Daniel Tarullo ponders leverage. Photographer: Andrew Harrer/Bloomberg
Daniel Tarullo ponders leverage. Photographer: Andrew Harrer/Bloomberg

Yesterday U.S. banking regulators adopted new rules on the leverage ratio, and I briefly considered trying to add value by reading them and telling you what they said, but fortunately for me cooler heads prevailed. Other, better people, such as Peter Eavis and the good folks at Davis Polk, read the rules and explained what they say, so if that is of interest to you I recommend clicking those links. Or you could read the rules yourself, stranger things have happened.

Capital Requirements

But why read the rules, they're so simple: Big banks will be required to have Tier 1 capital equal to at least 6 percent of their assets. All their assets, with no messing around. Here is Eavis:

“The benefit of the leverage ratio is that it’s a simple tool, but the problem with the leverage ratio is that it’s a simple tool,” said Oliver I. Ireland, a partner at the law firm Morrison & Foerster.

Under other regulatory calculations, banks get to set capital levels based on the perceived risk of their assets. The process of assessing risk, however, can be complex, subjective — and vulnerable to abuse by banks that want to try to increase profits by holding less capital.

In contrast, the new rule demands that the same amount of capital be held against all assets, regardless of their perceived risk.

Yes, avoid complexity and subjectivity, good idea. Let's do some math to show how easy the leverage ratio is. Morgan Stanley, for instance, had $61 billion of Tier 1 capital as of the end of 2013, and $832.7 billion of assets.1 Divide that first number by that second number and you get 7.3 percent. Morgan Stanley has a leverage ratio of 7.3 percent. Wait:

Chubak also estimated that Morgan Stanley would have the most room to make up, with an estimated leverage ratio under the proposed Basel rules of 4%.

Four percent is not 7.3 percent. It is rather less.2

Let's start over. Banks are required to have Tier 1 capital equal to at least 6 percent of their assets, but what does that mean? Like, you go measure a bank's assets. Take all the stuff a bank has, dump it on a big table, sort it out, and see how much there is. It is ... hard, no? The assets of a bank cannot be comprehended by a mortal mind. I told you a few paragraphs ago that Morgan Stanley had $832.7 billion of assets, but I was lying. That's just what Morgan Stanley says in its certified audited financial statements filed with the Securities and Exchange Commission. That's not actually the size of its assets. What would "the size of its assets" even mean?

Thinking about the actual size of a bank's assets will lead to madness. There's nothing actual about a bank. A bank is a pure creature of accounting; how much of it there is depends entirely on what you are trying to measure. One form of measurement is U.S. generally accepted accounting principles, which have their purposes, one of which is to be included in publicly released financial statements. Under U.S. GAAP, Morgan Stanley has $832.7 billion of assets, and so Morgan Stanley tells you that that's how many assets it has.

Another form of measurement is the thing used to calculate the leverage ratio. This manages to find rather more assets at Morgan Stanley. Where does it find them? Oh, places. Generally, the new rule counts "total leverage exposure" to include:

  • GAAP assets;
  • potential future exposures for derivatives;
  • other adjustments for derivatives to back out some cash collateral netting;
  • lending commitments;
  • sold credit protection;
  • "repo-style" transactions;
  • "all other off-balance sheet exposures"; and
  • probably some other things too?

So you just dump all those things on the table, without risk-weighting them, and ... wait no hang on that's not true either. The claim that assets are not risk-weighted for the leverage ratio turns out to be totally untrue. GAAP assets are, for these purposes, assigned a risk weighting of 100 percent, whether they're Treasuries or junk bonds. But everything else gets its own specialized weight. Lending commitments are risk-weighted at 10 percent of the amount of the commitment.3 Derivatives are weighted at 100 percent of notional, for sold credit derivatives,4 or 10 percent, for bought high-yield credit derivatives or 5+ year equity derivatives, or 8 percent, for 1-5 year equity derivatives or 5+ year precious metal derivatives (except gold, which is risk-weighted at 7.5 percent), or 6 percent, for 1-year-or-less equity derivatives, or 5 percent, for bought investment-grade credit derivatives or 1-5 year foreign exchange derivatives, or 1.5 percent, for 5+ year interest rate derivatives, or ... you get the idea.5 Or do you? Is there an idea?

I mean, obviously there's an idea. The idea is that some things are more worrisome to regulators than other things, or at least look more like they should be regulated. If you write a billion dollars of 5-year credit default swaps on Microsoft bonds, that counts as a billion dollars of "total leverage exposure," because credit default swaps are scary. If you commit a billion dollars to a 5-year credit facility for Microsoft, that counts as $100 million of exposure, because that's 90 percent less worrisome than CDS. If you write a billion-dollar 5-year total return swap on Microsoft stock, that counts as $80 million of "total leverage exposure," because that's 92 percent less worrisome than the CDS. Obviously Microsoft stock is riskier than Microsoft bonds but that is how it is. CDS caused a crisis; equity swaps did not.6

Now it is a little annoying that the assets that regulators use to compute the leverage ratio don't have much to do with the assets that banks report in their disclosures to investors. As a matter of transparency and consistency, you might want those things to be harmonized, either by changing the leverage ratio rules to require capital equal to X percent of publicly reported GAAP assets, or by changing financial reporting requirements (or GAAP) to require banks to report assets that match up with their leverage-ratio calculations.7 If the best way to measure a bank's size includes counting 6 percent to 10 percent of its equity derivative notional, then why shouldn't the bank's financial statements reflect that?

But what I like about the leverage ratio is its pedagogical function: It is a constant troubling reminder that there is no "best way" to measure a bank's size, or its riskiness, or anything else. A bank is an unknowable collection of quantities that might, if you're lucky, snap into some sort of coherence when you look at it in certain ways. But you'll only get coherence if you decide what you want to measure, measure that, and ignore the teeming complexities in all the stuff you're not measuring. And it's good to keep in mind that other people are measuring other things for other purposes. Otherwise you might start to think that you're measuring a real thing.

1 See pages 137 and 244 of the 10-K.

2 Morgan Stanley is not alone: The rules announced yesterday include a final rule fixing the leverage ratio requirement at 5 percent (for big bank holding companies) and 6 percent (for their bank subsidiaries), but also a proposed rule adjusting how "total leverage exposure" is calculated. Under the final rule, "CCAR data suggests that covered BHCs that would not have met a 5 percent supplementary leverage ratio would have needed to increase their tier 1 capital by about $22 billion to meet that ratio." Under the proposed rule, "these institutions would need to raise in the aggregate over $46 billion in tier 1 capital to exceed a 5 percent supplementary leverage ratio under the proposed definition of total leverage exposure, over and above the amount they would need to raise if the definition of total leverage exposure in the 2013 revised capital rule remained unchanged." So there's a total of $68 billion to be raised.

3 See page 11 of the proposed rule, and section 217.10(c)(4)(ii)(H) (page 66). It actually gets more complicated; some off-balance-sheet commitments have higher conversion factors.

4 See section 217.10(c)(4)(ii)(D) (page 62 of the proposed rule). This can be reduced by some netting transactions.

5 For the rest of these weights, and then some, see section 217.10(c)(4)(ii)(B) of the proposed rule (page 60), and the table in section 217.34 (page 62183 here):

6 A related purpose is that the leverage ratio is meant to backstop the risk-based capital measures, and you want it to work well with those measures. In particular, it is intuitive that you want risk-based capital to be more binding than the leverage ratio -- you want the leverage ratio to be the backstop, not the primary regulatory capital measure -- because the risk-based measures, y'know, try to reduce risk. Here is Daniel Tarullo:

Numerous commenters have noted the potential for skewing the incentives of the largest financial firms if the leverage ratio becomes the binding capital requirement in ordinary times. Board staff estimates do suggest that this rule would make the leverage ratio more binding, relative to the risk-based capital ratios, for certain U.S. systemic banking organizations. However, the impact would vary substantially among firms, depending on their business models, and analysis suggests that these organizations could manage their capital structures to help meet the standards through certain low-cost, systemic-risk-reducing actions. It is also worth noting that, if we increase the risk-based capital surcharge for U.S. systemically important firms to a higher level than the minimum agreed to internationally, such as by reference to dependence on runnable short-term wholesale funding, the supplemental leverage ratio would be less likely to bind in normal times.

7 Loosely speaking the leverage ratio calculations are closer to international financial reporting standards in some respects, though not identical there either. There is some support in the U.S. for making our derivative accounting more like IFRS, to make banks' derivatives books look bigger and scarier.

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.