Magazine article The RMA Journal

Are You Getting Enough Stress?

Magazine article The RMA Journal

Are You Getting Enough Stress?

Article excerpt

How to get more from your stress tests

Portfolio stress testing has been called difficult, confusing, and unhelpful, as well as intuitive, valuable, and efficient. The purpose of this article is to shed some light on the path to getting more out of credit portfolio stress testing through a three-step process. Perhaps in the process, risk managers will relieve some of their own stress.

"Insufficient stress" does not seem to be a problem for many risk managers. In fact, most seem to be perennially counting the days until the next unbreakable deadline, wishing they'd listened to their high school career advisor and become something less stressful instead--say, a cliff diver or a tightrope walker.

Not so their portfolios. For example, almost half the respondents to the Enterprise Risk Management survey, [1] drawn from a wide range of industries, do not perform stress testing. Furthermore, over a fifth of the respondents to the 1999 RMA consumer credit portfolio management study [2] indicated that they did not stress test their portfolios.

Given the limitations of most quantitative portfolio credit risk tools, [3] a couple of reality-checking stress tests would seem an essential part of a comprehensive risk management program. However, if they use them at all, most credit risk managers use them only grudgingly.

Why is this? Why would such an apparently useful tool, required by regulators and lauded by consultants, be subject to such ambivalence?

The answer seems to lie in the way in which stress testing is conducted. At most institutions, stress testing is a marginal activity appended to an already cluttered risk management process--one more box to check to keep the regulators at bay. As a result, the perception of stress testing ranges from vague intellectual interest to outright contempt.

In contrast, at those institutions where the risk managers expressed satisfaction with stress testing, it is an integral part not only of their enterprise-wide risk management process, but also of the strategic planning and capital allocation processes. As important, while its limitations are recognized, stress testing is accorded equal status with quantitative risk management methods, and, consequently, it attracts significant resources. Broadly speaking--as is true of many things--you get out of stress testing what you put into it.


Stress testing matters because most risk measurement tools are good at describing average loss levels but relatively poor at capturing the 1-in-20-year type of event that can wreak havoc on a portfolio. Therefore, stress testing methods are applied to explore the potential impact of extreme events that are beyond the applicability of standard statistical tools. Furthermore, regulators in most jurisdictions require it.

Stress testing quantifies the effects on a portfolio of "outlier" events. It gives insight on the portfolio behavior resulting from large moves in key market variables. For example, what would happen to the profitability or value of a portfolio if the Fed announced a 50 basis point increase in interest rates, or if the price of oil tripled (again)? These events, although very unlikely, are certainly possible.

Scenario analysis, a close relation of stress testing, typically goes beyond the immediate effects of predefined market moves and tries to explore the broader impact of a combination of events on revenue streams and portfolio value. The 1997 Asian crisis and the Mexican Peso devaluation in 1994 are examples of extreme scenarios where historically calibrated models performed poorly.

The principal advantages of stress testing and scenario analysis are:

* They are relatively intuitive ways of measuring and communicating risk (e.g. loss from 50bp shift in yield curve)

* They do not rely on historical co-movements of risk factors. Therefore, they provide valuable insights regarding what can happen if historical relationships break down. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.