A quick explanation of “fractional-reserve banking”

Elsewhere, Eric Posner, Walter Block, and Bryan Caplan debate “fractional-reserve banking.” Aren’t you glad you read UR, rather than these other blogs, with their little premasticated spoonfuls of brain? Here at UR we don’t send you away hungry, that’s for sure.

It pains me to have to say that all three participants in this debate are out to lunch. Posner and Caplan get the wrong answer for the right reason. Block, and the Rothbardians generally, get the right answer for the wrong reason.

The critical fact about FRB is that it’s a special case of our favorite financial solecism, maturity transformation. As I explain in more detail here, MT is the cause of the credit cycle, including of course our current unpleasantness. It is a source of grim hilarity to me that, almost a century after Mises first discovered this, most people still don’t get it.

100%-reserve banking, as Block proposes, is of course MT-free and thus not a cause of credit oscillation. So, again, he gets the right answer. But I suspect what strikes most readers about the dialogue is that Posner and Caplan seem more or less sensible, while Block seems to have some weird abstraction he can’t let go of. So let me provide a brief overview and try to straighten everyone out.

In classical “fractional-reserve banking,” a bank balances liabilities which are demand deposits with assets which are loans of nontrivial maturity. Let’s assume for simplicity that our currency is gold. So a fractional-reserve bank might issue notes redeemable for 100kg of gold, while having only 10kg of gold in the vault, plus bonds whose net present market price is 95 kg. Our bank’s “reserve ratio” is thus 10%, and its “leverage ratio” is 20 to 1 (because it has liabilities of 100kg and capital of 5kg). Rocket science, this ain’t.

Block asserts: the bank is fraudulent, because it does not have the gold to redeem its notes. Posner and Caplan assert: the bank is solvent, because it can sell its bonds for 95kg of gold. In a frictionless market, this transaction can happen as fast as customers can redeem.

Posner and Caplan are hereby directed to Arnold Kling, who seems to have finally understood the problem with MT. In short, the assumption that mark-to-market accounting and frictionless transactions imply that term loans can be liquidated at market price is wrong. It ignores the collective game theory of the bank-run problem.

Briefly: the price of a bond or loan is a function of its interest rate. Interest rates in a free market are set by supply and demand. In a bank run, all banks must sell future money for present money, driving yields arbitrarily high and loan prices arbitrarily low. An obvious, extreme, but historically common absurdity occurs when the financial system as a whole has issued more demand notes for present gold than the amount of monetary gold in the world. (Believe it or not, this was the case even for the 19th-century “classical gold standard.”) There’s simply no way it can redeem. Time travel is impossible; transmutation is impractical; matter is conserved.

So Block is right. In general. However, Block’s argument (not his, but originally due to Rothbard) is awfully strange, and bears only a remote resemblance to either logic or reality. I suspect that this confuses a lot of people into thinking that there’s no there there, FRB is fine, these are not the droids we’re looking for and the financial system may go about its business. Let me try to untangle Rothbard’s twisted libertarian reasoning for you.

Rothbard, certainly one of the 20th century’s top five philosophers, was a generalist and synthesizer of incredible breadth and power. What strikes the reader of Rothbard immediately is his razor-sharp consistency: hardly a piece is out of place. It is almost impossible to find a crack in his edifice.

So how does he get to this weird view of FRB? He starts with an ethical premise: slavery is wrong. If slavery is wrong, it must be unethical to sell myself into slavery. In the world of Rothbardian ethics, this is because I have inalienable rights which I cannot alienate (sell), namely, the control over my own body, which will always be mine no matter what I do.

(I part ways with Rothbard here. While hereditary slavery is more debatable, I don’t have a problem at all with selling yourself into slavery. For me, a contract is an enforceable promise; removing my option to make enforceable promises cannot benefit me. If you don’t want to make the promise, don’t sign the contract. And promising to be your faithful servant so long as you and I shall live is a perfectly normal, legitimate, and (in a sane world) common sort of promise.)

In order to keep slavery illegal, Rothbard has to reach for what I consider a very odd definition of a contract. To Rothbard, a contract is always a transfer of goods. So, for example, if you pay me $1000 in exchange for painting your house, I have not entered into a slave-like bond to faithfully paint your house. Rather, I have transferred to you the shadowy, metaphysical, and thoroughly virtual object of a paint job on your house. If I then fail to paint your house, I have in a sense stolen your paint job, and thus am a thief.

This is why the idea of a bank deposit as a bailment contract is so appealing to Rothbard. It fits his definition of a contract perfectly. It is also perfectly clear that if the bank and its customers agree that the contract is a bailment, and notes are warehouse receipts, a fractional-reserve bank is fraudulent. Historically this is a very common case, and Rothbard knew his history.

However, the suggestion that openly-agreed fractional-reserve banking is fraudulent is—as Posner and Caplan point out—untenable. (Remember, Posner and Caplan follow their correct argument to the wrong answer because, while open FRB is not fraudulent, as a case of MT it remains imprudent—for the depositor. Unless of course it is insured by a fiat issuer acting as a lender of last resort. Which is equivalent to the case in which the LLR is an LFR and just makes the loans itself. As it seems to be doing these days. But I digress.)

Moreover, Rothbard slips here. His argument is untenable even in terms of Rothbardian ethics. There is a crack in the great edifice.

Rothbard approves of the normal financial practice of term loans, or as he calls them “time deposits.” In a time deposit, A gives B $(X) at time (T), in exchange for the promise to return $(X+k) at time (T+u). Of course, as a Rothbardian contract, this promise becomes the shadowy construct of virtual future money.

The problem is that a demand deposit is an extreme, but qualitatively indistinguishable, case of a time deposit. A demand deposit is the limit of a time deposit as (u) approaches 0. For example, one might imagine a time deposit in which case (u) equals one second. When the loan times out, it is automatically renewed (rolled over), unless of course you are standing at the ATM and you want your money back. In which case you have to wait one second for your latest loan contract with the bank to time out, and the bank to return your money.

This clearly has exactly the same practical effect as a demand deposit. If a second is too long to wait at the ATM, we can make it a tenth of a second, etc. And it is clearly a time deposit. And there is certainly no way we can draw a line at any interval of time, and say that if the term of the loan is more than five minutes it is nonfraudulent, and if it is less it is fraudulent.

Thus, Rothbard and Block are just plain wrong. But they get the right answer anyway, because FRB is a special case of MT and MT is harmful. The real reason that 100%-reserve banking will defeat FRB in the free market is that depositors will avoid financial institutions that practice MT, and absent MT the natural rate of interest at a term of 0 is 0—because there are no productive investments that can produce an instant return on capital.