Jeff Speakes and Ray Hawkins
Banks utilize stress tests in order to estimate potential losses in extreme environments and to show that they have adequate capital to handle these potential losses. Advocates of stress testing argue that this is the best method to assure that the bank will continue to be viable even in a stressful environment. It can directly address the problem of tail risk that bedevils statistical models of risk. Critics make one of two arguments. One argument is that the scenarios that are often deployed are not sufficiently stressful, so that everyone passes. The other argument is that if severe scenarios are deployed after a major negative market event, then the effect of the tests is to increase capital requirements during the down cycle and thereby accentuate volatility.
In July, Federal banking regulators sought public comment from industry on proposed regulatory guidance for stress testing. This guidance will apply to banks regulated by the Federal Reserve, OCC or FDIC that have assets greater than $10 billion. Key elements of the proposed guidance include the following:
There should be clearly defined goals for the stress tests.
Management should clearly articulate the objectives for the post-stress outcomes.
The scenarios should be tailored to the bank’s business.
The framework should cover the full set of activities and exposures.
The results should be clear, actionable and well-supported.
The tests should assess the adequacy of capital and liquidity.
Commenters were requested to discuss the elements of the proposed guidance along with any additional elements that the commenters feel should be included. Comments are also encouraged on the challenges that may arise in implementing the guidance.
The following is our response to the regulators’ request.
Comment on Proposed Regulatory Guidance on Stress Testing
We support the use of stress testing in banking for risk management, capital adequacy and other purposes. We concur with many of the precepts contained in the proposed regulatory guidance; in particular that there should be clearly defined goals for the tests and that the results should be actionable. However, we believe that the set of stress scenarios needs further discussion and definition. In the proposed guidance, the onus is on each bank to define a set of stress scenarios that is appropriate for its own business strategy and risk profile. We believe the regulators should work with the banks and outside experts to define a small set of scenarios to be applied to each bank.
This regulatory guidance appears to be independent of the stress test directed by the U.S. Treasury in October 2009 for the largest financial institutions. The purpose of this test (the Supervisory Capital Assessment Program, or SCAP) was to ensure that the largest banks were adequately capitalized. Each bank’s portfolio was subjected to the same stress scenarios. We support the logic of the SCAP. This test was parsimonious, uniform across all applicable banks, and transparent. Indeed, we would take the SCAP example and extend it in certain respects.
Common set of scenarios
While we believe banks should be encouraged to conduct internal tests tailored to their specific business and risk profile, we also believe that there should be a common set of stress environments that is applied to all banks. This common set should be parsimonious and the results should be transparent. Different banks may use various methodologies to assess loss exposure in a given scenario. We believe the regulators can develop a useful standard by building the ability to carry out the calculations based on data supplied by the banks.
We know of at least two examples in which a common test is applied. First, many options and futures clearinghouses around the world have adopted the Standardized Portfolio Analysis of Risk, or SPAN, test to assess margin requirements. The purpose is to assess the worst case one-day losses on a given position. To accomplish this, the SPAN test utilizes just 16 scenarios.
Second, following the thrift crisis of the 1980s, thrift regulators adopted a uniform test of interest rate risk which is based on schedule CMR. Each thrift institution is required to estimate the change in net portfolio value under a variety of yield curve shocks. In addition, the OTS makes its own calculation of these sensitivities for each thrift institution using the Net Portfolio Value model.
The CMR report focuses on parallel yield curve shocks. More generally, scenarios should be based on positive or negative shocks to the slope of the curve as well as the level. Also, to address credit issues negative shocks to housing prices and positive shocks to the unemployment rate should be considered as well. This does not necessarily involve a large number of scenarios. In fact, four may suffice (short rates up or down, long rates up or down, each with unemployment up and housing prices down).
Capital adequacy
One reason for assessing potential losses under stress scenarios is to assess the degree of capital adequacy. We believe the standard methodology in this area is flawed. Typically, banks are held to a “AA” credit standard wherein the probability of default within one year is extremely remote, like 3 in 10,000. The problem with this metric is twofold. First, it is extremely difficult to precisely measure an event that occurs three times in ten thousand years. Second, the event is so remote that senior management relates neither to the concept nor the calculation. A better metric would be tied to the event of losing an investment grade rating. That is, capital should be maintained such that the chance of losing the bank’s investment grade rating over a reasonable time frame is small, something like 1 in 100 over a three month horizon.
Challenges
We recognize potential risks and challenges to our approach. We have just passed through an extreme economic downturn and instituting more severe capital requirements today could make recovery that much more difficult. Still, we should assess risk in a rigorous way, and if greater capital is required then this requirement can be implemented over time.
Observing the confluence of pro-cyclical factors exhibited in the recent financial crisis, it seems clearly desirable to establish regulatory standards that are counter-cyclical in nature. Yet, estimates of probable loss will tend to be greater following a major market setback than before. That is, during a boom perceived risk tends to be smaller than during the subsequent bust. This effect can be addressed in at least a partial way by using estimates of capital adequacy based on cycle average information.
Thirdly, it is sometimes argued that a common set of stress scenarios is subject to gaming by financial engineers. But this has not proven to be the case either with SPAN or CMR. The key is clever selection of the set of stress scenarios.
Finally, it should be recognized that the process of carrying out stress tests is complex and requires significant expenditure of resources. The largest banks in the country undoubtedly have adequate managerial and computational resources to effectively carry out the task, but the burden on smaller banks could be great. The guidance should address this issue by emphasizing a parsimonious set of scenarios.
The focus that the guidance places on designing stress scenarios that are tailored to the particular bank and address all the important areas of risk raises the raises the danger of scenario proliferation. This could put undue burdens on managerial and computational resources. It could also lead to lengthy and ultimately unproductive discussions between management and regulators regarding the rationale behind scenario selection. Again, we believe the best solution is for the regulators to develop a small set of uniform scenarios.
Summary
We believe bank regulators should follow the lead of the futures/options clearinghouses and the OTS in setting forth a common set of scenarios for all banks to use, and develop the regulatory ability to estimate losses in these scenarios. By doing this, bank regulators will push forward the standard of excellence in bank risk management.