Limits of Stress-Test Based Bank Regulation

Abstract

Stress-tests provide complementary information about banks’ risk exposures. Recent empirical evidence, however, has uncovered potential inaccuracies in stress-test based assessments. We investigate the regulatory implications of these inaccuracies. Without stress-tests, the regulator cannot observe bank’s type, and sets the same requirement across banks. Stress-testing provides a noisy signal about the banks’ types, and enables bank specific surcharges, which can improve welfare. Yet, when stress-tests are less accurate, they distort banks’ ex-ante incentives to improve their risk-return profiles. As a result, higher capital surcharges can lead banks to be more risky. The optimal surcharge depends on the accuracy of stress-tests, and can be zero.