Information Content of Mandated Bank Stress Test Disclosures
Atif Ellahie*
University of Utah, David Eccles School of Business
1655 E Campus Center Drive,
Salt Lake City, Utah 84112
atif.ellahie@eccles.utah.edu
First Version: November 27, 2014
This Version: November 2016
Abstract
Prior studies generally question the credibility and usefulness of the disclosed outcomes of
government stress tests of banks in the European Union during the recent global financial crisis.
In contrast, I document robust evidence that the geographic distribution of sovereign- and credit-
risk exposures disclosed in the EU stress-test results enables clear sorting of strong and weak
banks on the basis of expected future performance and sovereign spillover risk. Ex ante bank
specific exposure measures developed using test disclosures have predictive information content
for cross-sectional variation in subsequent equity and credit returns, return volatility, the yield
curve, and analyst forecast revisions.
Keywords: mandatory disclosure, stress testing, financial crisis, information content
JEL Classifications: G12, G28, M41
*
I thank Scott Richardson for his guidance and comments. I also appreciate helpful comments from Hina Baig, Joao
Cocco, Maria Correia, Emmanuel De George, Wayne Guay, Rachel Hayes, Anya Kleymenova, Ningzhong Li, Yun
Lou, Frederic Malherbe, Seda Oz (discussant), Emre Ozdenoren, Bugra Ozel (discussant), Stephen Penman,
Marlene Plumlee, Hélène Rey, Giovanni Ricco, Til Schuermann (discussant), Roger Silvers, Doug Skinner, Steve
Stubben, İrem Tuna, Oktay Urcan, Florin Vasvari, Vikrant Vig, and participants at London Business School
workshops, the 2012 AAA Annual Meeting, the 2012 Federal Reserve Bank of New York / Journal of Accounting
Research Conference and the 2012 LBS Transatlantic Doctoral Conference. I also wish to thank Hamid Mehran and
Abbie Smith (FRBNY/JAR Conference editors). I acknowledge London Business School and the University of Utah
for financial support. All errors are my own.
Electronic copy available at: https://ssrn.com/abstract=2685919
,1. Introduction
The recent financial crisis has seen increased demand for real-time information on the
financial condition of banks. Specifically, investors and regulators have expressed concerns
about the lack of transparent and reliable information on the composition of bank risks, the
resilience and capital adequacy of banks in downturns including their contribution to systemic
risk, and the potential contagion effect of bank failures. In response, regulators in the United
States and the European Union (EU) have conducted “stress tests” of banks and mandated public
disclosure of the results. In particular, the EU stress tests contain mandatory disclosures of
sovereign- and credit-risk exposures of European banks across over 30 geographic segments.
Thus the EU bank stress tests enable cross-sectional comparability of risk exposures and provide
a rich setting in which to examine the consequences of mandatory disclosure.1
An important question regarding the mandatory disclosure of bank stress test results is
whether the tests provide investors and regulators with useful information, and if so, to identify
the source of information content. Recent studies examining bank stress tests have either used
the disclosure event dates themselves, or the announced pass-fail outcomes of the tests, to study
the effect on the information environment for tested banks compared to untested banks (see
Bischof and Daske, 2013). However, there is a clear selection issue since primarily large and
systemically important financial institutions were subjected to the stress tests. Thus it is
impossible to identify a comparable group of banks that were not stress tested, and this challenge
significantly weakens inferences about the effect of the disclosure on the information
environment. Overall, prior studies conclude that the pass-fail outcomes and capital shortfalls
revealed by bank stress tests were not credible and failed to forecast subsequent bank
1
I do not use the US setting because only the 19 largest US bank holding companies were stress-tested, and the
detailed exposure information that I require in this study was not released.
1
Electronic copy available at: https://ssrn.com/abstract=2685919
,performance (also see Acharya, Engle, and Pierret, 2014). The usefulness of mandated stress test
disclosures remains a largely unanswered question.
Instead of focusing on the pass-fail outcomes or the effects of stress tests on the
information environment for banks, the motivation of this paper is to examine whether the actual
data disclosed in the stress test results had useful information content for capital market
participants. Specifically, I use the detailed geographic exposure data disclosed in the EU bank
stress test results to develop ex ante measures of sovereign risk and macroeconomic growth for
each tested bank. Also, I employ a cross-sectional research design that focuses only on the tested
banks in order to circumvent the need to identify a suitable control group of untested banks.
Using this approach, I provide robust evidence that the detailed exposure data has predictive
information content for capital market participants to (1) enable ex ante sorting of strong and
weak banks and (2) develop forecasts about the subsequent performance of tested banks.
To my knowledge, this is the first study to fully examine the predictive information
content of country-by-country risk exposures (i.e., exposure at default or EAD) disclosed in the
results of the 2010 and 2011 EU bank stress tests.2 I assume the perspective of equity and credit
investors and use the disclosed EADs to develop ex ante measures of exposure to sovereign risk
and macroeconomic growth for each tested bank. The approach is predictive in the sense that the
stress test exposure information enables identification of those banks that have the highest cross-
sectional exposure to sovereign risk and macroeconomic growth at the time the stress test results
are publicly disclosed. This ex ante identification of the differential exposures of banks enables
investors to form expectations about their subsequent performance after the release of results.
2
Exposure at default (EAD) is the total amount of risk that a bank is exposed to in the event of default by an obligor
and is an important credit risk parameter used to quantify expected loss (EL). Specifically, EL for a given maturity is
computed as 𝐸𝐿 = 𝑃𝐷 × 𝐿𝐺𝐷 × 𝐸𝐴𝐷, where PD is probability of default and LGD is loss given default expressed
as a percentage of EAD. Furthermore, EAD is used to calculate the risk weighted assets of a bank.
2
Electronic copy available at: https://ssrn.com/abstract=2685919
, Specifically, I examine cross-sectional variation in subsequent equity returns, CDS spreads,
realized equity return volatility, the slope of the credit term structure, and revisions in earnings
expectations. If the exposure information revealed by the stress tests has predictive information
content, then these performance expectations should be realized ex post. I find results that
support such a predictive role for the exposure information disclosed by the EU bank stress tests.
As part of the stress testing process, the European Banking Authority (EBA) mandated
the disclosure of sovereign EADs five times over the 2010 to 2013 period, and in 2011 also
mandated the disclosure of total credit risk EADs. In this study, I focus mainly on the EAD
disclosures made in the 2010 and 2011 stress tests. I expect these earlier disclosures in the midst
of the financial crisis to provide new information for capital market participants, which will
sharpen identification of the response. (See Appendix A for details on the 2010 and 2011 tests,
and see Appendix B for the five disclosure events.)3 Before the EU bank stress tests, detailed
information on the geographic distribution of exposures was not publicly available for a broad
cross-section of European banks. Alternative sources for similar information, such as Pillar 3 risk
reports under the Basel II Accords, are mostly narrative in nature and detailed EAD information
is not reported consistently or in a standardized format.
There are several arguments in favor of the hypothesis that the geographic distribution of
exposures would be useful for investors, including (1) non-redundancy of the disclosed
information, (2) objectives of the European Banking Authority to increase transparency, and (3)
heightened sovereign risk concerns in Europe.4 However, there are several features of the bank
3
The European Banking Authority (EBA) and its predecessor, the Committee for European Banking Supervisors
(CEBS), stress-tested 22 banks in 2009, 91 banks in 2010, and 90 banks in 2011. This paper focuses on the 2010 and
2011 tests as EAD information was not disclosed in 2009.
4
I am primarily interested in examining whether the geographic exposures across countries are valuable to investors
in developing beliefs about the future performance of banks, and not whether this information is redundant.
However, to some extent non-redundancy is a necessary condition for the exposures to be informative.
3
Electronic copy available at: https://ssrn.com/abstract=2685919
