Stress Test Keeps Getting Less Stressful for Big Banks

I find that once you make a spreadsheet of the stress tests it sort of takes over your life.

So much congrats, Mike Corbat!

Photographer: Simon Dawson/Bloomberg

For some reason I can't resist doing my own numerical guesswork about the Federal Reserve's bank stress tests and inflicting it on you. So here is my rough guess of:

  1. How much money each of the usual big six banks initially asked to return to shareholders,
  2. How much they asked to return after seeing last week's stress test results, and
  3. How much money the Fed would have let them return.

You can tell that this is pretty unscientific because some of my guesses of how much room the banks had to return capital -- number 3 above -- are off by billions of dollars from what they were last week. That is because this is more of an arithmetic game than a serious scientific estimate of any of those numbers. But it at least roughly suggests what is going on here. The point is: The gray bar (the bank's final capital-return request) has to be lower than the black bar (the Fed's calculation of excess capital), and if the blue bar (the bank's initial request) is above the black bar then it has to shrink to become the gray bar. Methodology and caveats in the usual place. 1

So what is there to say? First, yesterday I predicted that Goldman Sachs would pass the stress test by revising its capital-return request to some gestural minimum. 2 That's certainly what the chart looks like! To be fair, that's because I've only incorporated Goldman's five-cent dividend increase and not any share repurchases, of which there might be a few. Goldman's announced revised plan "includes the repurchase of outstanding common stock, an increase to its quarterly common stock dividend and the possible issuance and redemption of other capital securities," but unlike the more swaggery banks today, Goldman did not announce an amount of stock buybacks, and I suspect they will be quite small. There is just not a lot of room there.

Second, is there a stigma about revising and resubmitting? No, right? (The revise-and-resubmit banks' stock prices don't seem to have suffered.) Should there be? I guess I've expressed my view that the proper aim in taking the stress tests is to get as close as possible to the minimum capital without going under. From that chart, the revise-and-resubmitters seem to have done a pretty good job of achieving that goal. The banks that got it on the first try were very conservative about their capital return plans. One of them is Wells Fargo, which just seems to like being conservatively capitalized. The other two, Citigroup and Bank of America, had good reasons to be conservative: Citi's senior managers, after failing last year's stress tests, had already boxed up their belongings just in case they failed again. And Bank of America's qualitative capital-planning deficiencies earned it only a pass-with-an-asterisk this year, even without an aggressive capital-return request. Those two could not afford to get close to the line, and Wells Fargo didn't want to, but the healthy red-blooded big banks took full advantage of their two chances to squeeze as much capital return as possible out of this process. Which was the point of the process.

Of course, you don't have to agree with my premise that the proper aim in taking the stress tests is to end up just above the minimum. My Bloomberg View colleague Mark Whitehouse, for instance, disagrees:

The banks' brinkmanship illustrates a problem with the stress tests and with regulation in general: Instead of thinking about how much capital they need, executives are focused on how little they can get away with. This is unfortunate, because capital -- also known as equity -- should not be seen as a burden.

One possibility to consider is that the banks are focused on having as little capital as possible under the Fed's stressed calculations. But under their own stressed calculations, taking into account their own plans for revenue growth and asset planning, they wouldn't be all that thinly capitalized, and there's an argument that their calculations better reflect reality 3 than the Fed's do. If the banks' own plans show that they would still be very comfortably capitalized even in a second financial crisis, you can understand why they might think they can spare some cash to give back to shareholders.

But that aside. It seems clear enough to me that the point of capital regulation is to require banks to have X amount of capital, and that banks don't want to have any more than X, because if they did then you wouldn't need the regulation to tell them to. (Just as the point of tax law is to require people to pay Y percent of their income in taxes, and no one wants to pay any more than Y. 4 ) It also seems clear enough to me that the stress tests are simply another form of capital regulation. Therefore, the expected, and to some extent "correct," response to stress tests is to want to hit the minimum capital requirement as precisely as possible. If regulators want banks to have more capital than the minimum, they should make the minimum higher. 5 If they don't -- and they mostly don't -- then you can't really blame the banks for wanting to have the minimum.

This column does not necessarily reflect the opinion of Bloomberg View's editorial board or Bloomberg LP, its owners and investors.

  1. Methodology:

    1. We start with the actual announced capital return numbers, rounded up in this Bloomberg News article: Citi has announced up to $7.8 billion of stock buybacks, Morgan Stanley $3.1 billion, JPMorgan $6.4 billion, and Bank of America $4 billion; I've assumed that Goldman and Wells Fargo are doing zero. All the banks but Bank of America increased their dividends; I've incorporated five quarters of dividend increases into my chart, because that seems to be roughly the covered period. (I've assumed current share counts, ignoring the fact that buybacks will reduce those.) 
    2. I've ignored non-common-stock capital actions in these counts, which might be wrong; some of the banks seem to be interested in doing stuff with non-common capital securities.
    3. I've used those announced capital actions as the final ask.
    4. To the extent the final ask and the initial ask differed, I've calculated (as per last week's methodology, and using the numbers in the Fed's stress test release) the difference in implied capital amounts between the initial and revised asks, under the various capital regimes, and then used the maximum of those differences as the difference between the initial and final asks.
    5. The initial asks in the chart are therefore calculated by just adding the initial ask and that difference.
    6. Note that the delta between Morgan Stanley's initial and final ask seems to be explained by its dropping its request to redeem some preferred stock. I don't know what Goldman and JPMorgan revised.
    7. The room that the banks had to return capital is calculated by looking at the stress test results and estimating how much room the banks had (after their planned capital actions) above the minimums for the various capital ratios. I then took the minimum of those estimates -- that is, the closest they came to the minimum -- and added that to the revised ask, to get the total amount of room they had.

    This methodology starts with actual announced capital actions, rather than my estimates of room under the stress tests, and should (I hope) therefore produce guesses that are more accurate than last week's. But last week's caveats still apply; there are many avenues for error in these calculations.

  2. Disclosure, I used to work there. 

  3. I mean. Expected hypothetical stressed reality. That's a bit oxymoronic. Everyone has a revenue growth plan until they get punched in the face.

  4. Is capital a tax? Lots of people say no, but those people do not work at banks.

  5. As Whitehouse advocates.

To contact the author on this story:
Matt Levine at

To contact the editor on this story:
Zara Kessler at

Before it's here, it's on the Bloomberg Terminal.