Copula Methods and the Analysis of Credit Risk

Abstract

Banks' internal credit risk assessment may take into account not only publicly available ratings but also internal measures of credit risk. We present a credit rating model of credit risk that permits banks to employ their own risk assessment in addition to public rating information. Banks need to evaluate the credit risk of portfolios of risky assets. Ordinary correlation techniques fail to capture dependency relationships between marginal distribution functions. The copula can describe these relationships exactly. To model the credit risk of baskets of securities we examine dependency using copula methods paired with EVT techniques. We empirically estimate the dependency between different credit classes, and discuss the application of the empirical copula to valuation problems.

Keywords: credit risk management, default correlation, copula

JEL Classification: C10, G12, G21