Yesterday, the Federal Reserve announced the results of its annual stress tests. This was the first time since 2009 that the Fed had stress tested large banks during a period of systemic distress. In a new paper, Stress Testing During Times of War, forthcoming in The Stress Test Handbook (Cambridge University Press) and available here, I explain why stress testing after a negative shock is both more important and more fraught than stress testing during times of peace. I further address why time-limited government backstops may be critical to ensuring the tests are robust, relevant and disclosed – the criteria that must be satisfied for crisis-era stress tests to be useful. Based on these criteria, there were meaningful weaknesses in how the Fed approached its most recent round of stress tests.
As a threshold matter, it is important to distinguish stress testing during times of peace and times of war. Over the past decade, the focus has been almost exclusively on stress testing during times of peace. The Fed and other central banks have incorporated stress testing into their standard supervisory toolkit. In this role, stress tests have provided a useful forward-looking complement to more traditional, backward-looking capital adequacy requirements. As the Covid crisis has made clear, however, these regular stress tests can never assure that banks are adequately prepared for what the future may bring. The future is too uncertain, and even well-run stress tests can never fully capture the feedback loops that can take hold when crisis sets in. Thus, although regular stress tests should remain central to the supervisory toolkit, their results must always be taken with a grain of salt, and bank regulators must remain attuned to the possibility of threats and weaknesses not captured.
This backdrop helps to explain why stress testing during periods of systemic distress remains vital. When fault lines emerge, regulators have a much clearer sense of the type of adverse developments that may threaten stability, where weaknesses may lie, and the specific fears that they must address to restore market functioning. With respect to Covid, for example, the rate of economic contraction has been far faster than it would be in a traditional recession, and in this regard it was out of line with even the most adverse scenario the Fed originally proposed for the 2020 stress tests. The Covid crisis has also affected different types of businesses and other borrowers differently. By using the information available about the distinct ways that the pandemic and public policies designed to contain its damage may affect the value of bank assets and bank income in the future, the Fed could provide far more insight into how banks are likely to fare than any generic assumptions could yield.
The actual stress tests undertaken by the Fed make modest progress in this regard. The core stress tests were conducted using the scenarios that the Fed devised before Covid became a systemic threat. One reason that the Fed chose to run these stress tests without modification is that, earlier this spring, the Fed finalized rules incorporating stress test results into a new “stress capital buffer.” A primary aim of the new scheme was to simplify, modestly, the range of capital requirements that banks must face. But one effect is that the Fed may have felt that it had less flexibility to alter the stress testing to improve the relevance of the results while potentially making other compromises.
Nonetheless, the Fed recognized that to simply undertake the previously planned stress tests without any regard to Covid made little sense. Hence, it opted instead for a middle path. It added a new set of “sensitivity analyses” designed to assess how each bank would fare under three possible scenarios: a quick V-shaped recovery, a U-shaped recovery entailing a longer downturn before the recovery takes hold, and a W-shaped recovery, in which a second shock derails the recovery, perhaps because of a second wave of Covid-19 infections. These analyses, however, were not run with as much data or rigor as the Fed’s stress tests, hence the new name. More importantly, the Fed only disclosed aggregate results for how the banks fared inn these various scenarios.
The paper provides a frame for explaining why it is useful that the Fed at least undertook these sensitivity analyses. The evolution of the Covid crisis and its impact on bank health remain far from certain. One of the common ways that a crisis can grow is by a failure to recognize and redress weaknesses within banks in a timely way. The Fed has already taken criticism (and I have been among the critics) for its failure to staunch dividend payouts to shareholders, in contrast to the Bank of England and the European Central Bank. The sensitivity analyses could help reveal whether this type of intervention or other more aggressive action to increase banks’ capital cushions may be warranted. On a more granular level, the sensitivity analyses could also help alert the Fed to deficiencies that may not have been as pressing or apparent under the more generic assumptions used for the stress tests. The earlier that regulators and banks can identify potential weaknesses, the more likely they can address them before a full-blown panic.
This is where the more limited scope and lack of individualized results of the sensitivity analyses become more troubling. As the paper explains, one of the most important benefits of stress testing during periods of systemic distress is to provide market participants, including a bank’s counterparties and other creditors, credible information about the capacity of that bank to weather the storm ahead. In providing only aggregate disclosures, the Fed could well darken rather than dispel the cloud of uncertainty that hangs over the banking sector. The effect could thus be to increase, rather than reduce, fragility.
This does not mean that the Fed’s decision to provide only aggregate results was completely unjustified. A core challenge when it comes to stress testing during periods of systemic distress is that the situation is already so fragile. Any new, adverse information could well trigger a run, exacerbating the very crisis regulators are seeking to contain. The 2009 stress tests were possible only because the Emergency Economic Stabilization Act of 2008 had given the U.S. Treasury broad new authority to shore up the financial system, and thus the Treasury could credibly claim that it would recapitalize any bank that came up short and that could not raise capital without government aid. Although Treasury did not need to inject any new funds as a result of the original stress tests, the knowledge that it could do so was critical to providing the Fed and other regulators comfort that they should seek the truth, no matter how bad the news might be.
Neither the Fed nor Treasury currently has any authority to recapitalize banks should they need it. The Coronavirus Aid, Relief, and Economic Security Act, commonly known as the CARES Act, does give them authority to provide fiscal support for businesses, and the Fed has already instituted an array of emergency facilities, but neither body has anywhere near the discretion that Treasury enjoyed in 2009. The bigger point is that the Fed can only be expected to run rigorous stress tests after a crisis is underway if it is assured that regulators can contain the fallout of any information that may be revealed. Looking ahead, this is one reason it may be appropriate to create special, time limited authorities that regulators can invoke only during periods of systemic distress. The paper provides some suggestions and other guidance about how regulators can better prepare for stress testing during times of war.
Crisis-time stress testing will always be fraught, but it is also critical. The paper explains why these tests are so important, how to prepare and how to do them right when the time comes.
This post comes to us from Kathryn Judge, the Harvey J. Goldschmid Professor of Law at Columbia University Law School. It is based on her recent paper, “Stress Testing During Times of War,” forthcoming in The Stress Test Handbook (Cambridge University Press) and available here.