Stress-Test Modelling for Loan Losses and Reserves

Michael Carhill and Jonathan Jones

The widespread use of macroeconomic and financial factors in the quantitative models that banks use to forecast their credit losses has been an important development. The financial crisis of 2007–9, and the associated severe recession, underscored the need for banks to incorporate economic and market conditions into their retail and wholesale credit risk models in order to produce credible stress loan-loss estimates. Prior to the crisis, banks were unable to estimate, and apparently uninterested in estimating, the credit losses that would result from a recession, probably because a generation of bank executives had never experienced a severe recession. This left banks unprepared for the severe loan losses that occurred between 2008 and 2010. In onsite supervision of national banks’ credit-risk management systems during the past several years, the authors have observed that bank executives at the larger banks, in response to the 2007–9 crisis, have determined that models capable of estimating credit losses conditional on stress economic scenarios were necessary for enterprise-wide capital planning and stress testing, and so directed their staffs to begin model-development efforts.