If a financial institution has a culture or business model that affects its sensitivity to crises, then it is expected that the performance of the institution in one crisis will predict its performance in another crisis. The authors show that poor stock return performance of a bank during the crisis of 1998 is a strong predictor of both poor performance and failure during the crisis that started in 2007. Banks that relied more on short-term funding, had more leverage, and grew more than their peers are more likely to be banks that performed poorly in both crises.

Practitioners can use the findings to better navigate a future crisis. The authors’ risk culture hypothesis states that the persistence of a risk culture or aspects of the business model that affect a bank’s performance in one crisis predict its poor performance in the next crisis. Alternatively, the authors’ learning hypothesis states that a bad experience in a crisis will prompt a bank to change its risk culture or modify its business model so that it is less likely to face such an experience again. Assuming that banks learn from their crisis performance, their performance in one crisis would not make it possible to distinguish performance across banks in the next crisis.

The authors empirically test both hypotheses against each other and find evidence that supports the risk culture hypothesis. Their results hold regardless of whether investment banks are included in the sample. They find that for each percentage point of equity value lost in 1998, a bank lost an annualized 66 bps during the financial crisis from July 2007 to December 2008. The correlation of crisis returns is driven by the quintile of the lowest performers and is inconsistent with the learning hypothesis.

The authors explore various scenarios that could explain the results. The personality traits of the executive rather than the persistence in a bank’s risk culture do not explain their results, nor does the possibility that banks found it unnecessary to change their risk culture because of a strong rebound from the 1998 crisis.

The authors test the risk culture hypothesis and the learning hypothesis using a sample of 347 publicly listed U.S. banks and show that the stock market performance of banks in the recent crisis is positively correlated with that of banks in the 1998 crisis. After summarizing the findings from various related papers, they present an overview of the events that hit financial markets starting in the middle of 1998.

The authors build their sample by starting with all companies in the CRSP and Compustat databases that have SIC codes between 6000 and 6300, which identifies them as financial institutions, and that existed in July 1998. They then exclude companies with foreign incorporations and further reduce the sample to include only those firms that also existed with the same Compustat identifier (gvkey) or permanent CRSP company identifier (permco) in Compustat and/or CRSP at the end of 2006. They do manual examination when firms match on either the gvkey or the permco criterion but the names do not match. Mergers and acquisitions are included in the sample. Finally, they exclude firms that are not in the traditional banking industry, such as investment advisers.

The authors use buy-and-hold returns from 1 July 2007 to 31 December 2008 to investigate the determinants of individual banks’ returns. In an additional analysis, they split the sample into large and small banks by median 2006 assets and find that large banks emerged from the crisis of 1998 faster but that smaller banks tended to do better during much of 2000–2009.

The authors’ findings are mostly consistent with expectations, such as that the correlation of poor returns during the 1998 and recent crises is at least partially attributable to banks having a risk culture or business model that favors high leverage, more short-term funding, and strong asset growth during the boom preceding a crisis. The findings may help investors construct portfolios that are more resilient in a future crisis, assuming that these results can help predict which financial firms are more likely to survive a future crisis and which may rebound faster.