The basic problem with Menger's approach, from my perspective, is that he's concerned with the historical origin of money, whereas I am concerned with its logical origin. What Menger wanted to know is how money actually happened. What I want to know is how it can happen.Whatever his intentions, Menger's account today has much more value as a logical than a historical account of the origins of money. Thanks to modern archaeology we now know that money (or at least goods that before the rise of coinage were valued primarily as intermediate goods, which I call "collectibles", the main example being bead jewelry) emerged long before the efficient markets that Menger assumes. Humankind thus did not, with the exception of certain short and exceptional situations during the colonial era, ever pass through a stage of efficient barter markets that is Menger's setup. Indeed so long ago (more than 100,000 years ago) did collectibles start being used that they probably played an important role in the evolution of human cooperation, as I describe here.
But let's get back to the logic of whether and how money will emerge in a voluntary and efficient barter market. Moldbug first gives a great description of why money is not like a normal commodity:
...since buying and selling any good cannot fail to affect its price - ie, its exchange rate against other goods - we have a feedback loop. The herd selects an intermediate good based on its predicted exchange rate. But the exchange rate cannot be predicted without knowing the herd's selection. Problem!Moldbug describes his setup world of Nitropia, in which the storage and transport costs on which Menger based his analysis are eliminated:
anyone can trade with anyone, anywhere, by teleporting goods. In addition, we'll assume that all goods can be stored perfectly without any overhead.Moldbug calls this an equilibrium where there is no money, just barter. I don't agree -- given the mental transaction cost assumption (see below), money could emerge even here, in the complete absence of storage and transport costs. I'll explain why below.
Moldbug then breaks this barter equilibrium by introducing a good with a storage cost, fish, which rots if not soon eaten. Sven the fisherman wants to fish, sell the fish, save the income, and when he's saved up enough buy a Cadillac.
Moldbug claims that this eliminates the coincidence-of-wants problem from his scenario, but I don't buy it. A coincidence of wants problem is just what we have here: customers want to eat the fish while it's still fresh but Sven does not want to purchase the Cadillac until he has saved up enough income for it (he apparently prefers delaying gratification to incurring the interest costs of credit).
Since the storage and transport costs of intermediate commodities are still zero, these provide no reason for Sven to choose one particular such commodity over another as a currency. But because as Moldbug says, "[t]ranslating between standards is a pain in the butt," out of these intermediate commodities a single monetary standard will emerge. In other words, Nitropia assumes transaction costs that create an incentive to converge on one currency. Since Nitropia has costless storage and transport, this is just what I have called mental transaction costs -- the costs involved in making buying and selling decisions. These include the costs of keeping books and otherwise tracking and comparing prices, and the costs of mentally mapping preferences to budgets via prices.
But mental transaction costs are a problem even if there are no commodities like fish with storage costs. A world of pure barter has O(N^2) prices for N commodities, and the mental transaction costs in such a world are correspondingly much higher than a world with a single currency and O(N) prices. So even a market with no transport or storage costs for any commodity whatsoever, but with sufficiently high mental transaction costs, will converge on a single currency.
Not even the elimination of all storage and transport costs eliminates all coincidence of wants problems. If Sven's customers want the fish Sven caught today because they are hungrier today than they expect to be ten years from now, they will prefer to buy it today even if Sven could costlessly store the fish to be sold with equal freshness ten years from now. Even with zero storage and transport costs, time preferences for production and consumption create noncoincidences of wants, and these mismatches combined with mental transaction costs give rise to a currency standard.
Menger's analysis does not and cannot show that the coincidence-of-wants effect is the only force that can result in standardized money. Perhaps there is another? Indeed there is.I disagree. What Moldbug's argument, properly corrected, shows is that some of the costs that arise from the noncoincidence of wants occur even if there are no transport or storage costs, but only mental transaction costs. This contrasts with Menger's analysis, which defined the costs caused by the coincidence wants in terms of transport and storage costs.
Suppose Sven is choosing between only two possible intermediate goods - Ia or Ib. Say Ia is palladium, and Ib is rhodium. What is Sven's algorithm? It's actually quite simple. All Sven cares about is the change in the exchange rate between palladium and rhodium, across the time window T1 - T0 of the transaction. If (Ia/Ib)@T1 is greater than (Ia/Ib)@T0, he prefers palladium. If it is smaller, he prefers rhodium. In other words, he will prefer the I which will appreciate more across his monetary time window...Rhodium emerges as our standard, its price reflects its value as money on top of its value as an industrial commodity, and the price of palladium goes back to its mere value as an industrial commodity. Given the vagaries of markets, our blue-eyed MBAs would have bought palladium too high (anticipating the possible increase in its value as money if it would have become a monetary standard) and sold too low (as glut and bust will follow this monetary bubble), while the brown-eyed colluders reap the full benefits of investing early in the money standard that actual emerged.
[Now consider a population of Svens, each choosing an intermediate commodity]. Let's separate this herd into two strategies, by eye color. If Svens have blue eyes, they follow their proper MBA reflexes and diversify, buying equally priced lots of palladium and rhodium. But if they have brown eyes, they buy only rhodium.
Who does better? The brown-eyed Svens. Why? Because [the introduction of a commodity that can't be costlessly stored [but as I observed above this is not really necessary; what we need to introduce are coincidences of wants and mental transaction costs -- NS]] has created new demand for both palladium and rhodium. There was no monetary demand before we broke the equilibrium - now there is. Ceteris paribus, the price must go up.
But if we take this analysis further, our blue-eyed MBAs don't come off nearly as bad as Moldbug's scenario suggests. If mental transaction costs, in addition to storage and transport costs, are sufficiently low there is no convergence on a single currency, because in a world of sufficiently unpredictable and volatile prices and risk aversion a party indeed benefits from stockpiling multiple currencies, just as they teach MBAs about investments.
In Moldbug's competition of blue eyes versus brown, the dice were loaded: the agreement between brown-eyed Svens to standardize on rhodium, and the lack of any attempt to set up a competing standard, allowed parties who knew about the agreement to predict ahead of time which standard would win. If the outcome is significantly predictable, it pays to invest completely in the most likely winner. But let's add a third group: green-eyed Svens that use just palladium as their intermediate commodity. Green and brown eyes being of the same expected financial size (or of a completely unpredictable financial size), there is a 50% chance that palladium and green eyes will win, while brown eyes lose everything (except the original non-monetary value of the commodity, presumably negligible), and 50% chance of the reverse. For the blue eyes, if they diversified evenly it's basically a wash. The risk-neutral expected value of all three groups is the same, but if our players are risk-averse our blue-eyed MBAs have the strategy of highest expected value. 50/50 diversity is thus the optimal initial position when it cannot be predicted which of two commodities will gain value as money. But since it's impossible to discover a perfect 50/50 diversification, and mental transaction costs are sufficiently high, the equilibrium is unstable and will coverge to a single currency. Once one commodity starts to be favored, the optimal strategy is to move to that currency. So we have shown that sufficiently high mental transaction costs are sufficient to cause the emergence of a currency standard, even in the absence of storage and transport costs.
Reminder: we are neglectling coercive means such as legal tender laws and operating in a completely voluntary market. But Moldbug's scenario also explains why larger governments usually end up controlling the currency, even in the absence of legal tender laws. Markets will tend to standardize on whatever the dominant transactor, the party that controls the largest plurality of cash flow, standardizes on, and in most historical societies the dominant transactions were tax collection and the payment of those taxes to soldiers. More recently government bonds and even more recently welfare payments have joined the fray, biasing the outcome still further. By standardizing on its own currency, a large government can gain revenue from an "inflation tax." This process is far easier for the government with a fiat currency than a currency based on natural unforgeable costliness (such as a precious metal). Before the advent of modern currencies, inflation and the resulting revenue could only be obtained tediously via the slow substitution of less scarce for more scarce metals in the government coinage. With paper the physical process is trivial, and only the matter of how the new money enters the economy is at issue.
All the foregoing assumed sufficiently high mental transaction costs. This has been the historical norm, because trying to shop or otherwise do business in a world of multiple currencies, much less of pure barter, has always led to confusion, error, and overly complex accounting, and would do so even given the costless teleportation of Nitropia. But with sufficiently low mental transaction costs and sufficiently unpredictable exchange rates, it pays to hang on to multiple currencies, and a world of multiple currencies is the equilibrium. At the extreme of zero mental transaction costs, zero storage costs, and zero transport costs, we have a pure barter market, with no need for money at all.
Now for a more radical claim: in some cases, computers can drastically reduce the mental transaction costs of comparing prices in multiple currencies, which along with the "costless teleportation" of online markets allows multiple currencies or in some cases even barter to become the equilibrium. I'm quite a bit more fuzzy on just what those circumstances are, or just what software with what user interfaces said computers must be running, but you can see some of my ideas here and here. The general idea is that most of the mental costs of mapping of preferences to budgets via prices, in order to make buying or selling decisions, are offloaded onto a software agent, via a user interferface and a complier that translates high-level preferences to detailed "binary" contracts.