Comment on the OFR brief “Capital Buffers and the Future of Bank Stress Tests”

Comment on the OFR brief “Capital Buffers and the Future of Bank Stress Tests”

On Tuesday, researchers at the Office of Financial Research (OFR) published a brief — “Capital Buffers and the Future of Bank Stress Tests.” In the brief, the OFR suggests that the Federal Reserve’s stress tests should be tightened. The stress tests have been an important and successful addition to the post-crisis supervisory regime, but they are already designed to ensure that banks can withstand a recession more severe than the great recession, a major meltdown in global financial markets, and the failure of the banks’ largest counterparty. The OFR note simply asserts that tighter is better and makes no consideration of the negative consequences for economic growth and employment that would result. Moreover, the reasons the note provides for their suggested change are myopic or flawed.

The Federal Reserve conducts an annual stress test of the 34 largest banks to assess whether those banks have enough capital to withstand a severe recession over a 9-quarter horizon. The very largest of these banks must also show they can weather the default of their largest counterparty, and for six of the largest banks a global market shock, and still have enough capital to function well. While we have argued elsewhere in favor of changes to how the tests are conducted (see a complete list of our analysis at, we nevertheless continue to think that on the whole, stress tests are a beneficial addition to the post-crisis regime used to supervise banks.

The stress test complements the standard risk-based capital requirements that banks must meet on a continuing basis. The core of those requirements, which focuses on common equity tier 1 capital, includes four components: a 4.5 percent minimum requirement, a 2.5 percent capital conservation buffer, a surcharge for the eight largest U.S. banks (the global systemically important bank holding company, or GSIB surcharge) and the countercyclical capital buffer for the advanced approaches institutions. The GSIB surcharge depends on the size, complexity, and a few other factors of the bank as determined under the Federal Reserve’s rule and varies from 1.5 percent up to 3.5 percent. Currently, JP Morgan Chase has the highest surcharge among U.S. banks.

In the note, researchers at the OFR argued that the Federal Reserve should add the GSIB surcharge to the post-stress hurdle rates included in CCAR, in line with a proposal by a Fed Governor in a speech last fall, but that the Fed should not change the stress test assumption that bank balance sheets continue to grow over the stress period, as opposed to assuming the balance sheets remain constant. There are several problems with the OFR analysis.

The reasons provided in the OFR note to add the GSIB surcharge to the stress test hurdle rate are unsound.

The OFR paper lists four reasons why the authors believe GSIB surcharges should be added to the stress test hurdle rates.

  1. It would make banks more resilient during stress.
  2. Banks need to be required to have enough capital now so that post-stress they still pass all the day-to-day requirements; otherwise banks will cut back on lending during the downturn.
  3. If the GSIB surcharge is not included, the tests are tougher on non-GSIBs, who do not face day-to-day GSIB surcharges and so will be more likely to fail.
  4. The leverage ratio is the requirement that often binds post stress, and a binding leverage ratio is undesirable because it encourages banks to take on more risk.

The first reason is incomplete, and the remaining three are mistaken. We address each in turn.

  1. It would make the financial system more resilient under stress.This rationale is myopic — it focuses on the benefits of higher capital to overall resilience while ignoring all other consequences. Adding the GSIB surcharge may make the eight largest U.S. banks more resilient under stress, but it also reduces the availability of credit and increases lending rates, thereby causing a reduction in economic activity and employment in normal times; the question that needs to be answered is whether the benefit of increasing capital requirements further exceeds its cost. As documented in a recent TCH research note stress tests establish the binding capital requirement for the majority of institutions that participate in the stress tests. If banks are required to pass a higher hurdle rate after losses are projected over the stress episode, then they will have to hold more capital day-to-day. Setting capital requirements always requires making a choice between greater financial stability or less growth. The BIS standard study on the costs and benefits of capital and liquidity requirements indicates that the cost of higher capital begins to exceed the benefit when capital requirements are in the range of 10 to 13 percent. We estimate that passing the quantitative portion of CCAR currently requires banks to hold a little more than 10 percent capital, on average, given the CCAR results in 2016. If the hurdle rate is raised by the GSIB surcharge, which averages 2.8 percent, capital requirements will move up to levels estimated by the BIS to potentially have negative returns.In addition, if risk migrates from the GSIBs and into the rest of the banking system or the shadow banking system, adding the GSIB surcharge to CCAR hurdle rates may not increase the resilience of the financial system at all.
  2. Banks need to be required to have enough capital now so that post stress they still pass all their day-to-day requirements and can keep making equity distributions; otherwise banks will cut back on lending during the downturn.If the hurdle rate for the stress test is increased, banks will have to hold more capital on a day-to-day basis to be able to pass the stress test. Specifically, as discussed above, banks have to maintain about a 10 percent capital ratio on a day-to-day basis, on average, to pass the stress tests, and that number would rise to about 13 percent if the GSIB surcharges were added. But those higher effective continuous requirements would apply during and after a downturn as well as in normal times. As a result, banks will lend less, not more, if the hurdle rate is raised, both in ordinary times and in a downturn. More generally, while it is true that when banks have more capital they tend to make more loans, a recent TCH research note demonstrates that this positive relationship between bank capitalization and loan growth is driven by the amount of capital banks have in excess of capital requirements. We find that raising capital requirements slows lending both directly and because it reduces banks’ capital surplus.
  3. If the GSIB surcharge is not included, the tests are effectively tougher on non-GSIBs, because non-GSIBs have lower day-to-day capital requirements and so will be more likely to fail the stress tests.The OFR note is apparently arguing that failing to make the stress test tougher on a set of banks that are separately being held to a higher standard implicitly makes it tougher on another set that is not held to those separate higher standards. The extent of that illogic is even deeper when one recalls that GSIBs actually already face a tougher stress test than others because they are subject to the assumption that their largest counterparty fails and most of them are subject to a global market shock.
  4. The leverage ratio is the requirement that often binds post stress, and a binding leverage ratio is undesirable because it encourages banks to take on more risk.We agree that the leverage ratio hurdle is the post stress hurdle that often binds (see TCH research note “Capital Allocation Inherent in the Federal Reserve’s Capital Stress Test”). We also agree that a binding day-to-day leverage ratio is undesirable because it mismeasures balance sheet risks and encourages banks to take on more risk (see TCH research note “Shortcomings of Leverage Ratio Requirements”). However, as explained in the TCH research note on the stress tests, a binding post stress leverage ratio hurdle is much less likely to encourage banks to take on more risk, and more accurately reflects real risk because banks with riskier portfolios incur higher losses during the stress episode. As a result, a post-stress leverage ratio hurdle acts much like a risk-based capital requirement.

The OFR note’s argument that assuming banks’ balance sheets expand during stress will encourage lending in a crisis is flawed.

As noted above, currently under CCAR, the Fed calculates each bank’s post-stress regulatory capital ratios under the assumption that banks’ balance sheets continue to expand under the severely adverse scenario. The OFR argues that the Fed should not adopt recent suggestions that the stress tests instead assume that balance sheets remain unchanged. Assuming that risk-weighted assets and total assets remain constant over the stress horizon is a natural assumption to communicate and implement, and prevents banks from improving their post-stress regulatory capital ratios through excessive reductions in their balance sheets during stress. The OFR argues, however, that continuing to assume balance sheets expand will ensure that banks have enough capital in normal times to support increased lending in a downturn.

In practice, this assumption of an expanding balance sheet over the stress test stress episode ends up reducing banks’ projected post stress capital ratios and so is equivalent to an increase in the stress test hurdle rate. As a result, to pass the stress test, banks have to have even higher capital ratios at the start of the test. Following a stress episode banks would have a strong incentive to still be able to pass a stress test, thus avoiding a restriction on their ability to pay dividends. So boosting the requirement as proposed would result in less, not more lending in a downturn.

Disclaimer: The views expressed in this post are those of the author(s) and do not necessarily reflect the position of The Clearing House or its membership.