Facts and myths about bank leverage ratios

This is kind of a techie article, but if you’re interested in banking at all, you probably ought to read it. Because if you’re interested in banking, and you’ve been following the policy debates of the last few years, there is a decent chance that you might believe some very untrue things about the way in which people measure capital ratios.

Ever since about mid 2008, when everyone had a massive panic about bank capital, there has been a community of people who wanted to ignore the Basel Committee on Banking Supervision, and their “Tier One Ratio” (based on equity divided by “risk weighted assets” or RWA). Instead, people like Anat Admati or Andy Haldane have suggested that we should just take the equity in the balance sheet, and divide it by the assets to get a “leverage ratio” without any of this complicated risk weighting stuff.

Are they right? In my opinion, yes to the extent that they want to look at the leverage ratio, but very much no to the extent that they want to get rid of risk weighting. It’s kind of like the flat tax argument — it’s true that our tax code is absurd and full of loopholes, but if you think that the complicated thing about it is that income tax has a progressive schedule, I’d say you’re not concentrating. Or even better, I’d make an analogy to sex education. It’s true that risk weighted assets and the Basel framework have gone badly wrong in the past and even contributed to some failures. But to take that as a reason for throwing the whole system away is to act rather like someone whose daughter just got pregnant and who is now demanding that everyone be taught abstinence only.

So, I’ll come back to the good things about leverage ratios at the end. But first, a little exercise in myth versus reality.

MYTH — Leverage is a “simple” and objective calculation

FACT — Only if you think that calculating a bank balance sheet is simple and objective

A balance sheet is fundamentally not a measure of risk, or even really of financial condition and ownership. It is what the name implies — it is the cumulative record of the transactions carried out in the other financial statements, which ensures that the debits and credits “balance”. Because of this, balance sheets often don’t contain important items like undrawn overdrafts, potential future derivatives liabilities, guarantees and so on, because they haven’t given rise to any transactions which are there to be balanced. There are various fudges and solutions which the accounting profession has put together over the years to cope with this, but it’s still very much a matter of judgement as to what sorts of things can be considered certain or quantifiable enough to put them in the balance sheet.

Historically, auditors have done what I would call a pretty reasonable job in making these decisions (some critics of the profession would say that on some occasions they did a somewhat worse than reasonable job). But historically, the size of the balance sheet has not been a particularly contentious matter. If you turn it into the basis for the most important regulatory ratio there is, then you’re putting more pressure on this relationship, and you can expect that pressure to show up on the auditors.

MYTH — Leverage is a more conservative standard than RWA

FACT — Unless you make major adjustments to the “simple” leverage ratio, it misses whole categories of risk

How much capital should a bank set aside to cover the risk of having to make a massive legal settlement for LIBOR manipulation? How much to cover the variability of trading P&L? How much to cover “wrong way risk” — the correlation between the likelihood of having to make recourse to collateral, and a decline in the collateral’s value? How much for the credit risk on in-the-money derivatives positions? If you want to use a “simple” leverage measure (like the one proposed in the mercifully scrapped Brown-Vitter bank reform bill), then your implicit answer to these questions is “none at all”. The Basel Committee’s adjusted leverage measure (a proposal, which they’re in the process of finalising) addresses all these questions, but it isn’t a simple measure.

This matters, of course, because it’s related to the previous point. Sometimes banking reformers make a nod in the direction of derivatives by saying that they want “all off-balance sheet risks taken into account”. But they never say what they mean by this, because they can’t — in order to require capital for risks which aren’t on the balance sheet, you need to have a whole load of formulae and risk modelling which isn’t given to you by the balance sheet, and you lose the ability to pretend that the risks faced by a big bank are actually very simple and only made to look complicated by nefarious bankers with their “complicated” formulas.

MYTH — Leverage ratios can’t be “gamed” by the banks.

FACT — Leverage ratios are very often gamed

A useful way of thinking about a “simple” leverage ratio is that it is a risk-weighted asset measure with a very specific weighting scheme — a weighting of 0% for anything that you can get off the balance sheet and 100% for everything else. When you think about it in these terms, it’s easy to see what the biggest incentive for a bank is when it’s regulated according to balance sheet leverage — to try and get off-balance sheet treatment for things that don’t really deserve it.

Here’s an example — before the crisis, American banks had to meet leverage ratio requirements. So they wanted to get a load of subprime mortgages off their balance sheet. So, among other bright ideas, they invented the Special Investment Vehicle (SIV). In order to persuade people to buy these things, they then invented the concept of the “liquidity put” — a provision which ensured that if a SIV ever went bankrupt, its only creditor would be the bank that had sponsored it in the first place.

There was no economic sense in which the risks associated with the SIV had been removed from the bank. But the assets were reduced, and the leverage ratio accordingly improved. This is a really important thing to remember — if and when RWA calculations go wrong, you can always look at balance sheet leverage as a backstop. If the leverage ratio is miscalculated, then this means that the financial statements themselves have been corrupted, and there is nothing left to help you. Stefan Ingves, the former head of the Basel Committee, used to be fond of pointing out that the subprime mortgage crisis did not blow up as a result of the Basel capital standards — it happened in a country (the USA) which had failed to even implement Basel II, and in a set of institutions (US broker-dealers) which were exclusively regulated by leverage ratios.

MYTH — Risk weighted asset calculations are always fudged and faked by the banks

FACT — No evidence has been found of this despite a dozen studies

The international regulatory authorities suspected in around 2010–11 that there might be some overly aggressive assumptions made in the calculation of risk weighted assets, and so they launched a big exercise in assessing their comparability. There have now been something like a dozen reports produced by the Basel Committee and by national regulators on this subject, and the results have been fairly consistent. First, there is variability in the RWA assigned to assets. Second, that the biggest driver of that variability is differences in the way in which the supervisors themselves have drawn the rules up. Third, there is really almost no evidence of systematic gaming of the system. In the vast majority of cases, the standard deviation of the capital requirements on a given portfolio has been around a third of the mean. And there is one real statistical tell-tale — as the exercises have progressed to include more and more portfolios, the variance has tended to decrease as a proportion of the average. This is what you’d expect to see if the variation was being driven by random errors. If the RWA were being systematically and intentionally miscalculated, you’d expect to see the opposite result — the errors wouldn’t cancel each other out, because they would all be made in the same direction.

This means that the issue of RWA variability is more theoretical than real. There is considerable variation in the way in which the RWA amount is calculated for different kinds of assets, but when all the different calculations are added together, it looks like banks make about as many overestimates as underestimates, so the Law of Large Numbers is working in the regulators’ favour. In principle, banks might be able to falsify RWA calculations, but in practice there’s no evidence that they do. Contrast this theoretical worry with the definite, observable fact that leverage ratios have been fudged and you start beginning to worry whether people have been chasing up a wrong tree for the last five years (and, a cynic might suggest, a systematically wrong tree which has tended to have the effect of making big US banks look better than European competitors).

So, why is “simple” leverage really important?

So, given all this, what’s the benefit of including a leverage ratio in bank regulations? Well, I’m going to assume that by “leverage ratio”, one means a sophisticated measure like the Basel Committee adjusted standard, otherwise the answer is “none at all, and possibly a great deal of harm”. And there are two, related, legitimate reasons why the leverage ratio is a big step forward.

First, it acts as a checksum. Most of the things that a bank might do to fool the risk-weighted asset ratio would have the effect of making the balance sheet bigger and the leverage ratio worse. Most of the things that a bank might do to fool the leverage ratio would have the effect of piling up tail risks and making the risk-weighted assets ratio worse. It’s comparatively difficult to think of measures which can fool both ratios at the same time.

Second, and related to this, it helps to avoid “corner solutions”. What you don’t ever want to do in banking is to create the impression that a particular line of business has a zero capital requirement. Because if it has zero regulatory capital, it looks like it’s got a return on capital which is close to infinite. And if the return on capital is infinite, then you can bet that banks with bad internal systems (ie, those that effectively allow their capital allocation to be set by the regulator) are going to want to do a hell of a lot of it. That’s why, for example, UBS ended up owning a mountain of super-senior CDO tranches. Or Laiki Bank ended up owning a slug of Greek sovereign debt. (Or, indeed, from the other point of view, how Citigroup ended up writing so many liquidity puts on SIVs). If you have more than one ratio, you have much less chance that any activity is going to get a very low capital charge.

So that’s the real point of the leverage ratio. It’s a good idea. But steer clear of anyone who treats it as if it might be a panacea. And definitely, the moment that the word “simple” is used to describe a bank leverage ratio, you should mark the speaker down as somebody who has never really thought hard about a balance sheet in concrete rather than abstract terms in his or her life.