Numerical estimates of risk factors contingent on credit ratings

Assuming a favorable or an adverse outcome for every combination of a credit class and an industry sector, a binary string, termed as a macroeconomic scenario, is considered. Given historical transition counts and a model for dependence among credit-rating migrations, a probability is assigned to each of the scenarios by maximizing a likelihood function. Applications of this distribution in financial risk analysis are suggested. Two classifications are considered: 7 non-default credit classes with 6 industry sectors and 2 non-default credit classes with 12 industry sectors. We propose a heuristic algorithm for solving the corresponding maximization problems of combinatorial complexity. Probabilities and correlations characterizing riskiness of random events involving several industry sectors and credit classes are reported.