Migration Rates and Credit Risk Economic Capital

ABSTRACT Transition matrix methods for credit risk migrations of borrowers follow three hypotheses. From one hand, it is assumed migrations are consistent with a homogeneous Markov process: heterogeneous borrowers of the same rating class have the same probability to migrate to a given rating class in the next future, regardless of their histories. From the other hand, it is assumed the default state is absorbing. Theoretical intuition suggests what empirical evidence puts forth: both assumptions are questionable. Moreover, it is commonly neglected that the universe of borrowers changes with time with business cycle relevance: some borrowers exit while some new ones enter into the portfolios of credit institutions; hence, standard transition matrix methods does not take care of portfolio's renewals and a complete recovery. To overcome these axiomatic constraints, the present study suggests the sample of borrowers to be open, with new entries and exits at each time, beyond the only recombination of present time borrowers over different credit risk classes, as it happens if the sample is closed once and for all. Opening the sample allows considering the configuration of the system is changing through time, either by recombination and renewal. What the paper argues is that a more realistic estimation of the credit risk economic capital is feasible by coupling it with the business cycle. To the end of application purposes, if diversified provision rates were given to different rating classes, while taking care of renewals and recombination as driven by the business cycle, the total amount of funds a credit institute should provide as economic capital at risk can be statistically estimable, while obeying regulatory norms. Accordingly, the credit activity can be made more sustainable with regard to macro-prudential and systemic financial stability criteria along the business cycle.