Big Banks Use a Footnote to Look Smaller
One thing that makes banking regulation hard is that it's impossible to know how big a bank is. There are lots of regulatory reasons why you might want to know that -- is it too big to fail? too big relative to its capital? -- but you just can never quite measure it. You get out your giant tape measure, and you try to wrap it around the bank, but the bank squirms and oozes and you can never quite see all of it at once.
This shimmering evanescent quality is, to the connoisseur, part of what makes banks so beautiful. But not everyone is a connoisseur. Some people find banks ugly, and the bigger, the uglier. But these people, too, run into the measurement problem: How can you know how mad to be at a bank, if you can't tell how big it is?
The Wall Street Journal had a fun story on Thursday about "footnote 151." This is at its heart a story about how to measure bank bigness. But since you get -- or get rid of -- what you measure, it is also a story about how regulation can make banks simultaneously less risky but also more creative.
But first, measurement. Back in the olden days, before the financial crisis, bank regulation was mostly a matter for connoisseurs. Bank regulators knew that it was impossible to measure banks, but they also knew that the doomed but noble struggle against that impossibility was enough to give a life meaning. They developed something called "risk-based capital regulation." The idea here, as the name implied, was to measure banks based on the riskiness of their assets. A bank that had $1 billion of corporate loans was, the regulators decided, twice as risky as a bank that had $1 billion of mortgage loans, and five times as risky as a bank that had $1 billion of mortgage-backed securities. And this risk-based measure became, for regulatory purposes, the main way of determining a bank's size, the denominator for its capital ratios: The corporate bank would have $1 billion of "risk-weighted assets," the mortgage bank $500 million, the mortgage-backed-securities bank $200 million. For regulatory purposes, the corporate bank was twice as big as the mortgage bank, and needed to have twice as much capital.
This is just the tip of the iceberg; you could devote your whole life to understanding the nuances of risk-based capital, and lots of people did.
But then the crisis happened, and it rather discredited risk-based capital. There are two reasons for this. First, risk-based capital regulation seems to have contributed to the crisis. If mortgages require more capital than mortgage-backed securities, and if capital is expensive for banks, then banks will want to turn their mortgages into mortgage-backed securities. So they did. You know the rest of the story. It turns out that regulators are not perfect at measuring risk. And by enshrining their risk measures in capital regulation, they created regulatory arbitrages: If you can make something look less risky for capital regulation, without actually making it less risky, you will be tempted to do so. And that will make the system more risky
Second, though, the crisis took bank regulation out of the hands of the connoisseurs and made it a matter of general public interest. Since 2008, bank regulation overall has been considerably less appreciative of the aesthetic qualities of big complex banks. A lot of people think those banks should be smaller and less complex, and are not so interested in devoting their lives to nuances of bank risk and measurement. And now those people have some influence over bank regulation. One upshot of that has been a focus on the leverage ratio, which was intended to be sort of like risk-based capital, but without the nuance. Instead of lovingly assigning an individual risk weight to every sort of asset that a bank might own, the leverage ratio tries to take a straight-ahead approach to measuring how big a bank is. Regulators just add up all the assets the bank has, without weighting, and then the bank has to have some minimum amount of capital as a percentage of those assets.
Critics of bank complexity like this approach because it seems simple. There is no need for regulators to estimate the risk of assets. The banks just have to have enough capital, relative to their size, to weather those risks. You don't need to know how risky the assets will be; all you need to know is how big the bank is.
But remember: It is impossible to know how big a bank is! Banks have assets that you can measure relatively easily, like bonds and loans. But they also have other things that are harder to measure, like derivatives and loan commitments. A $100 million interest-rate swap isn't quite the same thing as a $100 million bond, either in terms of its risks or in terms of its treatment under U.S. generally accepted accounting principles. And yet that $100 million interest-rate swap isn't nothing, either. It does feel like it should be measured somehow for capital and leverage purposes.
And so it is. The general idea of the leverage ratio appeals to non-connoisseurs, but the details of its implementation were left to the same sorts of people who built risk-based capital regulation, and they went and built a surprisingly elaborate system for measuring derivatives exposures for purposes of the simple leverage ratio. So, for instance, that $100 million interest-rate swap counts as $1.5 million of assets, if it has a maturity of more than five years, or $500,000, if it's one to five years, or nothing if it's less than a year. Why those numbers? I don't know. To re-introduce some mystery to bank capital regulation, maybe.
Footnote 151 is, as its name implies, a footnote to all of this. Here is what it says:
For a derivative contract that is structured such that on specified dates any outstanding exposure is settled and the terms are reset so that the market value of the contract is zero, the remaining maturity equals the time until the next reset date. For an interest rate derivative contract with a remaining maturity of greater than one year that meets these criteria, the minimum conversion factor is 0.005.
And here is what it means:
Instead of selling a 10-year swap to hedge oil or interest rates, for example, banks are now proposing to structure such deals as a series of swaps that settle every day for 10 years. The impact on the capital banks have to set aside is significant: A cleared 10-year interest-rate swap ties up three times as much equity as a one-day contract.
If you and I do a 10-year swap, and interest rates move in my favor, then you owe me money. In a sense you owe me the money in 10 years, when the swap expires, but the way things normally work, you post collateral to me pretty regularly, so I don't have to spend 10 years wondering if you're good for the money. Still, collateral isn't a perfect solution: There can be delays in collecting it, and if you default, we can have days of gamesmanship and years of litigation to figure out how much of the collateral I get to keep. If we replace the idea of "collateral" with the idea of "we close out the contract at fair value, and then immediately open a brand-new one at fair value," then some -- some -- of those issues go away. Now there's no question that I get to keep the collateral, because it isn't collateral; it's a payment that is mine for all purposes forever. And the amount of collateral, sitting in an account somewhere to secure the contract, can never get all that big: Once you pay me the money, I can use it, instead of keeping it in a collateral account and fretting about it. (On the other hand, if you default, we still need to figure out what the last settlement payment is, which can still lead to gamesmanship and litigation.)
More importantly, from a bank's perspective, recharacterizing the transaction this way lets it look smaller for leverage-ratio purposes, and thus require less capital. Not much has changed economically -- the bank still has the same 10-year exposure to interest rates -- but now it needs less capital. It is, dare I say it, a regulatory arbitrage. This guy dares:
“This is classic regulatory arbitrage,” said Marcus Stanley, policy director for public-interest group Americans for Financial Reform, which advocates for tougher financial regulations. “The risk is identical.”
And Thomas Hoenig of the Federal Deposit Insurance Corp. agrees: "If banks are shrinking their capital buffers without equally reducing risk, he said, 'I call that gaming the system.'"
Substantively, I suspect this is partly true and partly false. It is true in the sense that the interest-rate risk in the two contracts is identical. To the extent that the higher capital requirements for a 10-year swap than a one-day swap are based on the higher interest-rate risk in the 10-year swap, structuring that 10-year swap with daily settlement doesn't change the risk, and shouldn't get you a lower capital requirement. It is false, though, in the sense that the daily-settlement contract probably does avoid some unpleasant risks, like the risk of ballooning both of our balance sheets as the swap gets bigger, you post more collateral to me, and we become more and more intertwined financially. Settling up and reducing contracts to zero every night "will reduce risk across the financial system," says the associate general counsel of the Securities Industry and Financial Markets Association, and that is probably right.
But whatever you think of that substantive debate, the thing to remember is that the rule explicitly says that banks can do this. The banks are not exploiting a typo or a mistake: There is a specific footnote advising them to settle up their contracts frequently, so that's what they're doing. "Regulators, through rulemaking, have indicated a preferred legal method," says the Sifma lawyer, and "banks are conforming their practices." Presumably the regulators prefer frequent settlement because they think it makes the banking system safer, which is at least a plausible belief, and they've decided to encourage it by giving the banks a break on capital for doing it. This isn't a story of banks gaming a regulatory system. This is a story of regulators holding game night at their house and inviting the banks to come play.
But at a deeper level it is a story about how and why you measure banks. Even what is supposed to be the simple measurement of bank size, the leverage ratio, is not a simple or objective matter. It has a political and risk-management purpose, and so it is influenced by political and risk-management arguments. Regulators manipulate it to encourage the banks to be safer, and banks manipulate it to make themselves more profitable. The old concerns -- regulators trying to anticipate risks and build them into the measurement, bankers trying to arbitrage those measurements for profit -- continue in a new regime. The banks, their size and their risks remain elusive.
Also a story tailor-made to appeal to me, insofar as it is about (1) derivatives, (2) bank regulation and (3) footnotes. Obviously those three things go well together.
It is. There are counter-arguments, which I find unpersuasive, but even the counter-arguments admit that capital is expensive for bankers, who after all make the decisions about how much capital to have.
A good book telling this version of the story -- that is, the version where risk-based capital regulation is a major culprit -- is Friedman and Kraus's "Engineering the Financial Crisis."
That's Table 19, "Conversion Factor Matrix for OTC Derivative Contracts," in the capital and leverage ratio rules. There's a relatively (relatively!) straightforward explanation on slides 14 and 15 of this Davis Polk presentation. I am ignoring netting agreements, collateral, etc. Also when I talk about the "leverage ratio" I really mean the Dodd-Frank "supplementary leverage ratio."
Most importantly, I am ignoring mark-to-market. On Day One, if you and I enter into an interest-rate swap, it will have a value of around zero: If it's struck at fair value, it won't be an asset either to you or to me. As time goes by and interest rates move, it may be an asset to you, or to me, and that mark-to-market exposure will show up on our balance sheets and be included in the leverage ratio. But that is straightforward; of course on-balance-sheet assets count against leverage. The $1.5 million, or whatever, is different: It's an add-on to the regular assets to reflect possible future exposure.
In "foreign exchange and gold" derivatives, the difference is even bigger: A contract of more than five years gets a 7.5 percent weighting, while one of less than a year gets a 1 percent weighting, so a 10-year contract requires 7.5 times as much capital as a daily-settled version. In most other classes of derivatives, though, it's not as big a difference. In oil ("other"), the 10-year contract gets a 15 percent weighting, the daily 10 percent.
Leaving aside interest on the collateral, etc.
These capital requirements -- which, remember, are in addition to just requiring capital against the mark-to-market exposure amount of the swap -- can be thought of as sort of like initial margin on a derivative: The idea is to cover you against the risk that the position moves dramatically in one day and your counterparty will default without posting the final collateral payment (or, here, the final settlement payment). A 10-year swap has more interest-rate exposure than a one-day swap, and so should have a higher initial margin. Daily settlement doesn't change that calculation.
And then I guess being a bit petulant when they lose?
In a statement Thursday to The Wall Street Journal, the Office of the Comptroller of the Currency, the regulator of federally chartered U.S. banks, said the practice “raises serious safety and soundness concerns” and the agency “does not support actions that weaken bank capital in ways that could affect a regulated institution’s ability to withstand market disruptions or economic downturns.”
This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.
To contact the author of this story:
Matt Levine at email@example.com
To contact the editor responsible for this story:
James Greiff at firstname.lastname@example.org