In the debate over financial reform, one of the major themes has been the need to protect the public purse from the cost of future bailouts by requiring too-big-to-fail banks to hold a larger buffer of private capital. In support of this requirement, and against bank lobby claims that it would raise their cost of funds and hence the price of credit, economists have weighed in with a standard argument from corporate finance, the Modigliani-Miller theorem.
Unfortunately, as suggested in a forthcoming article in the Harvard Business Law Review, this argument may not work as well as its proponents think it does because banks are, in important ways, not like the ordinary firms to which the theorem is intended to apply.
The Modigliani-Miller theorem is sometimes summarized as saying, in effect, that cutting a pizza into more slices, or into square slices instead of pie-shaped slices, does not change the size of the pizza. More technically, under certain assumptions, it can be shown that the value of a firm is not affected by its capital structure, in particular by the fraction of its capital structure that is equity.
Applying this line of thinking to banks, it has been argued (most prominently by Anat Admati here and here) that requiring banks to fund themselves with more equity and less debt, especially short-term debt, should not in principle change the value of the banks, and hence should not change their cost of funds. If in practice it turns out that the cost of funds does go up a bit, that would only be because the market has been including the implicit call on the public purse as an asset of the firm, a calculation that it is the appropriate purpose of financial reform to eliminate.
I have to say that, when I first heard this argument, it sounded wrong to me, but I couldn’t immediately see why, so I put the matter aside to think about later, when I had more time. Last week, that time came during a stimulating intellectual interchange at a pro-seminar on “Re-theorizing Liquidity” sponsored by Harvard’s Institute for Global Law and Policy.
For that gathering, Morgan Ricks, a Visiting Assistant Professor at Harvard Law who spent much of the crisis working at Treasury, supplied as background his forthcoming paper “Regulating Money Creation after the Crisis”. On page 527 of that paper he asks the crucial question “Does the Modigliani-Miller theorem hold true for maturity-transformation firms?” The answer, he argues, is “No”.
Now Ricks is a lawyer, by training and by practice, but his reasoning is impeccable economics. He appreciates, as pizza slicing analogies do not, that the underlying logic of Modigliani-Miller relies on arbitrage. The reason that capital structure does not matter for the value of the firm is that (by assumption) investors can undo the capital structure decision of the firm managers by their own portfolio decisions. Specifically, if they want more risk than an outright equity position offers, they can add leverage to their holding of the equity.
Ricks argues that this arbitrage argument fails in the case of banks precisely because banks can fund themselves with monetary liabilities while investors cannot. If banks were forced to term out their funding, i.e. fund their assets with longer term and equity claims, they would no longer be banks, they would be finance companies. And the former holders of money claims would not be able to do anything about it, since their own liabilities do not pass as money.
Investors may be able to roll their own leverage but in general they cannot roll their own liquidity. Quite the opposite in fact, investors are demanders of the liquidity that the banks supply by issuing money claims. Nothing of this is captured in the world of Modigliani-Miller.
“Standard capital structure principles do not necessarily apply to these firms, because most of their funding comes not from the capital markets, but from the money market.”
The key point is not so much that banks fund themselves in a market inaccessible to the rest of us, since we know from shadow banking that non-banks also have access. The key point is that money market funding is at lower rates than capital market funding. Society is apparently willing to pay a premium (lower yield) to hold money claims rather than capital claims. Why is that?
Ricks does not say, and frankly standard economics and finance does not either. The resulting asset price “anomalies”—the failure of the Expectations Hypothesis of the Term Structure—remain puzzles, but leave that for another time.
For present purposes, the pressing question is whether this means that the bank lobby is right to oppose requirements for increased equity funding? NOT AT ALL. It just means that we need a different intellectual framework to think about the problem. We need the money view.
More generally—and going beyond what Ricks argues—from a money view perspective a bank is a special kind of dealer, so the question of appropriate bank capitalization is actually a larger question about the appropriate capitalization of dealers more generally, who fund themselves primarily in the wholesale money market, using repo for example.
I don’t know the answer to that larger question, but it seems to me definitely the right entry point into the problem. Models (like M&M) that abstract from liquidity, the very business of dealers, cannot be expected to address the question of appropriate dealer capitalization very satisfactorily. New thinking is required.