Now the extent of that part of the demand for a medium of exchange which is displayed on account of its service as a medium of exchange depends on its value in exchange. This fact raises difficulties which many economists considered insoluble so that they abstained from following farther along this line of reasoning. It is illogical, they said, to explain the purchasing power of money by reference o the demand for money, and the demand for money by reference to its purchasing power.
The difficulty is, however, merely apparent. The purchasing power which we explain by referring to the extent of specific demand is not the same purchasing power the height of which determines this specific demand. The problem is to conceive the determination of the purchasing power of the immediate future, of the impending moment. For the solution of this problem we refer to the purchasing power of the immediate past, of the moment just passed. These are two distinct magnitudes.
— Ludwig von Mises, Human Action (Auburn: Ludwig von Mises Institute, 1998), pp. 405–406.
Many people interpret Mises’ regression theorem as claiming that all money must have some value as a non-monetary commodity. This is only true for the first currencies. It’s possible for a new currency, such as bitcoin, to be first introduced as a substitute to an already circulating currency (e.g. the U.S. dollar), therefore basing its purchasing power on that of the dollar (by setting an exchange rate and basing prices on U.S. dollar prices [or Euro, et cetera, prices]). From there, bitcoin can go its own direction and become a competitor, rather than a substitute. An accurate reading of Mises — at least, in my opinion — suggests that bitcoin doesn’t violate the regression theorem any more than the occurrence of fiat money does (which first arose as a substitute to outside money and then went its own direction).
Where initial substitution isn’t possible, a potential currency must have some use value. As a commodity is more widely traded, its purchasing power can be partially decided by this use value. That is, as the first monies spontaneously emerge, money prices will have to initially be based on the value these commodities carry as producers’ or consumers’ goods, and then a liquidity premium is added as the commodity circulates more widely. These are currencies that can’t base their purchasing power on already existent currencies, by initially acting as substitutes.
If you’re still having trouble explaining bitcoin, think of the U.S. dollar, backed by whatever assets the Federal Reserve owns. From where did the U.S. dollar derive its value? Some say from taxes. The problem is that taxes are monetary transactions, and it represents the U.S.’ demand for money. But, for the U.S. government to know how much it should tax, it needs to have some basis to calculate its demand for money — without money prices it can’t do this. Instead, the U.S. dollar — say, after 1971 — based its purchasing power on the already existent dollar when it was still redeemable, which in turn continued to base its value on prices of the recent past. We can trace this back to the national banking era, where the U.S dollar was first introduced as an alternative substitute for outside money. Bitcoin has essentially the same history, except that it’s a step beyond fiat money, in that it originally used (if it still doesn’t) fiat money as a means of estimating its initial purchasing power.
The vehement rejections of bitcoin based on the regression theorem are all misreadings of the latter. Of course, this doesn’t mean the bitcoin is preferable to alternatives as a general means of exchange. I’ll continue to use the U.S. dollar for the foreseeable future, because it’s less volatile and less sensitive to speculation. But, the regression thereom is not an adequate basis for a critique of bitcoin.