WASHINGTON — Moody’s Analytics is claiming it has developed its own stress test based on publicly available Federal Deposit Insurance Corp. data that more accurately examines bank performance in stress conditions than models overseen by the Federal Reserve.
The research firm is set to release a report Tuesday that it says previews the results of the Fed’s 2017 Comprehensive Capital Analysis and Review stress test results — which are to be published Wednesday — by delivering the findings of its own stress test of 16 noncomplex superregional firms based on quarterly call report data published by the FDIC.
Tony Hughes, a managing director of Moody’s Analytics and one of the authors of the report, said the firm’s stress test avoids the conservative bias that the Federal Reserve uses in its own models and instead emphasizes accuracy.
“It’s much more important that the models be built strictly for accuracy and that any conservative overlay is added later,” Hughes said. “From a regulatory standpoint, you can understand why the Fed wants to take a conservative view of the banks’ capital plans. But the Fed also wants the stress test to be woven into the fabric of the management of the bank. At that point ... it becomes hard to make truly informed business decisions using something that was built only for conservatism.”
The Fed’s stress testing program has been widely viewed — for better or for worse — as among the most consequential supervisory exercise to emerge from the financial crisis. CCAR, along with the Dodd-Frank Act Stress Test, examine the performance of balance sheets of all banks with more than $50 billion in assets under a hypothetical stress scenario. Banks who fall below the CCAR minimum capital levels, or who are deemed to have qualitative shortcomings in their models, can have limitations placed on their dividend payments.
Research has been conducted before to determine how banks and other firms perform under hypothetical stress. Stress tests themselves have been employed by banks themselves for years, and some academics, notably the New York University Stern School of Business, have undertaken efforts to develop independent examinations of systemic risk.
The Moody’s model builds on earlier research that developed a method for parsing out bank-specific data based on both aggregate industrywide data and information about an individual bank’s market share and portfolio makeup available in quarterly call reports to the FDIC. The firm then takes that information and runs an approximated balance sheet through a future stress cycle, much the way the Fed’s stress tests do.
To be sure, there are important differences. The Fed runs its tests using actual details of each bank’s portfolio, rather than educated guesses based on call reports. And CCAR, at least, also factors in the bank’s own capital management plan rather than an assumed plan, which makes the Moody’s test more closely resemble the Dodd-Frank Act Stress Test.
Different banks have alternate business models and management methodologies, Hughes said, making a direct head-to-head comparison between banks that are not in direct competition more challenging. In Tuesday’s report, Moody’s examined the 16 superregionals because they represent the largest peer group among the 34 CCAR firms. The results show how the model gives banks a better sense of how they might perform among their peers.
“We just wanted to build a tool that would allow … any bank to compare itself to any other bank without worrying about methodological consistency as an issue,” Hughes said.
The report found that the banks it examined performed well and would experience chargeoffs that are far less severe under the Fed’s scenarios than were experienced during the 2008 recession. But some metrics are easier to reverse-engineer than others — bank-specific post-stress capital levels are “among the hardest to forecast because they rely on many inputs and assumptions,” the report said.
A Moody’s spokesperson said the firm’s stress test can calculate firm-specific post-stress capital ratios, but would keep those results confidential to the banks themselves. It did release aggregate data, however, that estimates post-stress Tier 1 capital minimums of around 11.1% and losses of $21 billion.
Hughes said the other advantage of the test is that it can be applied to any bank of any size — or even potentially applied to nonbanks or even nonfinancial companies, all of which might want to know how they would perform in a severe recession.
“We try to take the best publicly available view into all of the banks in the system, and then we try to model that data as effectively as we can with a view to creating a forecast and a stress test of every bank in that database,” Hughes said. “This is the first attempt to conduct a complete stress test in a manner that is similar to the way the banks are going about it on their own.”