The Coordination Role of Stress Tests in Bank Risk‐Taking

• Published date: 01 December 2019
• Author: CARLOS CORONA,GAOQING ZHANG,LIN NAN
• Date: 01 December 2019
• DOI: http://doi.org/10.1111/1475-679X.12288

DOI: 10.1111/1475-679X.12288

Journal of Accounting Research

Vol. 57 No. 5 December 2019

Printed in U.S.A.

The Coordination Role of Stress

Tests in Bank Risk-Taking

CARLOS CORONA,

∗LIN NAN,

†AND GAOQING ZHANG

‡

Received 26 July 2017; accepted 8 October 2019

ABSTRACT

We examine whether stress tests distort banks’ risk-taking decisions. We study

a model in which a regulator may choose to rescue banks in the event of con-

current bank failures. Our analysis reveals a novel coordination role of stress

tests. Disclosure of stress-test results informs banks of the failure likelihood

of other banks, which can reduce welfare by facilitating banks’ coordination

in risk-taking. However, conducting stress tests also enables the regulator to

more effectively intervene banks, coordinating them preemptively into tak-

ing lower risks. We find that, if the regulator has a strong incentive to bail

out, stress tests improve welfare, whereas if the regulator’s incentive to bail

out is weak, stress tests impair welfare.

JEL codes: G01; G21; G28; M40; M41

Keywords: stress test; stress-test disclosure; bank regulation; bank risk-

taking; bailout; coordination

∗TepperSchool of Business, Carnegie Mellon University; †Purdue University; ‡Department

of Accounting, the University of Minnesota.

Accepted by Haresh Sapra. We thank an anonymous referee, Sanjay Banerjee, Qi Chen,

Pingyang Gao, Zeqiong Huang, Aneesh Raghunandan, and seminar participants at 2018

American Accounting Association annual meeting, City University of Hong Kong, Columbia

University, Chapman University, 2019 FARS Mid-year conference, Hong Kong University,

Hong Kong University of Science and Technology, International Corporate Governance Con-

ference at Tsinghua University, Junior Accounting Theory Conference (George Washington),

Pompeu Fabra University, Rice University,Washington University at St. Louis, Southwest Uni-

versity of Finance and Economics (SWUFE), and University of Pennsylvania (Wharton), for

valuable comments.

1161

CUniversity of Chicago on behalf of the Accounting Research Center, 2019

1162 C.CORONA,L.NAN,AND G.ZHANG

1. Introduction

The severity of the 2008 financial crisis compelled the Federal Reserve to

step in and bail out troubled financial institutions in order to avoid a fi-

nancial meltdown. To prevent future financial crises, one of the measures

adopted by bank regulators was to institute stress tests to assess and publicly

certify the stability and resilience of the largest banks to adverse macroe-

conomic scenarios. An important matter still under debate is whether the

public disclosure of stress-test results is socially desirable. Although advo-

cates of stress tests argue that disclosing stress-test information enhances

bank transparency, others argue that the disclosure “may actually have un-

intended consequences” (Goldstein and Sapra [2014]). In this light, our

analysis reveals a message of caution: the public disclosure of stress-test re-

sults can potentially impair welfare.

The key economic mechanism behind the potentially detrimental effect

of stress tests relies on the interaction between the disclosure of stress-test

results and the possibility of a future bailout. The regulator’s inability to

refuse to bail out upon the concurrent failure of a sufficiently large number

of banks raises the issue of whether the banks’ anticipation of the bailout

creates the problem in the first place. The prospect of a bailout reduces the

expected loss of a failure for banks and, therefore, may induce them to co-

ordinate to take risks that make such a concurrent failure more likely. Our

analysis reveals that stress-test information can facilitate such coordination

by helping banks assess the likelihood of a bailout more accurately. If stress

tests reveal the presence of a large number of weak banks, the inferred high

likelihood of a bailout aggravates the banks’ excessive risk-taking, inducing

the regulator to intervene preemptively. However, absent stress-test disclo-

sure, these same banks would not have taken so much risk nor merited any

intervention, and would have invested in projects generating positive social

surplus. Essentially, our analysis shows that the disclosure of stress tests may

impair welfare by facilitating the coordination among banks to take exces-

sive risk, thereby eliciting an otherwise unnecessarily broader preemptive

regulatory intervention.

Our paper studies a setting with a continuum of banks in which each

bank can invest in a risky project (i.e., a loan) and privately choose the level

of project risk. A bank can be either prone to failure (i.e., a low-type bank)

or resilient to failure (i.e., a high-type bank). If a sufficiently large number

of low-type banks fail, the regulator bails out the failing banks (or a fraction

of them) to prevent the ensuing social cost. The anticipation of the possi-

bility of a bailout induces low-type banks to take more risk than is socially

desirable. However, the regulator can preemptively intervene by banning

banks from taking risky projects, but this may be socially undesirable

for two reasons. First, banning high-type banks from investing is socially

undesirable because, by assumption, high-type banks will neither fail nor

require a bailout. Second, banning low-type banks from investing could

be socially undesirable because, if absent intervention these banks would

THE COORDINATION ROLE OF STRESS TESTS IN BANK RISK-TAKING 1163

coordinate on an equilibrium in which each of them takes a low level of risk,

they would generate a positive social surplus; banning them from investing

thereby leads to a loss of that surplus. We examine the welfare impact of

stress tests by comparing two scenarios. In the first scenario (henceforth,

“no-stress-test scenario”), the regulator does not conduct stress tests and is

thus restricted to intervene without knowing the banks’ types. In the second

scenario (henceforth, “stress-test scenario”), the regulator conducts stress

tests and publicly reveals the results for all banks. In this case, the regulator

can selectively intervene banks based on the stress-test information.

In our study, stress tests play two main roles. First, the disclosure of stress-

test results facilitates the coordination of risk decisions among banks by

providing information about the likelihood of a bailout. Under some con-

ditions, the disclosure of stress-test results may make risk decisions in the

banking industry more extreme. If the stress-test results reveal the pres-

ence of a small number of low-type banks, low-type banks infer a low like-

lihood of a bailout, which in turn coordinates them into taking a low level

of risk. However, if the stress tests reveal a large number of low-type banks,

the inferred high likelihood of a bailout aggravates banks’ excessive risk-

taking. The potential detrimental effect of stress tests on welfare is then

attributable to such a coordination role. In particular, if absent stress tests

low-type banks would coordinate on an equilibrium in which they take low

risk, then the disclosure of the test results can induce them to coordinate

instead on an equilibrium in which they take high risk. This forces the reg-

ulator to ban some banks from investing, which in turn leads to a welfare

loss compared with the no-stress-test scenario, because, absent stress tests,

the banned banks would have generated a positive social surplus.

Conducting stress tests, on the other hand, also facilitates regulatory

intervention. In the no-stress-test scenario, not knowing the type of each

bank, the regulator must intervene indiscriminately and thus potentially

incurs the social cost of banning some high-type banks from investing. On

the contrary, conducting stress tests provides the regulator with informa-

tion about banks’ types and thereby allows the regulator to fine-tune the in-

tervention policy to depend optimally not only on the individual but also on

the aggregate stress-test outcomes. We show that in equilibrium the regula-

tor never bans high-type banks from investing. As long as the total number

of low-type banks is sufficiently large, the regulator only bans some low-type

banks from investing, with the purpose of coordinating the remaining unin-

tervened low-type banks on a lower-risk equilibrium. Therefore, conducting

stress tests yields the social benefit of allowing the regulator both to inter-

vene selectively and to discipline the risk decisions of unintervened banks.

The overall effect of stress tests on welfare is determined by the aggre-

gate effect of the two roles they play. We find that stress tests improve social

welfare if the regulator has a strong incentive to bail out. This occurs when

the bank regulator faces a steep social cost of bank failures or when the so-

cial cost of a bailout is low. However, if the regulator has weak incentives to

bail out banks (i.e., if the social cost of bank failures is low or the social cost