Stress Tests for Retail Loan Portfolios

Bernd Engelmann and Evelyn Hayden

We introduce a simple pricing framework for retail loans, where the value of a loan is determined as the expected value of all future discounted cashflows. This framework is suitable for loans where no market information like bond spreads or credit default swap (CDS) spreads is available for the debtors. One common way to obtain multiyear default probabilities in this framework is the use of a Markov chain, which is calibrated to a one-year transition matrix of a bank’s rating system. Stress tests in this framework are often based on heuristic scenarios like downgrading every debtor by one rating grade and measuring the impact on the portfolio value or the regulatory capital. In this chapter, we show how heuristic scenarios can be replaced by econometric scenarios based on a statistical model of the total loan portfolio’s default rate related to macroeconomic variables. A stress of the macroeconomic variables leads to a stress of the portfolio’s default rate, which is transformed into a stressed transition matrix. This matrix is used to measure the impact of stress on the portfolio.

A loan is probably the most traditional and still the most important banking product. From a bank’s