If you want to read the new Basel III leverage ratio rules, you have a stronger stomach than I do. Davis Polk helpfully ran a redline against the previously proposed versions of the rules, so that would be one place to start; it gets pretty red pretty quick, but the first sentence of the introduction is delightful:
That's not a feature, that's a cause! Anyway. I like the leverage ratio rules, though not enough to read them, because they are a neat little parable of financial regulation. That introduction goes on:
In many cases, banks built up excessive leverage while apparently maintaining strong risk-based capital ratios. At the height of the crisis, financial markets forced the banking sector to reduce its leverage in a manner that amplified downward pressures on asset prices. This deleveraging process exacerbated the feedback loop between losses, falling bank capital and shrinking credit availability.
The Basel III framework introduced a simple, transparent, non-risk based leverage ratio to act as a credible supplementary measure to the risk-based capital requirements.
And the rest of the leverage ratio rule is just two more paragraphs. No, I'm kidding, it's 19 pages, and it'd be longer except that it incorporates whole swathes of the (68-page) Basel III framework by reference.
Here's sort of how Basel III works:
- You make a list of a bunch of asset types and figure out -- through various efforts by banks, regulators, ratings agencies, whatever -- which ones are riskier.
- Some things are super safe, and you can finance them with 100 percent debt, no equity at all.
- Some things are pretty pretty safe, and you can finance them with about 1.7 percent equity.1
- Some things are the normal amount of risky, and you have to finance them with about 8.5 percent equity.2
- Some things are really quite risky, and you have to finance them with 17 percent equity.
- Other things are very extremely risky, and you have to finance them with 100 percent equity.
- A whole bunch of things have a risk based on your internal models, and you have to finance them with an amount of equity that is, loosely and conceptually speaking, based on your worst-case expected loss based on historical data, plus some buffer.
You can see why that would make people angry. Who's to say what's risky? In hindsight, a lot of the risk in 2007-2008 turned out to be in the things that regulators treated as having zero, or zero-ish, risk. That's not a surprise. It's a ... feature? Equity is costly, and so you might as well buy as much as you can of the stuff that doesn't require any equity -- even if, objectively, it's kinda risky. "Kling’s law is that the capital measure used by regulators will, over time, come to be outperformed by a measure that the regulators are not using."
Also, just, all these rules, man. It's all so complicated. It lends itself to gaming. Couldn't we have something simple and hard to mess with?
The leverage ratio is designed to do just that, as a backstop to the complicated risk-based rules. Here's how it's advertised:
- All of your things, whatever the risk-based capital rules say about them, need to be financed with at least 3 percent equity.3
But here's sort of how it works in practice:
- Some things are really super duper safe, and you can finance them with 100 percent debt, no equity at all, even under the "simple, transparent, non-risk based" rules.
- Some things are really disfavored, and so you need to have more than 3 percent equity even under those "simple, transparent, non-risk based" rules.
- A lot of things are lumped into the big 3 percent bucket, sure.
The story of the rules is that a new version was released today, and the big changes between the previously proposed rules and the new rules are mostly that some things have been moved from the 3 percent-ish bucket to the zero-ish-equity bucket. So you can net certain repo obligations: If I owe you a bunch of cash on overnight repos, and you owe me a bunch of cash on different overnight repos, and legally we can offset those amounts, under the revised leverage ratio rules I can treat only the net amount as an asset that requires 3 percent equity financing. Purely offsetting nettable matched-maturity repos now count as zero-risk things for the leverage ratio. The same is true for some cash-collateralized derivatives.
Other things that banks wanted moved to the zero risk weight bucket -- like banks' "liquidity buffers" of safe liquid assets -- stayed in the 3 percent bucket. A lot of things in the greater-than-3-percent bucket stayed there too.4
These changes seem sensible to me -- matched repos really are pretty risk-free! and shutting down that market would be bad! -- but that's not really the point. The point is that if you think of the leverage ratio as a pure, objective, simple, unbiased, non-risk-based thing, you are lost. The leverage ratio, like the risk-based capital rules, favors some activities (financing government bonds, in the case of risk-based capital, or running repo books, in the case of the leverage ratio) and disfavors others (writing credit default swaps, for one thing). It is based on judgments about which activities are safe and which are risky, which can be made more expensive in the interests of safety and which need to be preserved as is. Making regulation simple may be a worthwhile goal, but it's not simple.
1 Ehhh. So that's (1) a 20 percent risk weight times (2) an 8.5 percent capital requirement. Things like U.S. government-sponsored entities and depository institutions get 20 percent risk weights. Basel requires 6 percent "tier 1 equity" plus a 2.5 percent "capital conservation buffer." Plus other stuff. Plus there are other ways to count; like replace that 6 percent Tier 1 with 4.5 percent common equity or 8 percent total capital (including Tier 2). Who cares, I'm saying 8.5 percent equity.
2 Like I said, that's a number that I picked, but you could pick 4.5 percent or 7 percent or 10.5 percent or some other number.
3 Obviously, in the aggregate. Money is fungible, you can finance some things with too little capital and some other things with too much, as long as the math works out over all.
4 Back when I was younger and braver I gamed some of that out:
If you have a five-year interest-rate swap with $100mm notional that has a positive value to you of $5 million, that’s … $5.5 million of “assets” for the leverage test. (If its value to you is negative $5 million, it counts as $0.5mm of assets.) If you buy protection via a five-year credit default swap with $100mm notional that has a positive value to you of $5 million, it’s $10 (IG) or $15 million (HY) of assets. If you sell protection via a five-year credit default swap with $100mm notional that has a positive value to you of $5 million, it’s … I think $105 million of assets? In any case, the answer is never $5 million: derivatives assets are never measured at what they’re worth.
Got it? So if you have $5 million of swaps, you need to finance that with equity equal to 3.3 percent or 6 percent or 9 percent or 63 percent (3 percent of $105mm divided by $5 million) of their market value on your balance sheet.