We survey both academic and proprietary models to examine how macroeconomic and systematic risk effects are incorporated into measures of credit risk exposure. Many models consider the correlation between the probability of default (PD) and cyclical factors. Few models adjust loss rates (loss given default) to reflect cyclical effects. We find that the possibility of systematic correlation between PD and LGD is also neglected in currently available models.
3
A Survey of Cyclical Effects in Credit Risk Measurement ModelsIt has long been recognized that banking is a procyclical business. That is, banks tend to contract their lending activity when business turns down because of their concern about loan quality and repayment probability. This exacerbates the economic downturn as credit constrained businesses and individuals cut back on their real investment activity.In contrast, banks expand their lending activity during boom periods, thereby contributing to a possible overheating of the economy that may transform an economic expansion into an inflationary spiral.The proliferation of credit risk measurement models in banking may accentuate the procyclical tendencies of banking, with potential macroeconomic consequences. That is, the models' overly optimistic estimates of default risk during boom times reinforces the natural tendency of banks to overlend just at the point in the business cycle that the central bank prefers restraint. Moreover, if credit risk models are unduly pessimistic during recessions, then even the most expansionary monetary policy may not encourage banks to lend to obligors that are perceived to be poor credit risks. Recent BIS proposals to utilize credit risk models such as CreditMetrics as a basis for bank capital requirements may further accentuate the procyclical nature of banking unless the credit cycle and its effect on credit risk are appropriately recognized in the model structure. If banks are constrained by risk sensitive (as measured by internal models) capital allocations and regulatory requirements, they may be unable to lend during low points in the business cycle and overly encouraged to lend during boom periods.1 This is because risk sensitive 1 To the extent that external credit ratings provide "through the cycle" estimates of default risk smoothed across the entire business cycle, it is the internal ratings-based approaches of the New Basel Capital Accord In this paper, we examine the treatment of cyclical factors in both academic and proprietary credit risk measurement models. 4 In section 2, we begin by discussing what is meant by procyclicality. We then divide our survey of credit risk measurement models into four sections. Section 3 surveys how various credit risk measurement models incorporate cyclical effects into the estimation of default probability (PD). In Section 4, that is most likely to exacerbate the procylical tendencies of banking. However, if credit ratings behave procyclically [as shown by Ferri, Liu and Majnoni (2000), Monfort and Mulder (2000) and...