This post is an attempt to figure out what ROEs, returns on capital, and capital structure really tells us about the risk of banks. It is partly a response to Martin Wolf’s FT blog post What do the banks’ target returns on equity tell us? but also more generally a comment on the whole banks can have more equity and it won’t cost much meme.
First, we should define some terms. By ROE or return on equity, I mean the return on equity based on the current market price. So if I buy a share for a dollar, and my total return (price appreciation plus dividend yield) in a year is seven cents, I have a 7% ROE.
Return on regulatory capital is slightly different. This is because capital is not the market price of a firm’s equity; rather it is* shareholders funds. This includes money raised from issuing equity, retained earnings, and a few other (typically smaller) items. Thus the return on capital is based on the price of equity at issue, plus whatever earnings have been retained, and a few other bits and bobs. Firms typically cannot issue meaningful amounts of new equity at the current price, as new equity dilutes existing shareholders. It also usually irritates them. Therefore it is usually easier to increase capital by retaining earnings rather than by issueing new stock.
The naive ‘banks can easily have more capital’ crowd typically argue thus:
- The Miller Modigliani theorem says that capital structure does not matter; if you have more equity, you are safer, hence your debt trades tighter, hence the extra cost of the equity is made up for in having cheaper funding;
- Remember that more equity makes banks safer?
- Get more of it.
The problem with this is that the Miller Modigliani theorem is false. Amongst other things, it assumes no taxes; it assumes that investors are risk neutral; and it assumes that risk is perfectly known by all investors. None of these things are true, and thus those who trade capital structure based on MM alone are known as ‘bankrupts’. For me, the biggest issues in MM are the linked ones of no risk premiums and perfectly known risk. Of course if I know the distribution of a firm’s earnings precisely then I can price the stock. It is the very fact that I don’t that makes equity investment difficult: it means that investors demand a variable but high premium for taking equity risk. Because an equity issuer has to pay this extra premium whereas a debt issuer doesn’t (or at least doesn’t have to pay as much of a one, assuming we are not in high yield territory), equity cannot be substituted for debt without cost.
Note that the existence of risk premiums means that you cannot assert that just because a bank’s target ROE is 15%, it is highly risky. Wolf makes this mistake, saying that mid teens ROEs demonstrate that ‘these desired returns must represent the result of extreme risk-taking’. Of course, he may be right, but one cannot conclude that from the evidence available.
What happens if you raise capital requirements above current levels? Clearly a bank has two choices:
- Raise more capital; or
- Reduce risk (as measured by capital requirements).
The first of these is typically done by retaining earnings, unless the new capital required is large, due to the aforementioned pissing-off-the-stockholders feature of issueing new equity.
It is the second I want to focus on, though. What would we do to reduce risk? Well, clearly, we need to cut those businesses with a low return on regulatory capital. And that, sadly, means commercial lending. Both retail banking (because it has relatively low regulatory capital requirements for the risk) and investment banking (because it, historically at least, has high returns) look better under return on reg cap than commercial banking. So that’s where you cut. And you cut savagely as your whole portfolio generates regulatory capital, so to meaningfully affect it, you have to make a dramatic change to the speed of origination of new business.
This is why I think that meaningful changes to banking regulation are best done in good times, not bad ones. In a boom, a decrease in the supply of credit is probably a good thing. It is certainly less bad than it is in a recession.
Note, by the way, that both increasing capital and reducing risk have the effect of reducing ROE. The first because either there will be more shares for the same earnings or because increasing retained earnings will be reflected in a higher stock price; the second because reducing risk will typically reduce earnings. The effect is not linear and not easily modelled though, especially capital tends to be sticky. For a large bank, raising another hundred million dollars of capital will often not budge the stock price at all, but a couple of billion will move it substantially.
At the end of the day, economics should be about what works. Yes, we want a more stable financial system. But we also want economic growth. Whether increasing bank capital requirements should or should not lead to decreased lending in your model or mine does not ultimately matter. What matters is whether it in fact does, and what the macroeconomic consequences of that are. It would be truly irresponsible to attempt banking reform in the depths of a recession without at least being alert to the possibility of adverse consequences, and moving quickly to address them if they occur.
*For simplicity here I have ignored deductions and a few other wrinkles which mean that common equity tier 1 is not quite the same as shareholder’s funds.