Liberty Matters

A Response to Hülsmann, Hummel, and Selgin

I thank Professors Guido Hülsmann, Jeff Hummel, and George Selgin for their thoughtful commentaries. I am broadly in agreement with Hummel’s and Selgin’s comments, but not surprisingly have some serious differences with Hülsmann’s views on banking theory and his interpretation of Mises. (See our previous exchange on free banking in Hülsmann 2003 and White 2003.)
Hülsmann writes that “All media of exchange need a nonmonetary value component.” For fiat money, he says, this component is the “threat of violence,” meaning the collection of taxes and the enforcement of legal-tender laws. This seems to me an unnecessary concession to the state theory of money. Some important evidence weighs against the view that state enforcement is a necessary condition for the continued circulation of a fiat money once it has been launched. In particular, the Somali shilling continued to circulate in Somalia even after the state disappeared, ending the state’s ability to collect taxes or enforce legal-tender laws (Luther and White 2013).
Interpreting Mises’s argument in The Theory of Money and Credit, Hülsmann writes: “Mises insists that changes in the supply of and demand for money do not benefit the economy as a whole.” In fact, this is true only of a change in the nominal quantity of fiat money. It was Mises’s view (and it is my view) that a gold standard is different because monetary gold is costly to produce.
Mises explained how a reduction in the demand for monetary gold due to the development of fractional-reserve banking is beneficial under a gold standard. Consider the ordinary case of the global economy in which the stock of monetary gold continues to rise due to gold mining. Because there is no benefit to the economy from increasing the stock of monetary gold in the world as a whole, the labor and capital sacrificed to produce an increased stock is a cost without a benefit. Banking developments that reduce the demand for monetary gold accordingly benefit the economy by avoiding this cost. With respect to the development of fractional-reserve banknotes and checking accounts as substitute media of exchange in a specie economy, Mises accordingly wrote (Theory of Money and Credit, Book III, Chapter 17, para. 4): “If metallic money is employed, then the advantages of a diminution of the demand for money due to the extension of such other means of payment are obvious.” Historically, given the growth of real economic activity,
the tremendous increase in the exchange value of money, which otherwise would have occurred as a consequence of the extension of the use of money, has been completely avoided, together with its undesirable consequences. If it had not been for this the increase in the exchange value of money, and so also of the monetary metal, would have given an increased impetus to the production of the metal. Capital and labor would have been diverted from other branches of production to the production of the monetary metal.... [In consequence] the welfare of the community would have suffered. The increase in the stock of precious metals which serve monetary purposes would not have improved the position of the individual members of the community, would not have increased the satisfaction of their wants; for the monetary function could also have been fulfilled by a smaller stock. And, on the other hand, a smaller quantity of economic goods would have been available for the direct satisfaction of human wants if a part of the capital and labor power that otherwise would have been used for their production had been diverted to mining precious metals.
Mises went on to note that this resource-cost-saving argument applies to a commodity money regime, but not to a credit money or a fiat money.
The statement just quoted comes from the third part of The Theory of Money and Credit. It is thus not correct to write, as Hülsmann does, that “In the third part of The Theory of Money and Credit, he [Mises] refutes one by one the claims purporting to show that the creation of fiduciary media by fractional-reserve banks could be beneficial from an overall point of view.” In the above-quoted passage Mises makes an important claim of exactly this sort himself.
Likewise it is incorrect to attribute to Mises the (unwarranted) view that “not only are fractional-reserve banks useless from an overall point of view, but they are also in fact harmful to the economy.” Compared to a system where payments can only be made in specie or in transferable titles to specie issued by money warehouses, fractional-reserve banks – far from being useless – provide payment services at lower cost both from the individual transactor’s point of view and from the overall point of view.
In my lead essay I credited Mises with improving economists’ understanding of how adverse clearings limit the quantity of bank-issued money. Hülsmann, by contrast, attributes to Mises the view “that the issue of fiduciary media was in principle unlimited,” which I take to be a paraphrase of Mises’s statements (Theory of Money and Credit Book III, ch. 17, para. 26) that “[t]he circulation of fiduciary media ... is only elastic in the sense that it allows of any sort of extension of the circulation, even completely unlimited extension,” and that “[t]he quantity of fiduciary media in circulation has no natural limits.” I quoted the second statement and pointed out that we will miss Mises’s key insight that it does not apply to a competitive free-banking system if one ignores, as Hülsmann does, the immediately following paragraph. There -- I quote it again in hopes that it will not be overlooked again -- Mises writes:
Of course, all of this [lack of a natural limit to bank-issued money] is true only under the assumption that all banks issue fiduciary media according to uniform principles, or that there is only one bank that issues fiduciary media. A single bank carrying on its business in competition with numerous others is not in a position to enter upon an independent discount policy. If regard to the behavior of its competitors prevents it from further reducing the rate of interest in bank-credit transactions, then -- apart from an extension of its clientele -- it will be able to circulate more fiduciary media only if there is a demand for them even when the rate of interest charged is not lower than that charged by the banks competing with it. Thus the banks may be seen to pay a certain amount of regard to the periodical fluctuations in the demand for money. They increase and decrease their circulation pari passu with the variations in the demand for money, so far as the lack of a uniform procedure makes it impossible for them to follow an independent interest policy. But in doing so, they help to stabilize the objective exchange value of money. To this extent, therefore, the theory of the elasticity of the circulation of fiduciary media is correct; it has rightly apprehended one of the phenomena of the market, even if it has also completely misapprehended its cause.
What is the referenced “theory of the elasticity of the circulation of fiduciary media” that has “completely misapprehended” the cause of the demand-elasticity of bank-issued money? It is the pair of erroneous Banking School doctrines known as the Fullartonian “law of the reflux” and the Real Bills Doctrine. The correct apprehension of the cause is the theory of adverse clearings, as partially understood by some Free Banking School writers in the 19th century, better understood by Mises, and as most thoroughly explained in recent years by George Selgin in The Theory of Free Banking (1988) and other works.
Under what conditions is it true that “the issue of fiduciary media [is] in principle unlimited”? The issue of fiat money is unlimited by economic forces, but “fiduciary media” refers to redeemable bank liabilities and not to fiat money. Redeemable liabilities are not gratuitously issued but are costly for any bank to keep in circulation in a competitive environment. The issue of fiduciary is unlimited only in the analytical limiting case (never historically realized) of a single world banking system with a single issuer of fiduciary media (or a perfect cartel among all issuers) and zero public demand to hold the medium of redemption (which, by the way, is inconsistent with gold being the medium of redemption), so that the issuing bank’s risk of reserve loss is zero no matter how great the volume issued. Short of that case, the risk of adverse clearings or over-the-counter reserve losses strictly limits the ratio of bank-issued money to bank reserves to a finite number.
Hülsmann seems to think – despite all historical evidence (Dowd 1993, White 1995, White 2003) – that a perfect cartel among all banks of issue is a reasonable approximation to historical systems in which banks were free to issue fiduciary media. He even cites Mises in a way suggesting that he thinks that this was Mises’s position: “After all, fractional-reserve banks have an interest in agreeing on uniform policies to facilitate credit expansion. They do not need governments and central banks in this regard (see Mises 1912, p. 426; 1953, p. 397).”
This statement is mistaken in three respects. First, if banks could collude, their interest would lie, like any monopolist’s or monopoly cartel’s, in restricting industry output and thereby getting a higher price and a monopoly profit. In banking, this means that a profit-seeking bank cartel would seek to charge higher interest rates on loans (and to pay lower interest rates on deposits). To do so requires the cartel to move up the industry demand curve for bank loans to a smaller volume of loans, which implies a smaller volume of liabilities (holding constant banks’ other interest-earning assets and non-deposit sources of funds) than under competition. Second, again like firms in any industry, there is no reason to think that banks can successfully collude without government (via the government central bank or otherwise) to enforce the cartel’s prices. When the rest of the industry is charging loan rates above the competitive level (the interest rate on deposits plus the marginal cost of intermediation), any individual bank can profit by undercutting the cartel, lending more at a slightly lower loan rate, thereby undermining the cartel’s attempt to fix the loan rate. Third, Mises quite clearly found it absurd to suppose that a banking industry could collude to bring about a uniform proportional expansion in every bank’s liabilities, so as to avoid the adverse clearings that constrain any non-uniform expansion. Mises noted in Human Action that the self-interest of reputable banks lies in not cooperating with less responsible banks. He wrote (Mises 1966, ch. 17, para. 159):
But, some people may ask, what about a cartel of the commercial banks? Could not the banks collude for the sake of a boundless expansion of their issuance of fiduciary media? The objection is preposterous. As long as the public is not, by government interference, deprived of the right of withdrawing its deposits, no bank can risk its own good will by collusion with banks whose good will is not so high as its own.… Under free banking a cartel of banks would destroy the country’s whole banking system. It would not serve the interests of any bank.
Let me turn to Jeff Hummel’s contribution. Hummel notes that historically some governments have launched new monies without first giving them a fixed spot-redemption value, by promising future redemption in gold or receivability for taxes. These, he rightly notes, are “credit money” in Mises’s terminology (as translated). He considers this a correction of my statement that the launching of new government fiat monies has always proceeded by ending redeemability of a previously spot-redeemable money. I do not dispute these cases, and I would have worded my statement differently if I had thought about them. However, Hummel and I are not in disagreement, because when I said fiat money, I meant fiat money, not credit money.
Hummel cautions that my essay “goes too far when he seems to imply that Mises had in mind the kind of free banking … system in which reserve ratios are extremely low and banks adjust the money supply to demand in a way that stabilizes velocity.” Mises did of course emphasize the restraint that free banking imposes on the volume of bank-issued money. Hummel refers to Mises’s “unambiguous desire to keep fiduciary media tightly constrained,” and appropriately quotes Human Action (Mises 1966, p. 443): “Free banking is the only method available for the prevention of the dangers inherent in credit expansion.”
I see no difference between Mises’s view and my own view in that regard – I also desire a system that keeps the volume of money tightly constrained to its warranted volume.
But Mises in Human Action (p. 446) does quote Cernuschi to the effect that free banking would have narrowed the use of banknotes considerably, and in other ways suggests that reserve ratios under free banking would be, as Hummel puts it, “up very high and possibly close to 100 percent.” If that is Mises’s prediction, then on this point I do depart from Mises. In my 1992 essay that Hummel cites, I criticized Mises for suggesting that free banking would produce reserve ratios close to 100 percent. The best historical evidence we have, from the Scottish free-banking system and other mature systems, shows reserve ratios below 10 percent. This appears to be another instance of the point Selgin made in his contribution: that an empirical question of magnitude cannot be settled by a priori reasoning. We need to look at the historical evidence.
Hummel highlights Mises’s division of bank-issued money into “money certificates” and “fiduciary media.” According to this conceptual scheme, if a bank issues 100 in banknotes against 10 in reserves and 90 in loans, the first 10 banknotes are “money certificates” and the remaining 90 are “fiduciary media.” I am puzzled by this way of dividing things. It seems anti-subjectivist, because from the point of view of a banknote’s holder, the “last” banknote issued by a bank with fractional reserves is identical to the “first.” It makes more sense to distinguish inside money from outside (base) money, or equivalently, bank-issued money from reserve money.
Hummel goes on to quote Mises’s statement that free banking “would, it is true, not hinder a slow credit expansion, kept within very narrow limits, on the part of cautious banks.”
It isn’t entirely clear what proposition Mises is advancing in this sentence. Under a constant set of parameters, the equilibrium quantity of bank-issued money is determined. A slow credit expansion requires a slow change in one or more of the parameters. But exactly what parametric change is Mises supposing? A slow secular decline in the cost of topping up low reserves on short notice? A slow rise in the fraction of the public prepared to hold bank-issued media? If one of these two secular trends is assumed, then I am in full agreement the proposition that a free banking system would respond with an appropriately slow expansion in bank-issued money.
Turning now to George Selgin’s comments, I agree with most of what Selgin has to say about the regression theorem and Bitcoin. I would reemphasize the point I tried to make in my lead essay, that while Bitcoin’s appeal to antistatists in undeniable, that link does not determine the magnitude of the real purchasing power of one Bitcoin in the same way as a “rock-bottom” (to use Hülsmann’s term) purchasing power of gold is determined by the intersection of the supply curve with the nonmonetary demand curve for gold. To know where to draw the demand curve for real Bitcoins, we would have to suppose that the ideological desire is not merely to participate in holding a nonstate medium of exchange, but to hold so many dollars-worth.
Selgin’s hypothesis that Bitcoin’s founders were “making conspicuous leaps onto their own bandwagon, so as to encourage others to do so,” and in so doing were pursuing “a clever marketing strategy,” seems plausible, but awaits historical documentation. Since there is a public record of all Bitcoin transactions, it should be possible to discover whether there was an in-group pattern of offsetting trades in the early transactions. If true, we should expect to see the same pattern in the launchings of the 70 other cyber-currencies (listed at <>) that emulate Bitcoin, the largest of which is Litecoin.
Needless to say, given my own efforts (White 2013, 2014) to show that the gold standard has outperformed fiat money as judged by price indices and other historical statistics, I agree with Selgin that “reference to the best available statistics” trumps “mere assertions.”
Finally, I agree with Selgin that Mises was far too sweeping when he wrote that “Issuance of fiduciary media, no matter what its quantity may be, always sets in motion those changes in the price structure the description of which is the task of the theory of the trade cycle.” As I noted in a footnote in my 1992 essay, an increase of bank-issued money that matches an increase in the demand to hold bank-issued money is not disequilibrating but the reverse. It prevents a liquidity spillover from the market for money balances to the market for loanable funds, and thereby prevents a disequilibrating rise of the market interest rate above the natural rate, rather than causing the market rate to fall below the natural rate.