Do stress tests matter? A study on the impact of the disclosure of stress test results on European financial stocks and CDS markets

Abstract

During the recent sovereign debt crisis, the European Banking Authority conducted two stress tests on European banks in order to gauge their capital needs, core Tier-1 ratios and ratios of resilience to adverse shocks. We assess the informational content of the disclosure of the stress test outcomes. We conclude that the stress tests conveyed new information and that the outcomes were not anticipated by the stock market but were partially anticipated by the credit default swap (CDS) market. However, while the stock market reacted to the disclosure of the stress test outcomes, in the CDS market there is some evidence of a ‘reverse’ reaction. Moreover, the publication of the outcomes of the stress tests had a stronger impact on the stock prices of riskier financial institutions. A similar pattern is evident in the CDS market, albeit narrowed to one of the stress tests and amid the financial institutions with higher perceived credit risk.

Keywords:

• stress tests
• information disclosure
• banks
• CDS markets

JEL Classification:

• G14
• G21
• G28
• E65

Acknowledgement

The authors thank an anonymous referee for useful comments.

Notes

¹ Harrington (2006) finds evidence that the CDS market anticipates leveraged buyouts (LBO) prior to the LBO announcement, and this seems to be justified by informed trading.

² This association is supported by Kwan (1996), who finds that both stock returns and bond yield changes are driven by firm-specific information.

³ There is an intense debate in the literature concerning the stock market reaction to news. Some studies reveal that the adjustment of stock prices is too slow; others find evidence of overreaction and price reversals (DeBondt and Thaler, 1985; Jegadeesh and Titman, 1993). On the other hand, authors find evidence that the CDS market is efficient in processing the information of rating announcements, because market participants are well informed and sophisticated (Hull et al., 2004; Norden and Weber, 2004).

⁴ The analysis does not include stocks of listed financial institutions submitted to the stress tests whose business was discontinued.

⁵ Results for this group are not reported. ‘Failed’ means that the bank’s capital may fall below the threshold of 5% Core Tier 1 Ratio (CT1R) over the 2-year time horizon.

⁶ ‘Tangential results’ means a CT1R between 5% and 6% over the 2-year time horizon.

⁷ Except for the control group.

⁸ As a robustness test, we investigate whether our findings continue to hold when we control for the different sample sizes in the CDS and in the stock market. We select listed banks that are also reference entities of CDS contracts, and end up with a common sample (for the stock and the CDS market) of 23 banks in the treatment group and 15 in the control group. We repeat our analysis for this sample and our results (not reported) are essentially unchanged.

⁹ In addition, we undertake the following robustness test (we thank an anonymous referee for this suggestion), whose results are available from the authors upon request. We employ a battery of nonparametric tests that are robust to outliers and other nonnormality features of the data. Following Cowan (1992) and Campbell and Wasley (1996), we compute the t_grank. Following Kolari and Pynnönen (2011), we compute the GRANK test. Finally, we extend the Corrado and Zivney (1992) procedure and compute the sign rank test. These tests are also robust to event-induced volatility and cross-dependence. Overall, the results obtained using the parametric and the nonparametric tests are consistent, the conclusions being hardly affected under the nonparametric approach.