How risky are the big U.S. banks?
Readers of this blog know that we are great fans of the Stern Volatility Lab’s estimates of systemic risk. Like many observers, including leading regulators, we find market-value rather than book-value measures of bank equity more useful for timely monitoring of systemic risk created by individual intermediaries. Equity prices are available in real time, rapidly incorporate bank-specific and economy-wide information, and are forward-looking. This makes them particularly helpful in assessing the impact of big events, like this summer’s Brexit referendum (see our earlier post).
So, based as it is on market indicators of bank risk, not surprisingly we share the recent assessment of Sarin and Summers (expressed in their September 2016 Brookings paper) that the increase of book capital in the banking system since the financial crisis ought not give rise to regulatory complacency. We have argued repeatedly for raising capital requirements (see, for example, here) and, like those authors, believe that we need mechanisms for the virtually automatic recapitalization of banks in a crisis (see here).
What Sarin and Summers highlight is that, compared with their pre-crisis readings, a range of market measures of bank risk do not show improvement despite the rise of reported bank capital. Consider, for example, the following key indicators for the Big 6 U.S. banks (Bank of America, Citigroup, Goldman Sachs, JP Morgan, Morgan Stanley, and Wells Fargo) over two time periods: the pre-crisis years and 2015.
Market Indicators for U.S. Big 6 Banks: Pre-crisis average and 2015
* Tier 1 capital ratio is reported in the Sarin-Summers text. 2015 figure is the average for 2010-2015. Note: The pre-crisis period is 2002 to 2007, with the exception of implied volatility data that begin in 2005. Source: Table 1 Panel A in Natasha Sarin and Lawrence H. Summers, “Have big banks gotten safer?,” BPEA Conference Draft, September 15-16, 2016.
One would think that as capital rises, market measures of risk should decline. That is, we would expect that with larger buffers, equity price volatility, correlation with the equity market (beta), and the probability an equity price collapse (option delta) would all be lower. In addition, we would think that preferred stock prices should benefit from the drop in the risk-free interest rate between the two periods. Yet in some cases, what should have gone down, went up. The option delta, which indicates the probability of a 50% plunge of the stock price over the next year, rose on average for the Big 6 from 3.6% before the crisis to 4.6% in 2015. And, instead of rising, the average price of preferred stock fell. Even when the direction of movement is consistent with theory, as in the case of equity volatility, the observed change has been substantially smaller than reported increases in capitalization suggest they should have been.
There is a sense in which we don’t find these data all that surprising, as we have long observed a similar pattern in the Volatility Lab’s measures of SRISK. The figure below plots end-month SRISK both for all U.S. financials and for the Big 6. Comparing the five years prior to the crisis with 2015, we see that aggregate SRISK rose from an average of $240 billion to $341 billion (the black line), while Big 6 SRISK climbed from $68 billion to $193 billion (the red line). In other words, the increase in Big 6 SRISK more than accounts for the rise in total SRISK.
U.S. SRISK: Total of all financials and Big 6 banks (U.S. dollars in billions), June 2000-September 2016
Source: Stern Volatility Lab.
As Sarin and Summers note, several factors could account for this pattern: (1) overstatement of the increase in bank capital; (2) pre-crisis optimism; or (3) a loss of bank franchise value. Like the authors, we doubt that the cause is an overstatement of the increase in bank capital. If anything, the opposite is more likely: since the crisis, regulators have tightened the definition of Tier 1 capital substantially (see, for example, Table 1 here). Put differently, regardless of whether reported capital ratios today are overstated, their overstatement prior to the crisis was surely much larger.
We are more sympathetic to the possibility that pre-crisis optimism played a role in buoying market-based measures of bank risk. There are numerous reasons to think that investors today perceive the world in general and banks in particular as riskier than they did a decade ago. As evidence, we highlight two key assets for which the crisis altered perceptions of downside risk—residential housing and corporate capital—both of which matter significantly for banks.
Consider first the probability distribution of annual percent changes in U.S. nominal house prices. As the following chart shows, over the more than half century from 1950 to 2007, economy-wide house prices fell in exactly one year (1991) and then by less than 1%. Partly reflecting that record, former Federal Reserve Chair Alan Greenspan expressed doubt as late as 2005 that there was a residential housing bubble in the country as a whole. Yet, from their peak in 2006, nationwide house prices fell for five consecutive years, resulting in a cumulative plunge of 24%. The only comparable period was the 26% collapse during the Great Depression of the 1930s. Why should we care about nominal (rather than real) prices? One reason is that many lenders in the run-up to the 2007-2009 crisis relied on rising collateral values (rather than income documentation) to roll over high loan-to-value subprime mortgages that had been originated with low, but temporary, teaser rates. When house prices began to drop, this practice became infeasible and defaults surged.
U.S. residential housing prices (annual percent change), 1950-2015
Source: Robert Shiller spreadsheet from Irrational Exuberance, 3rd edition, Princeton, 2016.
The second is the return on corporate capital (see next chart). Kozlowski, Veldkamp, and Venkateswaran (KVV) estimate that from 1946 until 2007 the real return to capital in the U.S. corporate sector averaged 12.4% annually with a standard deviation of only 4.1%. In 2008, that return dropped to a new low of +0.4% before plunging to -9.4% in 2009 at the height of the crisis.
Annual returns on U.S. nonresidential capital (percent), 1946-2015
Source: Revised and updated data provided by KVV, based on Figure 3 here.
KVV go on to argue that, once people update their views about the true probability of asset returns following the occurrence of a rare disaster like the plunge in house prices or in the return on nonresidential capital, their perceptions of risk can change for a very long time to come. And, from our perspective, since people are likely very averse to disaster risk, the impact of an event like the 2007-09 crisis can have a significant impact on required returns. Considering the revised perceptions of disaster probabilities together with profound disaster aversion, it is surely possible that the required return on bank equities has risen more than enough to offset the favorable impact of higher bank capitalization.
Finally, there is the possibility that the fall in equity valuations is a consequence of a fall in bank franchise value. Why might franchise value have declined? First, a range of factors—including the challenge of overcoming costly conflicts of interest, the obvious post-crisis need to improve risk controls, and heightened regulation—has boosted big-bank operational costs and closed off some profitable but risky business activities. Second, the pre-crisis franchise value for the Big 6 and other systemic intermediaries presumably incorporated the implicit taxpayer subsidy from their perceived status as too big to fail (TBTF). While TBTF remains an important issue, there is some evidence that the subsidy has diminished (see, for example, the 2014 report of the U.S. Government Accountability Office and Chapter 3 of the April 2014 IMF Global Financial Stability Report). Third, as Philippon has written, new technologies favor competition that squeezes the returns on traditional banking services like payments and consumer credit.
So, what to conclude? One doesn’t need to agree precisely on the reasons for the warning signs emanating from market indicators of bank risk. These warnings are sufficient to warrant continued close scrutiny from regulators and sustained efforts to increase the resilience of the financial system. We share the Sarin and Summers view that enhanced regulation, including Dodd-Frank, has made the system safer than it was before the crisis. We also agree that these improvements are insufficient, and that among other things, banks need more equity capital than they have today.
Where the most systemic banks are concerned, it is surely better to be safe than sorry.