Estimating Credit Migration Matrices with Aggregate Data - Bayesian Approach
|Issue||W - 30|
credit migration matrices, Bayesian inference, inequality constrained regression, truncated normal vector
This paper studies the Gibbs sampler developed in Rodriguez-Yam et al. (2004) that can be used to estimate the parameters of inequality constrained regression. The aim of the paper is twofold. First, to present an efficient estimation methodology that has not yet been utilised in credit risk literature, and second, by using this method to estimate a migration matrix governing the dynamics of loans to corporate sector in Croatia when only aggregate (supervisory) data on bank loans in every rating category is available. The results of the analysis suggest that the Bayesian estimator used in this paper has some important comparative advantages over the estimators previously used in the literature.