Econometric Estimation of Credit Rating Transition Matrices

The article presents an econometric approach to estimation of credit transition matrices by using a number of explanatory variables such as geographic region, industry, business cycle, state, and borrower’s credit history. The hurdle or-dered probit model is in the core of the proposed method. The hurdle specification of a model allows performing an estimation procedure in two steps with default probabilities characterizing an objectively observed default event estimated in the first step followed by estimation of transition probabilities describing transitions of borrowers’ credit ratings subjectively assigned by credit managers. A data source is a unified database containing detailed information about the borrowers in the credit portfolio of a large German banking alliance.