I bet Michael Corbat looks like this a lot. Photographer: Andrew Harrer/Bloomberg
I bet Michael Corbat looks like this a lot. Photographer: Andrew Harrer/Bloomberg

If you are a bank, and you have no idea what you're doing, what should we do with you? The Fed has an answer:

Practices with specific deficiencies included Citigroup's ability to project revenue and losses under a stressful scenario for material parts of the firm's global operations,1 and its ability to develop scenarios for its internal stress testing that adequately reflect and stress its full range of business activities and exposures. Taken in isolation, each of the deficiencies would not have been deemed critical enough to warrant an objection, but, when viewed together, they raise sufficient concerns regarding the overall reliability of Citigroup's capital planning process to warrant an objection to the capital plan and require resubmission.

The Fed's answer is: If you are a bank (specifically, Citi), and you have no idea what you're doing, we'll make you hold on to your shareholders' money, because it's better for you to have that money than for your shareholders to have it.

This answer is not immediately intuitive! One might very intuitively say, if you are a bank, and you have no idea what you're doing, maybe your money would be a bit safer in the hands of your shareholders than in your hands. But, nope.

The Fed's answer comes from the Comprehensive Capital Analysis and Review, released yesterday afternoon. I said last week, when the Dodd-Frank Act Stress Test came out, that the DFAST wasn't where the action is, that the place to be is the CCAR. But that isn't quite right; really the DFAST and the CCAR are sort of self-reinforcing oddities. The point of the whole stress test process is to decide:

  1. If things get bad, do the banks have enough capital?
  2. If so, can they maybe get rid of some of that capital? If so, how much?

The DFAST answers the first question, and the CCAR answers the second question, but they're the same question. There is an amount of capital that is "enough" and it is 4 percent of your assets.2 If, in a stressed scenario, you would have $100 billion of assets and $4 billion of equity, then you're okay. If you'd have $5 billion of equity, then you can hand back $1 billion of cash to shareholders and still be fine.

But that simple arithmetic is conducted in a fog: When you make your CCAR submission, you don't know your DFAST results. You know how much capital you want to return, but you don't know how much stressed capital you have. You know how much capital you think you'll have, but that's not how much the Fed thinks you'll have. So you have to imagine the Fed's number, see how much it's above 4 percent, and then ask to pay back the difference.

This is sort of weird and annoying; why don't you just tell me the name of the movie you'd like to see? But it's intentionally annoying. The goal of the banks, in these stress tests, is to return as much capital as they can, but that is obviously not the goal of the Fed. The goals of the Fed are subtler and more multifaceted: It wants banks not to return too much capital, but it also wants them to be sober and serious about the process. So the banks are rewarded for doing their math right, but they're also rewarded for being conservative and careful. Nothing encourages conservatism like fog.3

The stress test is not just a test of capital; it's a test of morality. And that's the test that Citi failed. The Fed says it "objected on qualitative grounds." Citi thinks that it would have enough capital in a crisis, even after raising its dividend and doing a $6.4 billion stock buyback. The Fed also thinks that; its numbers show Citi passing the stress test with a 5.6 percent leverage ratio and 6.5 percent Tier 1 common ratio after taking into account its capital plan. But the Fed worries that Citi's thought process to get to that result was wrong, even though the result was right, or at least right enough.

So the Fed, which is confident that it can predict how banks will perform in a crisis,4 is less confident that Citigroup can predict how it will perform in a crisis. And the remedy is: Citigroup needs to hold on to shareholder money.

What is the mechanism by which that remedy makes Citi better? One obvious answer is, capital solves a lot of problems; a bumbling clueless highly capitalized bank will do less damage than a bumbling clueless minimally capitalized bank. This is true as far as it goes, but in this case it doesn't go very far at all; the Fed calculates that Citi's stressed Tier 1 leverage ratio would be 5.6 percent if it raised its dividend and bought back stock, and 5.7 percent if it didn't.5 That's not a much bigger cushion.

A more plausible mechanism is:

  • Shareholders looooove capital return:6 Citi was down 4.3 percent as of 10:30 this morning, while Bank of America, which does get to give back money, was up 0.9 percent.
  • Shareholders are more or less the boss of banks.
  • When the Fed tells shareholders they can't get money back, the shareholders won't get mad at the Fed; they'll get mad at Citi.
  • When the Fed tells shareholders "you would get money back, except that your bank can't figure out how much money it'll lose in a crisis," the shareholders will go to the bank's managers and say, hey, you know, get on that.
  • The managers will listen to the shareholders and return, chastened, to their lairs to figure out ways to project their losses in a crisis.
  • It's something, right?

There are a lot of steps along this road, and an obvious alternative is for the Fed to disclose the cluelessness directly. Instead of that suggestive but short paragraph I quoted at the top of this post, the Fed could go ahead and tell shareholders what it knows about Citi's planning process, and where that process is deficient. "Hey, shareholders, you might want to look into that whole Mexico situation," the Fed might quite reasonably say. The Fed could motivate shareholders to remedy Citi's deficiencies directly, rather than by the symbolic and sort of misplaced punishment of leaving their money in Citi's accident-prone hands.

I guess the worry there is that shareholders might not care: Accident-prone banks are not necessarily great for shareholders, but aggressive and heedless banks might be. Sober reflection on where you might lose money in a crisis doesn't fit well with aggressive expansion into risky and volatile businesses, and risky and volatile businesses are good for bank shareholders.7

The Fed's job is to protect banks' creditors, and taxpayers, from bank failures. Using shareholders as a lever to do that job has its problems, because shareholders are looking out for themselves, and their interests are not the same as, like, society's. So the Fed has to translate its complaints about bank management into language that shareholders understand. Like taking away their money.

1 Ooh ooh can you guess what "material parts of the firm's global operations" betray a weak "ability to project revenue and losses"? I have a theory.

2 I mean, whatever, it's a whole matrix of numbers that you can find on the bottom of page 13 of the CCAR, but the most binding ones are the 5 percent Tier 1 common equity ratio and the 4 percent Tier 1 leverage ratio.

3 This is somewhat undermined by the fact that the Fed gives you a chance to revise and resubmit your capital plan if you ask to give too much money back. Bank of America and Goldman had to resubmit and I would think that, if you're a shareholder, and your bank didn't have to resubmit, you might be a bit peeved. It means it wasn't aggressive enough.

4 Right? Implicit in the idea of comparing the banks' stress tests to the regulatory ones, and then penalizing the banks if theirs differ from the regulatory ones, is the notion that the regulators' predictions are right and the banks' are wrong. Isn't that weird?

5 Compare page 13 of the CCAR with page 96 of the DFAST. Other metrics are further apart; e.g. the minimum Tier 1 common ratio is 6.5 in the CCAR and 7.2 percent in the DFAST, for a 70 basis point difference. I can't really explain why those metrics are so different.

6 One loose way to put this is: Citi's tangible book value per share is $55.31, and Citi's stock closed at $50.16 yesterday (and is down further today), so a dollar in Citi's hands is worth only like 90 cents. So every dollar that you free from Citi's clutches is worth an extra 10 cents to shareholders.

7 I mean, that is totally debatable in practice, but the theory is straightforward. A bank is a pile of assets mostly financed with debt. The equity of the bank -- which is just the top 4 or 5 or 6 percent of its financing -- functions as a call option on the value of those assets. Call options are worth more if the underlying assets are more volatile.

To contact the writer of this article: Matt Levine at mlevine51@bloomberg.net.

To contact the editor responsible for this article: Tobin Harshaw at tharshaw@bloomberg.net.