-
All 31 firms that took this year's Dodd-Frank Act Stress Test had enough capital to withstand the Fed's hypothetical severe economic scenario, but several of the largest banks were teetering on the edge of the leverage and risk-based capital requirements.
March 5 -
The Fed's annual dissection of each of the 31 largest U.S. banks is a painstaking and enormously complicated endeavor. This is an inside look at how it works.
February 27 -
Though stress tests are widely viewed as a successful and critical exercise, there are growing concerns that regulators and the banks themselves may have become too reliant on them, overshadowing other aspects of the supervisory process.
March 2
Add one more thing to the list of concerns following last week's Dodd-Frank stress tests.
The Federal Reserve projected much lower revenue for six major U.S. banks than the banks' own models. The calculation in question estimates pre-provision net revenue over a nine-quarter period following a severe economic shock. The gap that attracted substantial discussion among analysts was at JPMorgan Chase.
The Fed said JPMorgan's revenue would be $30.4 billion, or 38% less than JPMorgan's own estimate of $49.5 billion.
To be sure, differences between the Fed's numbers and company-run numbers are to be expected because of the structure of the test. The Fed gives the banks the scenarios, but otherwise it provides little guidance in an attempt to prevent the companies from managing to a number. Yet the large gap between the Fed's projected revenue number for JPMorgan and the company's own number is a negative in the eyes of analysts, who are still trying to figure out how bad a problem it is.
"What does this mean? Frankly, we don't know," said Mike Mayo, an analyst at CLSA. "It is certainly not good. I'm just not sure where on the not-good scale it goes."
The big question now one that will be answered in short order is if the variance could affect the company's results in the comprehensive capital analysis and review due out Wednesday. With the Fed projecting a smaller revenue base to absorb losses, would the company be able to return as much capital as expected? Could the gap between the Fed and JPMorgan's estimates cause the company to fail the CCAR on qualitative grounds?
"This does make me worry a little, because I think it makes you open to criticism if your model is significantly different from theirs," said David Konrad, an analyst at Macquarie Capital. "You don't want to be that far off from the Fed."
However, Konrad tempered his comments by adding that he is not suggesting that JPMorgan will likely fail the CCAR, just that "the wide variance creates some risk factors for the qualitative component" of the test.
Again, the stress-testing process is clandestine, so why various companies fail is largely speculative. Last year, for instance, Citigroup failed the qualitative component and Konrad said that the differences in its model versus the Fed's may have been a contributing factor.
Other analysts cautioned not to draw too many comparisons between the situations of Citi and JPMorgan.
"I think in the case of Citi, the two CCAR failures were penalties for a perceived weakness," said Christopher Whalen, senior managing director of Kroll Bond Rating Agency. "I think that is what the Fed was saying between the lines with those failures."
Whalen said the question of whether the variance matters highlights doubts about the value of the overall stress-testing process.
"The Fed's part is completely opaque," Whalen said. "The test is designed to be so subjective and speculative and it is driving the banks to terror. You have to tell us where these numbers come from. To me, this whole process is a circus that gets a lot of attention but has no analytical value."
Analysts tended to say that the gap between the two figures was likely a difference in the Fed's methodology, versus a massive overestimation by JPMorgan.
The Fed made few changes from 2014 to 2015 in its description of how it calculates pre-provision net revenue. Both descriptions said it includes operational risk events, mortgage repurchase expenses and other-real-estate-owned costs. Also, both are calculated by considering eight components of interest income, three components of noninterest income, trading revenue and three components of noninterest expense. The
Companies that rely on the market for a large amount of their revenue join JPMorgan in having a wide variance from the Fed. For instance, Morgan Stanley's projected revenue was 51% higher than the Fed's estimate, while Goldman Sachs' was 78% higher.
"The Fed modeled for more strenuous capital markets scenarios lower equity markets, greater volatility and higher corporate defaults," Mayo said. "That part we know."
Other analysts say that the Fed's modeling does not jibe with the way companies like JPMorgan and Goldman performed during the downturn.
"These companies went through a process like this not too long ago and there was a lot of capital raising and repositioning of their clients," said Marty Mosby, an analyst at Vining Sparks. "It is not as if their earnings fell off a cliff like the Fed is imagining."