Limits of stress-test based bank regulation

BIS Working Papers  |  No 953  | 
06 July 2021



How should supervisory risk assessments, such as stress tests, inform bank regulation when such assessments provide imprecise signals? What trade-offs do regulators face when redesigning assessments to improve accuracy? Does the disclosure of assessment results improve the effectiveness of capital requirements? We develop a theoretical framework to investigate these questions. A key element of our framework is the impact of assessment accuracy on the future behaviour of banks. This, in turn, is crucial for the design of risk assessments and for subsequent decision-making about capital requirements.


This paper strives to fill an apparent gap in the literature. Despite empirical evidence of noisy risk assessments, there is a lack of studies on what this noise implies for the effectiveness of capital requirements. We examine how capital regulation based on potentially inaccurate assessments affects banks' incentives to improve their risk profile, and derive the attendant optimal regulation. Our framework is robust and tractable. This also allows us to study trade-offs faced by regulators when making assessments more accurate and in disclosing results to investors. Moreover, we examine trade-offs involved in choosing optimal capital requirements when bank failure is socially costly.


Contrary to conventional wisdom, we show that higher capital requirements can create adverse incentives. This can lead to more risky banks when information frictions are present. As such, capital requirements must be less sensitive to assessment results when accuracy is lower. Where the regulator can enhance some aspects of accuracy only by worsening others, it may be optimal to impose lower capital requirements. We find that while disclosure of assessment outcomes can improve market discipline in general, when assessments are less accurate they can amplify risk-taking by banks. This further limits the effectiveness of capital requirements based on assessments. Regulatory trade-offs are aggravated when bank failures are more costly.


Supervisory risk assessment tools, such as stress-tests, provide complementary information about bank-specific risk exposures. Recent empirical evidence, however, underscores the potential inaccuracies inherent in such assessments. We develop a model to investigate the regulatory implications of these inaccuracies. In the absence of such tools, the regulator sets the same requirement across banks. Risk assessment tools provide a noisy signal about banks' types, and enable bank specific capital surcharges, which can improve welfare. Yet, a noisy assessment can distort banks' ex ante incentives and lead to riskier banks. The optimal surcharge is zero when assessment accuracy is below a certain threshold, and increases with accuracy otherwise.

JEL classification: G21, G28, C61

Keywords: capital regulation; stress-tests; information asymmetry; adverse incentives; disclosure policy; Covid-19.