Dependent Credit Migrations

This paper examines latent risk factors in models for migration risk. We employ thestandard statistical framework for ordered categorical variables and induce dependencebetween migrations by means of latent risk factors. By assuming a Markov process forthe dynamics of the latent factors, the model can be interpreted as a state space model.The paper contains an empirical study on quarterly migration data from Standard &Poor’s from the years 1981–2000, in which the ordered logit model with serially correlatedlatent factors is fitted by computational Bayesian techniques (Gibbs sampling). Apartfrom highlighting the usefulness of the Gibbs sampler for statistical inference in models ofthis kind, the survey in particular investigates the issues of rating-specific factor loadingsand heterogeneity among industry sectors, with emphasis on their implications in termsof implied asset correlations.