The Application of Credit Risk Models to Macroeconomic Regulatory Stress Testing

The application of credit risk models in Comprehensive Capital Analysis and Review and European Banking Authority mandated regulatory macroeconomic stress testing is of significant concern for banks. The credit models that are used to project stressed losses and impairments under macroeconomic scenarios are also instrumental in projecting interest income and accrual as well as the balances that go into projected risk weighted assets. In this paper we review both the significant role that the credit risk models play in the macroeconomic stress testing as well as demonstrating how different credit risk models can be efficiently implemented for stress testing execution with concrete application examples. Many of the credit risk models banks use in practice can be efficiently implemented through a very simple conditional Markov iteration. Examples include multi-factor models derived from the Merton structural approach and dynamic transition matrix models that depend on economic factors and are traditionally estimated on cohorts of loans. We also analyze the efficient implementation of more complex dynamic transition matrix models with the added feature of delinquency history tracking. Such models are frequently used for retail portfolios and can introduce significant past state dependence. Traditionally, such models are therefore deployed in stress testing using simulation of state transitions. However, in some important cases, such as quarterly models and monthly models with delinquency state indicator functions, the models can be solved more efficiently with an expanded conditional Markov iteration