The primary focus of post-crisis financial regulation has been to make large bank holding companies – that is, banks and their non-bank affiliates – more resilient and resolvable. The former includes sharp increases in capital requirements and first-of-their-kind liquidity requirements that have dramatically increased liquidity levels. The latter includes a reinvention of how such companies are resolved as a legal matter as well as the issuance of more than $1 trillion of so-called “bail-in” debt to make such resolution a practical reality. Those changes have focused almost exclusively on larger banks – those with $50 billion or more in assets. Much of the debate about post-crisis regulation is whether those requirements, when viewed holistically, are under- or overcalibrated and whether their costs are worth their benefits.
In this issue of Banking Perspectives, we attempt to answer a somewhat broader question: Has the enhanced resilience and resolvability of large banks reduced the chances of the next financial crisis – or actually increased those chances? Of course, those reforms certainly have reduced the chances that a large bank or its non-bank affiliate will cause or propagate such a crisis, which leads many to assume that the chances of a financial crisis are lower. But financial crises are about collapses in asset prices and unpredictable (and unpredicted) contagions and connections.
One safe assumption about the next crisis is that it will come as a surprise – to everyone. In their book A Crisis of Beliefs: Investor Psychology and Financial Fragility, excerpted in this issue, Nicola Gennaioli and Andrei Shleifer present considerable research on the beliefs and expectations of all relevant actors as the financial crisis loomed, and they use concepts from behavioral economics to provide insights about future crises. Importantly, they debunk what they refer to as the “moral hazard” or “too big to fail” theory of the crisis: the view that large financial firms saw the risk but took it with the knowledge that subsequent taxpayer support would limit their losses but not their gains. As they report:
[W]e have little evidence that the banks had any superior knowledge about the risk of the assets in which they invested, at least in 2005 and 2006 …. To the contrary …, banks showed substantial optimism and disregard for downside risks. Likewise, the study by Heng, Raina and Xiong … showed that securitization specialists were just as optimistic about real estate as others. This evidence points against the view that bankers knew something about the housing bubble that others did not. When the bubble collapsed, they lost money on their homes, they lost money on equity holdings in their firms, and many lost their jobs. This does not look like superior knowledge.1
That financial crisis came as a considerable surprise not only to the market but also to the Federal Reserve as the central bank and bank regulator. As Gennaioli and Shleifer relate:
Only six weeks before the Lehman bankruptcy, in early August 2008, both the Federal Reserve and professional forecasters predicted continued growth of the U.S. economy. Contrary to that prediction, the U.S. financial system nearly melted down after the Lehman bankruptcy, and the economy slid into a deep recession.
The [Federal Reserve’s] June 2008 Greenbook forecast next year  real GDP growth at 1.13 percent, which was actually revised slightly upward on July 30. … Perhaps most remarkable in this regard is a document prepared by the Fed Forecasting staff on July 30, 2008, for the August 5, 2008 FOMC meeting. As the governors were aware of the stresses in the financial system, the staff was asked to prepare the forecasts for the scenarios of “severe financial stress,” which was the worst case they considered. In this scenario the Fed forecasters expected -0.5 percent real GDP growth in the second half of 2008 with real growth rising to 0.5 percent in 2009 and 2.6 percent in 2010. The unemployment rate in this scenario was expected to peak at 6.7 percent in 2009. It actually peaked at 10 percent. Six weeks before Lehman, the Fed forecasters had no idea what was coming.2
From a distance, it is difficult to understand the monomaniacal focus of post-crisis regulation on large commercial banks and their non-bank affiliates, and the indifference or even enthusiasm with which regulators have watched much commercial banking and trading move outside the regulated banking sector. One of the many intriguing parts of the article in this issue by Anil Kashyap, Dimitrios P. Tsomocos, and Alexandros P. Vardoulakis is a disclaimer at the beginning: “Throughout, we will refer to these institutions as banks, but understand that non-banks may often also undertake some of these activities, and it is often conceptually hard to explain why regulations are based on organizational form rather than function.” Indeed. I’ve periodically asked people the following question, “Is Goldman Sachs subject to enhanced prudential standards and a G-SIB surcharge because of the risk that its commercial bank will impose a loss on the Deposit Insurance Fund or because of the systemic risk inherent in its non-bank broker-dealer?” The answer has been the latter, without exception. And yet, if that is the case, why are significant players in the financial markets unaffiliated with a bank not subject to such standards – and indeed not subject to any prudential standards at all?
I think it’s easy for a cynic to conclude that the focus of the federal financial regulators is either political or self-interested. Clearly, large banks are political bogeymen, and regulators will be held to account if they, as opposed to non-banks over which they exercise no jurisdiction, run into trouble. I think, however, the motivation may be subtler and more reasonable. Exactly because regulators cannot know what the cause of the next crisis will be or how it will spread across markets, they want to be absolutely certain that there is a safe port in the next storm, and they have chosen the banking system that they oversee as that port. That choice is understandable given banks’ stable deposit base, their now-unique access to the discount window, and the fact that the regulators regulate and examine them. Thus, so the thinking might go, we can worry less about the non-banks because we will count on banks to continue extending credit and making markets.3
Meanwhile, bank regulation becomes more and more complex over time, leading to more restrictions, whether through regulation (overt) or examination (covert). In their article, Kashyap and his co-authors find that when a standard economic model of banking is expanded to reflect a broader range of services that banks provide (for example, providing deposits and making information-intensive loans), the results indicate that multiple regulatory requirements are needed to get the socially optimal outcome. This seems intuitively correct.
The issue, though, is the cost of a multiplicity of rules aimed at only 15 to 20 financial institutions and their activities. There are two main costs. First is a reduction in the level of economic activity, which is inarguable, yet rarely acknowledged in regulatory proposals or academic work.4 (For example, while in their book, Gennaioli and Shleifer note that various factors lead to increased risk of financial instability – such as securitization and innovation – they do not acknowledge that there may be a societal cost to constraining those factors.) Second, and the focus here, is a possible shift (as opposed to reduction) of risk to those who may be less able to manage it – either non-bank financial companies or consumers or non-financial businesses – or to government-guaranteed securities for which taxpayers directly bear the risk.
We already see evidence of this in numerous markets. Non-banks originated more than half of mortgages in 2017 and account for about 80% of originations of mortgages insured by the Federal Housing Administration and the Department of Veterans Affairs. Moreover, non-banks are vulnerable to liquidity pressures both at origination and during the servicing of such loans; thus, non-bank failures could be quite costly for taxpayers. Banks are shifting out of holding the term part of leveraged loans, which are largely funded instead in collateralized loan obligations and non-banks now dominate home lending to low- and moderate-income (LMI) borrowers. Moreover, small-dollar lending to LMI communities is increasingly being done by non-bank lenders. Market depth in corporate securities markets has diminished and as bank-affiliated broker-dealers hold less inventory the capital cost of holding and hedging inventory has risen. In sum, post-crisis regulation has forced banks and their affiliates to manage fewer risks for the economy, leaving those risks either to be managed by non-bank financial companies; not to be managed at all, and thus retained by non-financial firms; or held by the government.
What is of even greater concern, though, is how markets will behave in the next financial crisis.
Consistent with the view that banks will serve as a safe port in the next financial crisis, regulators not only have imposed significantly higher minimum capital requirements on banks but also required them to hold so-called “buffers” above those minimums. One is a capital conservation buffer of 2.5%; another is the so-called G-SIB surcharge that currently ranges from 0.0% to 3.5%; and a third, though currently set at zero in the United States, is the countercyclical capital buffer. The Fed has also proposed a fourth buffer, the stress capital buffer, which would be based on hypothetical losses under the Fed’s most recent Comprehensive Capital Analysis and Review (CCAR) stress tests, which already require a bank to meet capital minimums and hold a buffer post-stress. A fifth, de facto, buffer may be forthcoming in the form of the current expected credit loss (CECL) accounting methodology, which will require banks at the origination of a loan to establish a reserve against all future expected losses (with no offset for future expected income). As Stephen G. Ryan notes in his article in this issue, “under CECL, banks will record far larger [provisions for loan losses] at inception for both heterogeneous and long-lived homogeneous loan types than they accrued under the [incurred loss model].”
The impact of these buffers can best be seen in this chart, which compares the projected peak-to-trough pre-tax net losses from the Federal Reserve’s recent CCAR stress (in red) with the amount of capital held by the CCAR blanks (in dark blue) and the amount of loan loss reserves and bail-in debt – that is, debt readily convertible to capital – held by those banks (in light blue).
Thus, if one reads the red bar as the amount of capital necessary for the largest banks to survive a financial crisis even worse than the global financial crisis of 2007–2008, which is how the Federal Reserve calibrates its stress scenario, then you can see that the affected banks hold a mountain of buffer. If those buffers will never be used, they represent a significant drain on banking activity and economic growth.
The rationale for these buffers is that when stress occurs, banks will treat these them as buffers, not legal or practical minimum requirements, and therefore allow their capital levels to drift down toward regulatory minimums in order to remain engaged in lending, market making, and other activities that support the economy. They will also use their buffers to acquire troubled firms, or their assets, lessening fire-sale risk. This belief is an important foundation of post-crisis capital (and liquidity) regulation. Indeed, one could say that is the most important foundation of post-crisis capital (and liquidity) regulation.
Unfortunately, I’ve never met a market participant who believes it. Rather, market participants universally believe that banks will never allow their liquidity and capital levels to drop, even if that means shrinking their balance sheets and turning away deposits. In his article, Douglas J. Elliott describes those reasons ably. He provides what I think is an important insight, namely, “[I]t’s possible that the key to predicting actions by banks and their key constituencies is not the total level of capital and liquidity but the margin of capital and liquidity above regulatory minimums. If so, the recession-related risks have likely become higher, not lower, as a result of the increased requirements.”5 My colleagues and I have a less eloquent phrase we use to express the same thought: the Great Buffer Fallacy.
In sum, when internal constraints or credit rating agency standards operate as the binding constraint on bank capital and liquidity levels, a bank might choose to operate at somewhat lower capital or liquidity levels under stress – continuing to lend and make markets. By contrast, when supervisory or regulatory constraints are the binding constraints, banks will be unwilling to operate at lower levels even if those constraints are temporarily reduced or “meant to be used.” This could make for a messy crisis. As Elliott states, “All in all, there is clearly a risk that non-banks, taken as a whole, could either choose or be forced to pull back on their activities more sharply in a recession than banks have historically done or would likely do in the future.… The aggregate impact of much tougher prudential requirements for banks without any significantly expanded regulation of non-banks that operate in similar markets has likely increased pro-cyclicality.”
We already have some hints of how markets might behave, based on rather low-stress events like reporting dates. At the end of last year, increased demand for repo to fund expanding government debt met reduced supply from banks concerned about their leverage ratios and G-SIB surcharges; as a result, Treasury repo rates traded above 7%, nearly triple their normal rate. The ramifications here are significant enough that they have reduced the Federal Reserve’s options in monetary policy: The Fed’s efforts to reduce the size of its balance sheet have run headlong into heavy bank demand for reserves, driven importantly by the liquidity coverage ratio and other liquidity requirements.6
BEHAVIORAL ECONOMICS AND FINANCIAL REGULATION
As an early adherent of behavioral economics and the work of Kahneman and Tversky in particular,7 I believe that work can have valuable contributions to financial regulation. We publish here an article that examines how our innate biases make it difficult for us to foresee crises – something readily observed in recent financial crises. I think a far more significant application could be toward how buffers would work in practice, which reflects how banks and regulators would behave. Why would a bank in the midst of market uncertainty and regulatory scrutiny choose to report that it was no longer in compliance – or, put another way, no longer able to comply – with its regulatory capital and liquidity buffers? For that matter, would a regulator in the midst of market uncertainty publicly allow a bank to do so – for example, by lowering a previously imposed countercyclical capital buffer? Anecdotal evidence and some back-of-the-envelope game theory seems to suggest that they would not.
To quote Amos Tversky, “People are not so complicated. Relationships between people are complicated.”8 So it is with financial regulation, and it seems that the failure to consider the relationship between the minimum requirements imposed on banks, the “buffers” imposed on banks (which are almost certainly de facto minimum requirements), and the absence of requirements on non-banks is an area ripe for study through the lens of behavioral economics and game theory. I fear we would not like what we would find (and further fear that may be why some people aren’t looking). n
1 Nicola Gennaioli and Andrei Shleifer, A Crisis of Beliefs: Investor Psychology and Financial Fragility (Princeton, N.J.: Princeton University Press, 2018).
3 The preamble to the Federal Reserve’s countercyclical capital buffer final rule states, “It is designed to increase the resilience of large banking organizations when there is an elevated risk of above-normal losses. Increasing the resilience of large banking organizations will, in turn, improve the resilience of the broader financial system” (emphasis added). “Regulatory Capital Rules: The Federal Reserve Board’s Framework for Implementing the U.S. Basel III Countercyclical Capital Buffer.” 12 CFR 217, Sept. 16, 2016. https://www.federalregister.gov/documents/2016/09/16/2016-21970/regulatory-capital-rules-the-federal-reserve-boards-framework-for-implementing-the-us-basel-iii
4 A reduction in economic activity is the cost of higher capital requirements used in studies of the optimal level of capital by the BIS (2010), the Bank of England (2015), the IMF (2016), the Minneapolis Fed (2017), and the Federal Reserve Board (2017).
5 It is worth noting that the countercyclical capital buffer was designed with the intention that regulators would lower it in times of financial stress, and Elliott therefore believes it is worth further study as a potential compromise in this area. Of course, this means that the buffer will be imposed sufficiently ahead of time – that is, that the same regulators who could not identify a looming crisis a few weeks ahead of its advent last time will do so a year or more ahead next time – and that they will have the political courage to tell the American people and the Congress that their response to an actual, ongoing crisis is to lower the capital requirements on large U.S. banks.
6 “The Future of the Federal Reserve’s Balance Sheet,” Vice Chairman for Supervision Randal K. Quarles, speech at the 2019 U.S. Monetary Policy Forum, sponsored by the Initiative on Global Markets at the University of Chicago Booth School of Business, New York, New York, February 22, 2019.
7 G. Baer and G. Gensler, The Great Mutual Fund Trap (New York: Random House, 2002), 170-75.
8 “The Undoing Project: A Friendship That Changed Our Minds,” Michael Lewis (W. W. Norton & Company, 2016).