Estimating Transition Probabilities from Aggregate Mortgage Arrears Data in Ireland

Abstract

This paper models mortgage loan quality in Ireland when only aggregate arrears data are available. A first-order Markov model of credit migration is estimated using constrained regression and Bayesian techniques. The estimated transition probabilities for the Irish mortgage loan book are broadly consistent with findings in the literature, obtained using different methodologies and data. The Markov model is also estimated for UK data, and the probability of transitioning into arrears is found to be higher in Ireland. Once in arrears, however, transition dynamics are found to be remarkably similar in both countries. The probability of falling into arrears is found to have peaked in June but widespread forbearance is likely to be exceptionally costly in Ireland.