In three previous posts, I have described the regression theorem, discussed its practical applications, and considered some misconceptions. In this post, I will consider the regression theorem in light of bitcoin.

The discussion around bitcoin and the regression theorem usually focuses on whether bitcoin has some intrinsic worth (read: non-monetary use). If bitcoin is intrinsically worthless, the regression theorem is invalid: an object might become money despite lacking a non-monetary use at the outset. If bitcoin had some non-monetary use at the outset, the regression theorem remains intact.

In a new working paper, I argue that bitcoin calls into question the *practical relevance* of the regression theorem *regardless of what one thinks about its intrinsic worth*. Recall from my earlier post that the practical relevance of the regression theorem is in distinguishing which items might emerge as money without government support and offering suggestions as to how the government might launch a money that could not emerge naturally. Here’s my argument in brief:

- If bitcoin is intrinsically worthless, the regression theorem is invalid. If the regression theorem is invalid, it has no practical relevance.
- If bitcoin has some non-monetary use, that use must be pretty trivial. If trivial uses suffice, the regression theorem has little if any practical relevance.

∴ Bitcoin calls into question the practical relevance of the regression theorem regardless of what one thinks about its intrinsic worth. Since (1) is widely understood, I will only elaborate on (2).

Those who claim bitcoin had some non-monetary use at the outset typically point to uses that are psychological or sociological in nature. Perhaps bitcoin appealed to those with anarcho-capitalists or futurist sympathies, who might signal their respective views by holding bitcoin, even if it would never be employed as a medium of exchange. Perhaps it appealed to those with a theoretical or scientific interest in money or cryptography or those appreciating a challenging programming problem, who might use bitcoin in testing the network. Perhaps it was a collectible. All of these non-monetary uses seem plausible. But they also relax the constraint the regression theorem places on would-be monies.

As I write in the working paper:

…maintaining that bitcoin has some non-monetary use preserves the validity of the regression theorem while simultaneously relaxing the constraint the regression theorem imposes on the set of potential monies. The inclusion of psychological or sociological uses—while perfectly reasonable—means the divide between fiat and commodity monies is not so large. Many items that would have been ruled out by a traditional understanding of the regression theorem suddenly become viable candidates to emerge without sovereign support. As Surda (2014, p. 9) puts it, “the threshold for the emergence of liquidity for goods like Bitcoin is relatively low.” Even pieces of paper of a particular dimension and design, which seem to serve no other use aside from their potential role as a medium of exchange, can be said to have intrinsic value because some people derive pleasure from their aesthetic features. Nothing seems to be ruled out—and, therefore, nothing is obviously excluded by the regression theorem. As such, arguing that bitcoin has some non-monetary use makes the regression theorem far less important than those working in the Austrian tradition have claimed. Perhaps that is how it should be. As Graf (2013b, p. 16) notes, “[t]he sometimes-touted industrial and electronic uses of gold and silver are all quite modern and therefore entirely irrelevant to the first emergence of these metals in a monetary trading role in various places many centuries earlier.” Originally, these commodities were mere collectibles. Indeed, the emergence of most commodity monies seems to have begun with “a few of mankind’s ‘crazy ones’ […] playing around with and collecting things that were useless for anything that would have been considered a generally ‘practical’ purpose at the origin phases in question, such as shell beads, shiny metals, or bitcoins (Graf 2013b, p. 29).”

Thinking about the regression theorem in light of bitcoin has led me to revise my view on its significance. At this point, I think Mises provided an especially useful extension of the Mengerian account to explain how an intrinsically worthless item might get off the ground. However, I think he went too far in claiming his explanation was the only explanation. I now give more credit to the view that explicit coordination might be used to launch an intrinsically worthless item. Such a view is in line with standard models of money employed by economists today. Coordination also seems to have played a role in launching bitcoin.