Last week, the Wall Street Journal Pro published a column, ”Creighton’s Take: Why Summers Questions the Fed’s Stress Tests,” which quoted an interview with Professor Lawrence Summers about the recent paper he coauthored with a Ph.D. student Natasha Sarin. Although we agree with some of the key takeaways from the Sarin and Summers paper we are concerned with a suggestion that the methodology employed in the U.S. stress tests is flawed:
‘The idea you could have 50% decline in the stock market and the value of banks’ equity would fall by a third, seems untenable,’ Mr. Summers said in an interview, referring to the sorts of severe scenarios in the Fed’s annual exercise meant to assess banks’ resilience in a crisis. ‘I would guess that a scenario remotely like the one described in the stress tests would see at least one major institution on the brink of insolvency … or [the] beneficiary of government assistance,’ he added. ‘Ouch.’” (Creighton’s Take: Why Summers Questions the Fed’s Stress Tests)
“Consideration needs to be given to approaches…that give more weight to market prices as indicators of asset values…” (Sarin and Summers, p.35)
Our reaction to Professor Summers’s critique of the U.S. stress tests is based on three main points: (i) the market value of equity is too procyclical and is not a very useful measure to assess the capital adequacy of banks; (ii) the sensitivity of the market value of bank equity to movements in the overall stock market assumed in Sarin and Summers is likely too large for well-capitalized banks; and (iii) the extremely severe supervisory scenarios used by the Federal Reserve in the U.S. stress tests would likely cause the failure of hundreds of smaller banks and destabilize many nonbanks, leading to an inflow of deposits and a migration of various financial activities to large banks.
First, the market value of equity has some serious shortcomings as a measure for assessing the capital adequacy of banks. Capital requirements based on the market value of equity would be procyclical and driven by changes in market sentiment and therefore would be a poor measure for assessing whether banks have sufficient capital to absorb losses during stressful conditions. Moreover, a procyclical capital requirement would force banks to deleverage excessively in response to adverse shocks to the equity market. That deleveraging would lead, in turn, to unnecessary swings in lending and real economic activity. As shown by the blue line of Figure 1, the market leverage ratio was approximately 12 percent for the largest 6 U.S. banks at the onset of the past financial crisis, higher than their current market leverage ratio value of about 10 percent. However, common equity tier 1—a measure of high-quality capital held by banks to absorb losses—nearly tripled from about $275 billion to $800 billion over the past seven years for the 6 largest U.S. banks. This makes clear that the capacity for U.S. banks to absorb losses has risen significantly despite the decline in the market value of equity. As indicated in the Sarin and Summers paper, the decline in the market value of banks’ equity is due to the imposition of more stringent banking regulation and the low interest rate environment.
Second, Sarin and Summers use a post-crisis beta estimate of 1.59 for the largest 6 U.S. banks to conclude that a 60 percent decline in the stock market “would wipe out its equity.” However, the beta estimate used by Sarin and Summers is up for some debate in the academic literature. Generally, the beta estimates are time-varying and its value depends on the time period chosen by researchers in their empirical analysis. For instance, Baker and Wurgler (2013) use a much longer-time series than the sample period used by Sarin and Summers, and report a beta of 0.50 for well-capitalized banks. This implies that a 60 percent drop in the stock market would result in banks’ market equity declining by one-third. This decline in the market value of equity is in line with the average decline in common equity tier 1 regulatory capital ratios across all CCAR banks under the severely adverse scenario in the Federal Reserve’s stress test. Moreover, the V-Lab leverage ratio under stress—a much more refined approach to monitor bank risk using market data—is just slightly below 5 percent for the largest 6 U.S. banks (red line in Figure 1), well above the minimum tier 1 leverage ratio hurdle rate of 4 percent in CCAR. As shown in a previous TCH blog post, the stress scenario under the V-Lab methodology assumes a 40 percent fall in the stock market over a 6-month period and market losses under V-Lab are about 20 percent higher than losses reported as part of CCAR 2016.
Finally, we believe that it is also worth considering how large banks would perform under a CCAR stress relative to the rest of the financial sector. We suspect that analysis would show that such a stress – which would include not only a 50 percent drop in equity prices but also a large and sudden increase in unemployment, as well as a dramatic widening in credit spreads – would cause the failure of hundreds of small banks, as well as destabilize market-funded lenders, securities firms unaffiliated with large banks, hedge funds, and other non-banks that are a large and growing source of funding for the U.S. economy. Given the elevated levels of high-quality capital held by the largest banks, our view is that retail deposits, various capital market activities and other functions would migrate from the failed financial institutions to the largest banks. These effects are currently not modelled in the U.S. stress tests and should be studied and considered before increasing even further the capital requirements of U.S. GSIBs.