Why Banking Leverage Requirements Are Not Enough

Hoenig’s plan calls for Congress to remove risk-weighted capital requirements, stress testing, and failure resolution planning, replacing all of the above with a 10 percent capital requirement and internal restructuring. (The internal restructuring is incorrectly described as a return to Glass-Steagall [1].)

In practice, a 10 percent leverage ratio by itself would be far too low to do what they want it to do; the Minnesota Fed recently argued for a 15 percent leverage ratio alongside increasing the additional risk-weighted ratio to 23.5 percent as a start. But it’s important to understand the problems with it in theory.

This debate has gone back and forth in the financial reform community. To summarize broadly, there are leverage ratios, which are equity divided by assets, and there are risk-weighted capital ratios, which are equity divided by assets adjusted for perceived riskiness. Some argue we need both; others argue that only having a leverage requirement is enough, or even better.

I think it’s important to touch all three parts of the balance sheet, and with three targets we need three tools: Risk-weighting, leverage, and liquidity (the ability to make payments in a crunch) requirements need to go hand-in-hand. I think abandoning liquidity requirements is especially dangerous. Others, like Aaron Klein of Brookings, use a chopsticks metaphor to describe the balance.

I think it might be easier to explain this using a simple two-dimensional chart. This will be broad and require some assumptions, but I think those assumptions are clear and easy to make.[2] Let’s give it a try.

Two Choices a Bank Makes

A bank has to make two choices given the size of a portfolio. It has to choose the riskiness of the portfolio, and it has to choose the leverage of that portfolio, or the mix of how it funds itself with equity and debt. Let’s assume a strictly increasing relationship between them; all else equal, more risk means you should be funded with more equity to absorb losses. Stakeholders — which include owners of the firm, deposit and deposit-like holders, and society as a whole — want to ensure that these levels match up. (I’m assuming away liquidity risks, but this same exact argument will exist in another dimension based on a banks’ ability to make payments in a credit crisis.)

Let’s draw where the capital level matches the riskiness level as a straight line. Above it, in the green safe zone, there’s more capital funding than the minimum we’d like to see for the level of risk. Below it, in the red danger zone, there is more risk than desirable for a level of capital, or less capital for a level of risk.

Let’s also assume that, at the margins, banks would like to be below the line, but stakeholders would like them to be above the line. Banks benefit more from bigger bets with leverage. Because of information asymmetries, taxes, and other deviations from Modigliani-Miller, cost of capital is slightly increasing with higher capital eating into profits, and as such for any level of risk banks would like to fund it more with debt. Because of their dispersed nature and desire to hold informationally insensitive private money, stakeholders such as depositors and capital markets are bad at disciplining firms to stay at the line. But the cost of failures has negative externalities for the economy as a whole in the form of contagion, panics, and runs. As a result, banks would like to wander into the red zone, and capital regulations need to keep them in or at the green zone.

One ideal is to have capital requirements that reflect the risks of the portfolio. If the risks increase, so do the capital requirements. Capital requirements would thus be the line itself, as shown in the above graphic. This is the idea behind risk-weighting.

Leverage in the Zone

What does a leverage requirement look like in this scenario? A leverage requirement is a capital level that exists independent of the risks of the portfolio. It measures the size of assets, and weights all assets equally. Let’s draw such a line here.

As you can see in the red-circled area, the leverage requirement permits a bank to have more risk than the capital structure would find appropriate.

If we assume that the banks want some level of risk above their capital, that they want to move some amount of space into the red zone given an initial spot on the line, they can always do that by moving straight to the right. They do this by increasing the riskiness, but not the value, of their assets. They could, say, replace an A+ rated sovereign instrument that has a risk-weight of 20% with a BB+ that would have a risk-weight of 100%, increasing the riskiness of the portfolio without changing the size. They could take that BB+ one and replace it with a CCC+ one that would have a risk-weight of 150%, all without changing their leverage ratio. A useful example from Daniel Davies is they can simply buy into structures that are themselves highly leveraged, like a CDO, choosing a much higher level of risk that doesn’t get reflected in the balance sheet. They can try and hide risk off-balance sheet and in other ways that don’t show up in asset size; trying to tackle what goes in the denominator of a leverage ratio has been a big international fight.

Under the Hoenig/CHOICE Act leverage requirement regime, banks can always take on more risk, move to right, no matter where the leverage requirement is set. The whole point of regulations is to reduce or eliminate this movement, but with only this requirement in place, the bank can, in theory, choose to always go further into the red.

Combined Together

Nevertheless, as a special case of risk-weighted capital requirements, leverage requirements have two important advantages that make them an essential component of regulation.

Risk-weighted requirements are flawed, and there’s reason to believe they are flawed in one direction. While we think regulatory risk-weighting gives us the dotted line above, chances are the actual line where the bank is shifted to the down and right. Risk-weighting is pro-cyclical, and can also impact the underlying dynamics of the instruments themselves. It is a poor substitute when there is a market-wide correction of a class of assets, where for an asset class there’s actually much more risk involved. This is modeled here as a shift of the curve to the right. Risk-weighted requirements can also create situations in which banks game them, massaging their portfolios to mimic a safer construction of assets for a level of risk, here a shift of the curve down. As David Davies notes, this gaming is found less in the literature than you’d think and leverage ratios are also suspect to gaming, but it is a relevant concern. In both scenarios, we think we are in the green but we are actually in the red. Regulators can take action against this by requiring analyses like stress tests, but this approach is still imperfect.

All else being equal, it’s also a good assumption that low levels of equity become more dangerous the less there is. Less equity means less of a chance for corrective measures. It means failure can happen quicker and with less preparation. It’s also more dangerous the further into the red you go. We model this in the graphic above by making the danger zone redder as equity approaches zero or goes further to the right.

In short, you can see that risk-weighting is dangerous because it can lead us to believe we are more safe than we are, and much of the associated risk is in a deep red danger zone.

But what if we add a leverage requirement?

A leverage requirement is more binding, as it requires less estimation of risk. This is because it simply ignores the risks of the underlying portfolio. While risk-weighting can be game to ignore the level of risk in a portfolio, a leverage requirement ignores them by construction. However, when combined with a regulatory instrument that does account for risk, both components become more powerful than they are by themselves. They each remove the worst case scenario of the other, preventing them from going deeper into the red. This is why Dodd-Frank incorporates both, and why we shouldn’t assume that a leverage requirement by itself would be harder on the banks or safer for any of us.

[1] To see why they aren’t the same, compare Hoenig’s plan with Glass-Steagall below. Hoenig keeps the financial holding company structure created under Gramm-Leach-Bliley rather than break apart the firms. Instead, Hoenig forces new holding companies at the intermediate level to be separately capitalized and resolvable.

Hoenig’s plan appears to abandon single point of entry, where a failure is handled through an FDIC interjection at the overall holding company level. Hoenig would instead focus on making banks go through bankruptcy at multiple independent points at the intermediate holding levels. It’s not clear why that would work better or be desirable. Much of the argument for Glass-Steagall derives from less complexity and breaking up political power. Abandoning the current work done on a single focal point of resolution to put multiple points at the intermediate level would likely increase complexity, and it’s not clear how this would weaken the political power of the firm itself.

[2] It’s okay if you don’t think this is real economics. Milton Friedman infamously SAID Markowitz’s CAPM two-dimensional risk-reward box, the basis of portfolio theory, wasn’t real economics either. (Ketchup economics forever.)