Friday, March 6, 2015

Are Bank Stress Test Results Too Predictable? OFC Publishes New Report

Yesterday the Fed released the results of the most recent stress test on banks. The result was fairly predictable -- they all passed. This shouldn't be a surprise according to the paper Are the Federal Reserve’s Stress Test Results Predictable? by Paul Glasserman and Gowtham Tangirala. Published by the U.S. Office of Financial Research the theory is a bold one with stark implications. Namely, that stress tests cease being a real deterrent to crisis if they are predictable because actual market crisis is anything but. As a result, the authors suggest that the scenarios used in future stress tests should be more diverse and numerous.

The current stress tests, as a function of regular economic policy, were created by the Dodd-Frank Wall Street Reform and Consumer Protection Act. The Comprehensive Capital Analysis and Review (CCAR) and the Dodd-Frank Act Stress Testing (DFAST) program are the two stress test programs in use. Both are designed to complement each other in application, but the process had become cumbersome for banks in both implementation and reporting which may have led to the use of the same scenarios every year to avoid creating any additional burden. According to the paper,
The annual execution of the CCAR/DFAST process has become an enormous undertaking for the banks covered by these programs and their supervisors. Despite the complexity of this process, using results made public across various stress tests we find that projected losses by bank and loan category are fairly predictable and are becoming increasingly so. In particular, losses for CCAR 2013 and 2014 are nearly perfectly correlated for banks that participated both years. 
In other words as the process has matured, stress tests have become more routine. "But whereas the results of stress tests may be predictable," the paper goes on to say, "the results of actual shocks to the financial system are not, and herein lies the concern."

The answer to the problem, as the authors point out, is easy -- greater diversity in the scenarios evaluated in a stress test. Providing the same two or three scenarios to every bank over and over again is like announcing a fire drill at 5pm every other Monday. It may help people to understand where to go in case of a fire, but it doesn't help to put the fire out any faster. Doing so requires a shift in focus to fire prevention and containment. It requires everyone at the bank to become "fire-fighters".  It is the ability to perform well under different scenarios that measures how well the bank is prepared for the unpredictable.

Only six Bank Holding  Company's undergo a stress test related to trading and counterparty risk, they are JPMorgan Chase (JPM), Citigroup (C), Bank of America (BAC), Wells Fargo (WFC), Goldman Sachs (GS), and Morgan Stanley (MS). These banks are also considered "Too Big To Fail" as can be read about in an article published yesterday on SeekingAlpha by TheGAFI entitled, What Investing In Too Big To Fail Means For Banking Investors. The paper shows a high correlation in projected losses from one year to the next. "This is more surprising than the corresponding results for the loan categories," says the paper, "because trading portfolios change much more quickly than loan portfolios, and trading losses should be more difficult to forecast than loan losses." In other words, while the predictability of stress test results for loan products is concerning, the predictability of stress test results for trading is even more troubling as it may point to some level of manipulation.

In a seemingly unrelated matter, and perhaps of more importance to investors, is that stress tests result in no reaction from the stock market. Here's what the paper had to say about this:
We then examine the stock market reaction to announcements of stress test results; consistent with the predictability of the results, we find no significant correlation between the severity of a bank’s reported stress losses and the change in its stock price relative to the market.
Conspiracy theorists might say this is the reason we don't have "real" stress tests; that is, a negative result could have a "real" negative impact on the market.

Bottom-line: Do stress tests as they stand today have value, yes. They allow for a thorough assessment of risks in terms of financial stability and put the human resources in place to help navigate through a financial crisis, but there are glaring holes in the logic of using the same two or three scenarios with similar variables for every bank and in every test. A dynamic set of stress scenarios is best to test the financial viability of banks in an unpredictable and complex market economy.