The Role of Stress Testing in Credit Risk Management
Journal of Investment Management (JOIM), Fourth Quarter 2012
Posted: 20 May 2013
Date Written: May 17, 2013
In this article, we outline some concepts relating to the use of stress testing in credit risk management. We begin by providing a simple taxonomy of stress scenarios and discussing the trade-offs that different approaches require for implementation. Our taxonomy is modeled after one that is common in the credit literature and involves con-cepts related to reduced-form and structural approaches to credit modeling. Recently, some have expressed the view that the use of distribution-based measures such as VaR and expected shortfall (ES) for credit risk management should be deemphasized in favor of stress-testing and scenario analysis. We consider this question in the main portion of this article. We discuss the benefits of stress testing and scenario analysis as well as describing some limitations of using scenario-based approaches as a sole mechanism for assessing portfolio risk. We provide a number of examples to illustrate these limitations. In particular, except in special cases, it is difficult to use stress scenarios alone, ex ante, for allocating capital across disparate portfolios. However, stress testing and scenario analysis are integral to prudent credit risk management and can complement measures such as VaR and ES, thereby better informing both risk assessment and business strategy development. While neither stress testing nor VaR type measures, in and of themselves, provide complete descriptions of credit portfolio risk, combining both approaches results in more robust risk analysis. This permits risk managers to integrate quantitative measures with managerial intuition and judgment to arrive at more comprehensive assessments of both portfolio risk and overall firm strategy.
Keywords: Stress testing, scenario analysis, portfolio analysis, systemic risk, model risk, VaR, macroeconomic analysis, credit risk
JEL Classification: G00
Suggested Citation: Suggested Citation