WASHINGTON — The Federal Reserve Board's annual stress tests have become "predictable," as evidenced by "nearly perfectly correlated" projected stress losses in 2013 and 2014 for banks that participated in both tests, according to a paper published Tuesday by the Office of Financial Research.

The paper, authored by Paul Glasserman and Gowtham Tangirala of Columbia Business School, was written while Glasserman was serving as a consultant with OFR, the data research arm of the Financial Stability Oversight Council.

The paper found that as the Fed's Comprehensive Capital Analysis and Review and Dodd-Frank Act Stress Test processes become routine, their results have become predictable over time. That is in part because Fed models and bank balance sheets have stabilized and banks have an incentive not to take on investments that incur higher capital costs, the paper said.

"But whereas the results of stress tests may be predictable, the results of actual shocks to the financial system are not, and herein lies the concern," the paper said. "To the extent that stress test results become more predictable, they become less informative."

The Fed each year requires systemically important financial institutions — that is, banks with more than $50 billion in assets, known as SIFIs — to submit large quantities of data to the central bank to assess their ability to withstand a severe stress event.

As part of the tests, the Fed sets up hypothetical stress events of varying severity — a baseline, moderate and severely adverse scenario — to test banks' ability to weather different conditions. (The moderate test, for example, may in fact be just as challenging to a bank as the severely adverse scenario, depending on its balance sheet.)

Those stress tests have become an increasingly important part of the Fed's regulatory operations, the paper says, but that value may diminish if the process becomes more routine, and by extension less illuminating.

The authors examined the projected loss amounts from banks and by loan category for the 2012 and 2014 CCAR tests, the 2014 DFAST test and the 2009 Supervisory Capital Assessment Program — a sort of proto-stress test undertaken before the passage of Dodd-Frank. The authors examined the moderate and severe tests and looked for indications that loss rate distributions had stabilized over time. The authors found that the results had stabilized to a "striking" extent.

"We would expect to see a more complex relationship between adverse and severely adverse outcomes," the report said. "The patterns appear to be an artifact of the stress testing process rather than an accurate reflection of potential bank losses. They suggest an opportunity to get more information out of the stress tests through greater diversity in the scenarios used."

The paper's publication comes as the Fed is slated to release the results of its 2015 DFAST test on March 5 and the results of the 2015 CCAR test on March 11.

Some observers have also voiced concerns that the Fed's stress testing regime is becoming more routinized and less rigorous. Til Schuermann, a partner at Oliver Wyman and former senior vice president at the New York Fed, told an OFR conference in January that banks' growing emphasis on passing the tests rather than using them as an opportunity to shore up vulnerabilities is troubling.

"What I do fear is that there is a real risk that CCAR is going to devolve into a compliance exercise," Schuermann said. "The stakes are really high, so the focus is on the outcome — passing if you will — rather than the process."

But Jaret Seiberg, an analyst with Guggenheim Securities, said the danger could be that the Fed comes back in 2016 and makes the stress testing program less predictable.

"A tougher test would impact the biggest banks the most as they already face the market shock test," Seiberg said. "Making the tests less predictable and harder to pass would be consistent with our view that the policy environment favors the regional banks over the megabanks."

Subscribe Now

Access to authoritative analysis and perspective and our data-driven report series.

14-Day Free Trial

No credit card required. Complete access to articles, breaking news and industry data.