WASHINGTON — The Government Accountability Office suggested minor tweaks to the Federal Reserve's stress tests, effectively endorsing the agency's models and disappointing bankers who had hoped the watchdog would push for greater transparency.
In a report Tuesday, the GAO said the Fed could improve its communication skills with respect to certain aspects of the stress testing regime and might do more to ensure the efficacy of its models.
But it was not the kind of rebuke that many Republicans and the banking industry were hoping for, something that could have provided ammunition to retool the annual tests in ways that the banks favor.
"The recommendations are really tweaks," said Mike Alix, a principal at PwC and U.S. risk and regulatory consulting practice leader. "It's not a major overhaul. It is consistent with the direction that the Fed was attempting to go in any case."
The report outlines findings related to three topics: communication related to the qualitative test; the process for creating and issuing stress scenarios; and examination and improvement of capital performance models and model arrays.
The Dodd-Frank Act mandated that the Federal Reserve execute an annual evaluation of all banks designated as systemically important financial institutions — that is, banks with assets of more than $50 billion. The Fed actually has two separate stress tests — the Comprehensive Capital Analysis and Review and the Dodd-Frank Act Stress Test. (Banks with assets of more than $10 billion are also subject to DFAST examination, but not CCAR.)
Both tests examine the performance of a bank's balance sheet under the conditions of three hypothetical stress scenarios over none consecutive quarters: one baseline, one adverse, and one severely adverse. DFAST works by running the banks' balance sheet through the scenarios using a standardized capital management plan, whereas CCAR uses the banks' own capital management plan. The Fed sees whether the banks' capital levels remain above the minimum capital requirements under all three scenarios, but banks cannot "fail" the DFAST test — the only penalty occurs if banks fail to meet minimum post-stress capital levels under CCAR.
CCAR contains both a qualitative and quantitative test; the Fed determines not only how low the banks' capital levels go under the scenarios, but how well the banks' internal risk management models perform as well. Banks can, and sometimes do, fail the qualitative assessment while maintaining minimum capital levels.
The report recommends that the Fed publicly disclose additional information related to the methodology for issuing its qualitative stress test determinations, including "the role of ratings and rankings". It also says the Fed should not object to a bank's capital plan on qualitative grounds unless it includes "additional information about the reasons for the determinations" and periodically issues additional information on what capital planning practices banks should follow in order to pass the qualitative test.
The Fed can also improve its model development process by applying the same principles its uses to improve individual models to the combined system of models that its uses to conduct its stress tests; improve communication to banks regarding the range and sources of uncertainty inherent in its models; and generally better examine model sensitivity and communicate tolerance levels for risks.
But the report also says the Fed could be even more imaginative in its development of the stress test scenarios upon which the program is based. The Fed's stress scenarios have been too limited and possibly too weak, the report said, and should examine a process to consider "levels of severity that may fall outside the U.S. postwar historical experience." The Fed has also limited the depth of its stress testing by only selecting a single severely adverse scenario per year, and should consider issuing multiple adverse scenarios in its annual examinations.
Alix said those recommendations, if anything, offer fuel to Wall Street critics who have been concerned that the stress testing program has been too lax.
"I don't think this is a road map for dismantling stress testing," Alix said. "They raise good questions about costs and benefits, but in some ways they suggest the Fed do more in envisioning different scenarios — that is kind of against the direction that was hoped for by the requesters of the report."
House Financial Service Committee Chairman Jeb Hensarling, R-Texas — who requested the report — praised it, emphasizing the faults the GAO found rather than the broad areas where the watchdog supported the central bank.
"The GAO report confirms the secrecy surrounding the stress tests makes it almost impossible to measure the effectiveness of the Fed's regulatory oversight or the integrity of the tests' findings," Hensarling said in a statement. "The changes recently proposed by the Federal Reserve to its stress testing process are inadequate given this report's devastating findings and recommendations."
The Fed's stress tests are among the most significant and effective supervisory tools to have emerged from the financial crisis, but have drawn frequent criticism from banks and some members of Congress. Banks have complained that the models that the Fed uses are too opaque and that the process for designing the stress scenarios lacks public input — some academics have even floated the idea of suing the Fed over the tests' possible violations of administrative law.
Fed Gov. Daniel Tarullo recently floated a blueprint for how to change the stress testing program to make it more fully integrated with the rest of the central bank's capital rules, particularly the additional capital surcharge on global systemically important banks. In the nearer term, the Fed issued a proposal to drop the qualitative assessment for all but the largest U.S. banks.