BankThink: The Stress Tests Themselves Are Untested

By Karen Shaw Petrou

The Federal Reserve Board’s stress test for the largest bank holding companies has been compared to a treadmill, on which patients are told to exert themselves so doctors can see how long they’ll last. But, is the stress test in fact a financial analog to proven medical science so that we can indeed judge just how healthy big bank holding companies are after the Fed gets done with them?

Or is it, as I fear, a case of an experiment taken too soon out of the laboratory? Just as Dr. Frankenstein sought to restore life when he hooked up his cadaver, the FRB is engaging in a well-intentioned effort that is, sadly, still false science based on complex formulas unproven by the rigorous validation true science requires.

That a cure to unwise capital distributions is needed is not the question before the FRB. In the run-up to the financial crisis, regulators erred on the shareholders’ side, with the most obvious grave mistake the decision by Freddie Mac’s regulator to let the GSE repurchase shares even as its capital was already flat-lining. The FRB hasn’t gotten the public opprobrium FHFA’s predecessor warranted, but it took a lot of heat behind closed doors from international regulators after Goldman Sachs kept the dividends flowing in the fourth quarter of 2008.

But, that correction to prior policy is needed doesn’t make right the cure chosen by the FRB. In fact, its faith in its stress-test procedures — over-complex ones, as even the FRB found out – and the undue focus only on capital adequacy may well set the Fed hunting for every little basis point of capital even as new risks with serious systemic potential are overlooked.

The Fed started the stress-test process in 2009 to judge whether or not systemic bank holding companies could survive the still-roiling crisis. The test then was first of its kind and best in class, helping both to calm financial markets and force recalcitrant bank holding companies to recapitalize. Based on this salutary experience, it continued with similar stress exercises that prevented complacency in the U.S. However, the FRB didn’t rest on these deserved laurels. Instead, it jumped to a far different stress-test standard and launched the 2012 Comprehensive Capital Analysis & Review.

CCAR began with a final rule last November, a rule that outlined the FRB’s approach and promised both detailed data templates and, thereafter, public release of the CCAR methodology. Bank holding companies with assets over $50 billion got the whopping data templates shortly thereafter and began figuring out which data met what Fed criteria, also running the numbers through stress scenarios detailed in the rule. The methodology wasn’t disclosed until March 12, just a day before the Fed was forced to release CCAR’s results. Then, on March 16, the FRB released several substantive corrections.

Unfortunately, these did not include a fix to one basic quantitative flaw in the test methodology. CCAR did not reflect the fact that planned capital distributions in the “baseline” scenario can be reduced or reversed under stress. Telling a bank holding company in Q1 that it can’t pay dividends or repurchase stock because the complex, untested stress analytics show a few basis points of capital vulnerability in Q8 is to put far too much faith in a still-nascent stress-test process.

In CCAR, the FRB tried a bit of basic science, but then built a complex edifice of untested variables evaluated with undue precision to make binding decisions about which bank holding companies are healthy enough to give shareholders a bit of their due. That tough stress tests are right doesn’t mean that the FRB’s stringent ones are also correct. They could be, although the quick correction the FRB needed to make to its announced results suggests that even it hasn’t fully got a handle on its complex methodology.

Although the CCAR final rule makes clear that the FRB will use qualitative judgments about the robustness of the capital-planning process in concert with its quantitative conclusions, the results seem premised solely on a few basis points here or there in the sometimes-unreliable stress-test equations. The FRB should go behind the numbers and evaluate bank holding companies from both a quantitative and qualitative perspective. A good place to start is with issuing guidelines on how in fact to do robust capital planning, focusing not on all the technical niceties in CCAR, but on fundamental board and senior-management governance procedures that can be readily implemented and, then, objectively judged by examiners and investors.

Too much faith in these stress tests could lead to an all-clear conclusion from the tests even as new risks are building. The current FRB methodology may make a BHC seem safe but, in fact, could undermine market capitalization in risky ways or, worse, push a BHC to go even deeper into “risk-free” sovereign assets.This will score better on the stress test, but will have done little for either its real risk or the economic recovery.

Because stress tests are new and untested, over-reliance on them leads to the second and, in some ways, worse problem: the FRB will be so transfixed by its fancy models that it misses all the other risks to which bank holding companies are heir. Many of these — the interest risk resulting from big holdings of sovereigns, for example — weren’t even in the stress test. Bank holding companies exposed to these risks should also husband capital, as well as build better risk-management and risk-mitigation capability, but the FRB may well be slow telling them so if all it watches are its own fancy capital meters.