Peacetime Stress Testing: A Proposal

Paul Calem, Jose Canals-Cerda, Arden Hall and Lauren Lambie-Hanson

In Calem and Hall (2013), some basic principles of bank capital stress tests regarding the objectives and limitations of stress testing, design of appropriate scenarios and the role of models were outlined. Particular attention was focused on stress-test modelling of balance-sheet loan loss. In that paper, we adopted a somewhat sceptical view of whether stress tests can reliably project credit losses for a hypothetical future downturn due to the need to rely on models that have been fitted to historical data. We opined that “there can be no assurance either that future behaviour under stress conditions will resemble past performance, or that the historical data incorporate a sufficiently diverse range of economic conditions to guarantee accuracy of the prediction”.

Accordingly, we expressed the view that the stress testing of bank capital is most useful as a “diagnostic” rather than a predictive tool. Stress-test models inherited from a previous downturn should not be expected to yield reliable loss projections, given how the nature and scope of banks’ risk exposures may have evolved over the intervening period. They are, however, useful for appropriately defined purposes. They