The annual Dodd-Frank Act Stress Test results for the nation's largest institutions have been released. This financial services regulatory version of the annual March collegiate basketball tournament largely provided an expected clean bill of health for the banking industry. It will support increased bank shareholder distributions based on the belief that banks have finally recovered from the financial crisis.
The tests, however, are deeply flawed, and the recovery belief is premature. Worse than ineffectual, the results are likely to create a false sense of security leading to overconfidence and unwise capital structure and dividend decisions. Just because things have gotten better does not mean they are good.
The Federal Reserve has not disclosed the details of their model. Thus, it is difficult to determine if they ran the appropriate tests and interrupted the results correctly. Nonetheless, it is still possible to offer some high-level concerns based on the general nature of scenario-based stress tests.
Stress tests are primarily backward-looking. Consequently, rare, yet-to-happen events are absent from them. As such, they underestimate what can, in fact, occur. This subjectivity in scenario selection suggests the tests are more about measuring regulatory risk beliefs not risk itself.
Next, the tests fail to include possible changes in risk behavior. Banks respond to earnings pressure by increasing their risk exposure. This is reflected in concentrations in higher yielding, easy-to-originate assets with a recent history of low losses. Previously, these assets included mortgages and commercial real estate prior to the last crisis. Currently, they could include the new fashionable asset class of commercial and industrial loans. Regulators will be unable to detect the increased exposures in the new asset classes of the banks who take them until it is too late.
Finally, the tests ignore the network impact of a large bank failure on the banking system, which is more than the sum of its member banks. The herding effect of just one troubled megabank can trigger a systematic crisis in ways not fully appreciated. For example, as Lehman Brothers demonstrated, the sudden decline of a major institution can cause investors to liquidate positions and increase their demand for cash, which, in turn, can create a system-wide bank run. The net result is the tests seriously underestimate the banking industry's collective risks.
Furthermore, their false precision suggests safety when there is none. Some may argue the tests, while not perfect, are better than no tests at all. No test, however, is better than a flawed test with misleading results. Banks will use their passing test results to take on more risk based on regulatory-induced overconfidence. Furthermore, despite the current undisclosed model details, banks will begin to discern the key model drivers from their results. Then, they will focus more on managing to the model than on managing risk.
Rather, than trying to predict the impact of possible so-called worst case scenarios on banks, regulators should focus on structural reforms that ensure the system can survive the next crisis. Instead of predicting the timing and severity of the next banking tsunami, we should concentrate on building seawalls to withstand them. This process means focusing on the consequences of institutional failures not trying to predict them, and involves two steps.
First, create banks strong enough to withstand a financial storm. Equally important, protect against network risk by ensuring the system can withstand a Black Swan event. The real enemy of banking stability is size. "Too big to fail" institutions represent unacceptable risk concentrations in our economy. The focus should be on making sure no one institution is so large that its failure would endanger the entire system.
