Predicting Annual Default Rates and Implications for Market Prices

Terry Benzschawel

This article was first published as a chapter in Credit Modelling (2nd edition), by Risk Books.

Forecasts of future corporate default rates are useful for estimating value-at-risk on credit portfolios, and for evaluating the attractiveness of credit market investments. In this chapter, we will describe a macroeconomic model developed by Yong Su and myself (Benzschawel and Su, 2014) to predict annual one-year high-yield defaults. That model builds upon on earlier work by Hampden-Turner (2009), which predicts monthly default rates using four predictors with various lags:

• Libor three-month/10-year slope;

• US Lending Survey;

• US funding gap; and

• GDP quarter-over-quarter (QoQ) growth.

The existing model is based on a rigorous analysis of variable lags as they affect predictions of default and a logistic transformation of predicted default rates. Also, the model is developed with strict attention to avoiding “look-forward” biases, so predictions of annual default rates are all out-of-sample. Finally, changes in predicted default rates from the model, but not current default rates, are shown to be predictive of future changes in credit spreads.

Rating agencies