US banking regulators have released new proposals to require banks to have higher leverage ratios, counterintuitively meaning lower leverage, and you can go read them here, or read about them here or here. Briefly: in addition to regular Basel III risk-based capital requirements, banks are also subject to a backstop equity-divided-by-assets0 leverage test, and internationally the minimum is 3%, but in the US it'll be 5% for the biggest bank holding companies and 6% for the biggest insured banks. The OCC estimates that the banks are in total about $84 billion or so short of that requirement, though they have five years to get there, so it's not, like, go sell $84 billion of stock right now or whatever.1
But here is a weird thing from the regulators' joint notice of proposed rulemaking: their estimate that banks' cost of equity capital is just 0.56% above their cost of debt:
To estimate the cost to insured depository institutions of additional capital associated with the proposed supplemental leverage ratio requirement, the OCC examined the effect of this requirement on capital structure and the overall cost of capital. The cost of financing a bank or any firm is the weighted average cost of its various financing sources, which amounts to a weighted average cost of capital reflecting many different types of debt and equity financing. Because interest payments on debt are tax deductible, a more leveraged capital structure reduces corporate taxes, thereby lowering funding costs, and the weighted average cost of financing tends to decline as leverage increases. Thus, an increase in required equity capital would require a bank to deleverage and – all else equal – would increase the cost of capital for that bank.
This increased cost would be tax benefits foregone: the additional capital requirement (between $84 billion and $123 billion), multiplied by the interest rate on the debt displaced and by the effective marginal tax rate for the banks affected by the proposed rule. The effective marginal corporate tax rate is affected not only by the statutory federal and state rates, but also by the probability of positive earnings (since there is no tax benefit when earnings are negative), and for the offsetting effects of personal taxes on required bond yields. Graham (2000) considers these factors and estimates a median marginal tax benefit of $9.40 per $100 of interest. So, using an estimated interest rate on debt of 6 percent, the OCC estimates that the annual tax benefits foregone on between $84 billion and $123 billion of capital switching from debt to equity is between $474 million and $694 million per year ($474 million = $84 billion * 0.06 (interest rate) * 0.094 (median marginal tax savings)).
The OCC does not anticipate any additional compliance costs for banks or costs to the banking agencies. Thus, the overall cost estimate for OCC-regulated banking organizations under the proposed rule is between $474 million and $694 million per year.
I'm sure this is old news to people who read a lot of OCC rulemaking on bank capital so the three of them should stop reading now, but: isn't that a weird way to think? I mean, no, it's got its theoretical justification and all, but it sure sounds weird. Like:
- The only cost of capital is missing out on the tax benefits of paying interest.
- So the higher your interest rate the cheaper your debt. (?)
- But equity is more expensive than debt because the interest you pay on it isn't tax deductible.
- Even though you don't pay interest on it.
- But just pretend you did, and it had the same interest rate as your debt.
- Which by the way let's just assume is 6%.
- Even though the actual average interest expense of big US banks is something like 0.64% of the face amount of their liabilities.
All of those things are pretty crazy? Anyway as you can see the OCC computes that $84 billion of new equity capital will cost banks an incremental $474mm, implying that the equity premium is 0.56%. That seems low? Like I might respond:
- Banks' current actual cost of debt financing is around 0.64% (true!).
- The perceived "cost of equity financing," conventionally defined as the required return on bank equity capital, is often stated (with some justification) to be something in the neighborhood of 10%.2
- 10% minus 0.64% is 9.36%.
- 9.36% of $84 billion is $7.9 billion.
- This regulation will cost banks $7.9 billion a year.
Which is more than $474 million. More like 10% of annual profits, for instance, as opposed to less than 1%.
Now I'm being totally unfair here: like I said, there's theory behind the OCC's math. It's good straightforward Modigliani-Miller: the cost of capital doesn't depend on capital structure; you can't slice a pie in such a way as to create more or less pie.3 I feel like a lot of people - particularly though not exclusively finance academics - would more or less endorse it.
But a lot of banks wouldn't! You can see why capital regulation would be a contentious topic. Banks tend to look at conventional cost-of-equity measures and say, gulp, $84 billion of equity would be really expensive. Those who want more capital, on the other hand, think it's basically free.4
Agencies Adopt Supplementary Leverage Ratio Notice of Proposed Rulemaking [Fed, and NPR]
U.S. Banks Face Two Ratios as FDIC Sets Capital Vote [Bloomberg]
U.S. bank regulators propose 6 percent leverage ratio [Reuters]
0.[Update: Oh technically (A) Tier 1 capital divided by (B)(x) assets plus (y) certain off-balance sheet items but feh, equity over assets.]
1.From the notice of proposed rulemaking:
[A]t the five percent supplementary leverage ratio for holding companies, QIS and CCAR data suggest that the capital shortfall will range between $63 billion and $113 billion. After making the scalar adjustments to estimate insured depository institution data, at the six percent supplementary leverage ratio for insured depository institutions, QIS and CCAR data suggest that the bank-level capital shortfall will range between $84 billion and $123 billion.
We talked a while back about a KBW research report trying to figure out the shortfall using various counting mechanisms; here's their table, with "Basel 3 Leverage" being - maybe! - the most relevant number and 5% being - maybe! - the target:
So Morgan Stanley, maybe in trouble.
2.Consider that JPMorgan trades at a ~9.8x PE, implying an earnings yield of ~10.2%, and a 2.76% dividend yield. If you buy a JPMorgan share you sort of do expect a return of 10%, loosely speaking. But at the very least you expect a cash dividend of 2.76%. 2.76% is still like 2.2% more than JPMorgan's blended interest cost.
3.If you're a Modigliani-Miller purist you say:
- the cost of capital is invariant to the capital structure, in a Modigliani-Miller world;
- the main (?) difference between that world and our world is taxes;
- so tax savings really are the only cost of capital imposed by capital regulation.
That is, the actual blended cost of funding (6% or whatever else) is otherwise independent of capital structure and, thus, of capital regulation. This doesn't help with the 6% versus 0.64% interest rate thing, but of course short-term rates now are abnormally low and you might want to estimate the long-run costs of regulation using something other than the current actual cost of financing.
Now, as I've said before, I sold capital structure stuff as my job, so I just constitutionally can't believe in Modigliani-Miller in any strong way, but obviously a lot of people do. They tend to select out of banking.