogeneity for rating transitions observed over seven years in a real corporate portfolio. Especially one-year transition probabilities increase over the first year after origination. This time effect suggests that banks should manage their credit portfolios with respect to the age of debt.
Keywords: Portfolio credit risk, rating transitions, Markov model, time-homogeneity, likelihood ratio.
Published in: Empirical Economics, Vol. 36, No. 3. (June 2009), pp. 575-596.