Bitcoin, The Regression Theorem, and That Curious But Unthreatening Empirical World
It must happen this way. Nobody can ever succeed in constructing a hypothetical case in which things were to occur in a different way.
—Ludwig von Mises on the regression theorem
Some observers of the Bitcoin phenomenon have been puzzled that it does not appear to conform to the monetary regression theorem, an explanation for the initial origin of the value of money as money. This has led to various claims and counterclaims. In essence, one camp claims that bitcoins are not “really” money (and will probably fall apart any time), while the other counters that bitcoins actually are here, and most likely here to stay, adding that if anything is broken, maybe it is these stodgy theorems, which we might have to revise.
In addressing this, I will attempt to account for the emergence of bitcoins in terms of the monetary regression theorem. In doing so, I will argue that 1) the existence of bitcoins does not and could not challenge the regression theorem and 2) the regression theorem does not constitute any particular problem for bitcoins in terms of economic theory. That said, 3) the investment analysis of bitcoins is a separate matter from the economic-theory analysis and is a good (but separate) topic for vigorous debate.
One early approach to this, for example, was, “Bitcoin does NOT violate Mises’ Regression Theorem” (27 July 2010) which argued that bitcoin values trace back to an initial translation with fiat currencies through early bitcoin exchanges. This explains how one currency can launch on top of another one once a conversion rate emerges or is set by fiat as in, say, the German mark or French franc to euro transition (btw, some people in Germany still back-translate euro to deutschmark prices—and are suitably horrified). Yet mere translation does not show any initial direct-use or direct-exchange value separable from expected indirect-exchange value, so this brings us right back to the regression-theorem/bitcoin paradox where we started.
While this fiat-currency exchangeability factor played, and is playing, a large role in the spread and growth of bitcoin use, the approach below considers direct-consumption values separable from the beginning of fiat-currency translation or other sources of rising market expectations of indirect-exchange value. An example of a recent attempt that started off nearer to the approach below was “Bitcoin and the Regression Theorem of Money” (7 December 2012). However, this explanation still mainly referenced speculative demand relative to existing fiat currencies and as such remained less than completely satisfying vis-à-vis the regression theorem (nevertheless, the author appeared to realize this. Several comments under that post were more in line with elements of my approach here, and I quote one of them below).
Are bitcoins “money” or not?
We begin with the relatively simple issue of whether or not bitcoins are “money.” Murray Rothbard (MES 2004, 192–93) can help us get started (we will discuss this use of the word “commodity” later):
A commodity that comes into general use as a medium of exchange is defined as being a money. It is evident that, whereas the concept of a “medium of exchange” is a precise one, and indirect exchange can be distinctly separated from direct exchange, the concept of “money” is a less precise one. The point at which a medium of exchange comes into “common” or “general” use is not strictly definable, and whether or not a medium is a money can be decided only by historical inquiry and the judgment of the historian.
If we say that money is a medium of exchange, then bitcoins are money. If we say that money is a “widely accepted” medium of exchange, then we have an interpretive and empirical argument about what “widely accepted” means and whether and at what point bitcoins might qualify, and for that matter, where. For example, if we say that money is “the most” widely accepted medium of exchange, bitcoins would fail to qualify, but then so would Mexican pesos, beaten out by dollars, euros, and others. So then we have to add, “in a given geographic area defined by a local monetary monopolist’s borders” to make our peso collection into “money” again.
The task of economics is to help explain phenomena of the market. In this sense, if we see bitcoins functioning as a medium of exchange, all our theorems cannot make them go away and cease doing so. Even if they were to collapse next week, they are still a medium of exchange now. People are acquiring them in order to save them or sell them or buy things with them or donate them to bloggers. Our job as economic theorists is to figure out how, in what sense, and to what degree, this is happening, not whether it is.
Besides the “widely accepted” element remaining debatable (literally; cases could be made either way), I might not call bitcoins money just yet because my impression is that many users still view them primarily through the lens of relationship to exchange rates against their local fiat money. Even Bitmit, a dedicated bitcoin-denominated auction site, automatically and helpfully translates all bitcoin prices into the local fiat currency instantly customized for visitor location after the bitcoin price.
In contrast, if significant pricing, planning, and economic calculation began to be conducted directly in bitcoin-denominated terms without reference back to fiat-money prices, and if some notable number of people began to earn income denominated in bitcoins, these could serve as interpretive landmarks for the “money” question. Interestingly, Jeffrey Tucker recently reported some signs of early stages of just such expansion in “Top Alternatives to Paper Money” (25 February 2013).
In conclusion, one could make a case for calling bitcoins “money,” or not doing so, but they clearly do qualify under the more precise concept of a medium of exchange.
Bitcoins? Regression theorem? No problem
The regression theorem explains how a good can gain an initial value as a medium of exchange. The theorem states that to do so, the good must have at a point prior to acquiring medium-of-exchange value some other value as a good in itself, a direct-use value and then also an actual or potential direct-exchange (“barter”) value. Yet many observers have found themselves challenged to find any “direct-use value” in bitcoins. These objects appear to have value only as a medium of exchange.
A few points should help clear up the confusion. I will first state these briefly, then explore each one and its implications in greater depth. First, the regression theorem is a praxeological statement. It does not admit of any qualifications in the form of value judgments on the part of observing economists. Therefore, the only challenge for the regression theorem/bitcoin relationship is to find any direct-use or direct-exchange value prior to and separable from the emergence of indirect-exchange value. Second, it is helpful to clarify and define what bitcoins are in economic theory terms as types of objects of human action. Third, the regression theorem specifically explains how a good first acquires value as a medium of (indirect)-exchange. This raises the question of whether bitcoins were ever used in direct exchange before also acquiring a medium-of-exchange use.
Not whether, how
As a praxeological statement, the monetary regression theorem is not threatened at all by the existence of bitcoins, nor are they threatened by it; the two merely gaze across the intellectual landscape at one another with knowing smiles. If we understand the regression theorem clearly, we already know that there must have been some direct-use and direct-exchange values, because 1) having them is a prerequisite for becoming a medium of exchange and 2) bitcoins are a medium of exchange.
Our challenge, then, is not to “test” our regression theorem as if it were a hypothetical “theory,” but rather to stretch our interpretive capabilities to the demands of the empirical case at hand. Our question as economists is not, “Was there a prior direct-use value?” but rather “What was it?”
As is typical, Mises already carefully addressed this specific relationship in 1949 (Human Action, p. 407 in 1998 edition), well before bitcoins, the internet, or even personal computers:
To explain an event historically means to show how it was produced by forces and factors operating at a definite date and a definite place…To explain a phenomenon theoretically means to trace back its appearance to the operation of general rules…The regression theorem complies with this requirement…It deduces a more special case from the rules of a more universal theory. It shows how the special phenomenon necessarily emerges out of the operation of the rules generally valid for all phenomena.
It does not say: This happened at that time and at that place. It says: This always happens when the conditions appear; whenever a good which has not been demanded previously for the employment as a medium of exchange begins to be demanded for this employment, the same effects must appear again; no good can be employed for the function of a medium of exchange which at the very beginning of its use for this purpose did not have exchange value on account of other employments. And all these statements implied in the regression theorem are enounced apodictically as implied in the apriorism of praxeology. It must happen this way. Nobody can ever succeed in constructing a hypothetical case in which things were to occur in a different way.
Thus, even if every single interpreter, including myself writing now, were to end up failing to find any prior direct-use or direct-exchange values, we would still know that bitcoins had had one. All that we would establish by not finding one would be the failure of our own interpretive efforts. That said, I will now proceed with my own attempt at such an interpretation along several lines.
First, one element to consider for intangible objects such as bitcoins are various “inherent” direct-consumption values that may be primarily psychological or sociological in character. Consider, for example, the geek value hackers find in creating and attempting to crack encryption codes of any kind: “Dude, look at this code; I bet you can’t crack it,” may indeed be more highly valued to some people in some contexts than certain “real” economic objects or specific quantities of fiat money. Regardless of any potential future indirect-exchange value, one can imagine such persons expending hundreds of hours of effort in creating and breaking encryption codes, just because they like to. This may be true, separate from any degree of dependence on any particular expectations of future exchange values of code objects.
The value of any such virtual object or system would then tend to rise, still entirely within this context of direct valuation, if it began to prove robust. It need have no other characteristic or uses whatsoever to still be valued as such among those persons who understand and value such things. We will discuss some further context for this claim and some other such psychological or social direct-consumption values further as we go along.
In an insightful comment under the post “Bitcoin and the Regression Theorem of Money,” mentioned above, “QBG,” writing on 8 December 2012, captured what I was thinking on this point nicely in relation to gold, an angle I had not thought of:
What good was gold thousands of years ago?…The only significant original direct use of gold was making pretty things. The original adopters of Bitcoin were the type of people (as a result of being interested in crypto-currencies) to see Bitcoin as something beautiful. As a result, Bitcoins had a small intrinsic value to them, and so it had its place on the value scales of those individuals. A system of barter was then able to emerge as people with different valuations traded Bitcoins for other goods.
Even now, well after their initial emergence, there appears to be a “mystique value” and a “curiosity value” attached to bitcoins among widening circles of newcomers who, compared with founders and earlier adopters, tend to understand the underlying mechanics of the system less and less, but have the impression that participation is a way to be proud and to send a message of being techno-savvy, up to date, in the know, etc.
In other words, mere possession, knowledge, and use can carry social membership signaling functions in various sub-cultures, much as wearing certain styles of clothing does. These are also direct-consumption values to those concerned with such signaling. Direct-use values, whether psychological or sociological, do not have to be recognized by anyone other than those in a given sub-culture actually doing the valuing (according to methodological individualism and subjective value).
Scarcity, not commodity, is the name of the game
This has several broader implications for the clear understanding of economic theory in our contemporary context. Goods do not have to be tangible to inspire the relevant types of human action; they only have to be scarce (a nice walk-through on this distinction is “Goods, scarce and non-scarce” by Kinsella and Tucker, 25 August 2010). A bitcoin is a scarce, intangible good. Being scarce, it already satisfies the minimum requirements of becoming an object of action, without also having to be tangible.
Some people may have become confused by the use of the word “commodity” in the treatment of monetary theory inAustrian economics classics. However, the quality of being a “commodity” as we usually think of one does not appear especially important in light of contemporary developments. It may be that these writers, thinking mainly of historical examples such as gold and silver, used this word as a result of having only such examples in mind, along with auxiliary examples such as tobacco, sugar, salt, cattle, nails, copper, beads, tea, and cowrie shells, to take Rothbard’s list from MES, p. 192. The distinction between scarcity and tangibility has since become clearer and clearer with the spread of digital goods.
But there is no a priori reason why any “good” or even “thing,” whatever it may be, cannot serve as a medium of exchange, provided it is scarce and satisfies other key monetary characteristics such as durability (not physical, but temporal!), divisibility, and fungibility (interchangability). The quality of tangibility was also characteristic of historical monetary-emergence patterns, but it is scarcity rather than tangibility that is the essential quality for a good. This should have become especially clear at the very point that we observed some intangible, scarce object actually serving as a medium of exchange!
If we do not want to call bitcoins a “commodity,” the solution is not to begin pretending that bitcoins are not “really” a medium of exchange. Instead, the solution is to select a word that can better encompass all those types of objects that actually serve, have served, or could ever serve as a medium of exchange. “Good” should serve in general. In this case, bitcoins are a “scarce, intangible, digital” good in particular.
In working within a science of human action, the stress is not on things, but on the structure of actions, on intents, means, and purposes. What counts is not what a thing is in objective or objectivistic terms, but what it is that we observe people doing with it, interpreted in terms of action.
To help contextualize this, witness the millions of gamers around the world expending tens, hundreds, and thousands of hours each of labor, planning, and at times socially coordinated effort, to obtain in-game cash and virtual objects (and sometimes glory), even though these have no or little indirect exchange value outside the particular game (with some exceptions, as we will see). The game Runescape is one prominent example that will serve. This game has “over 200 million accounts created and is recognized by the Guinness World Records as the world’s largest free MMORPG and the most updated game,” according to Wikipedia.
In this particular game world, available means of obtaining objects and in-game cash (called, appropriately enough, coins or gold pieces), include: original appropriation (finding things and picking them up); production of intermediate and final goods using various tools, machines, raw materials, and lesser intermediate goods; direct and indirect exchange with other players and non-player characters (“NPCs”); market speculation and price arbitrage; lending at interest; pickpocketing; and combat (the last two also being normatively acceptable within the context of the game). Some items are bartered, sold, or given away informally with other players; others are sold in public markets with listed market price trends.
Virtual in-game cash even has a tendency to spill outside of such game environments into real cash markets, sometimes against the rules of the game itself, sometimes systematically, as in the real-world activity called “gold farming.” Neal Stephenson’s 2011 novel Reamde plays with the fictional in-game cash dynamic by setting up a [moderate spoiler alert:] fictional MMORPG in which gold farming is NOT against the rules and the game interfaces explicitly with the “real” economy. It is basically Chinese youth mining in-game gold and selling it to richer Westerners who don’t want to do the in-game hard work, but decide they have more fiat cash than in-game cash to burn.
Bitcoins do have a wider element of future indirect-exchange expectations than, say, an in-game currency limited to within the game, but this may be in addition to other value components. It does not have to exclude the parallel presence of more intrinsic direct-use valuations, particularly among early actors. The creators built a new class of scarce intangible goods. In doing so, there was some degree of direct value to the relevant hacker-actors just for the sake of creating them as a new type of secure, scarce virtual object. Soon, there emerged an additional curiosity value among others in acquiring these objects, studying them, challenging their security, etc.
The particular security value of these digital objects sets up something similar to the kinds of in-game cash that gamers are already accustomed to, but makes it suitable to direct uses as a new “real-world” cash. Still, this new real-world cash functions almost entirely within a structural context of secure (more or less) computer-mediated communication that is remarkably similar on the user side to that of MMORPGs with cash economies: get online, log in, and conduct transactions among user accounts.
A bit of bartering?
There is also no reason that some initial transactions of bitcoins may not themselves have had characteristics of barter rather than indirect exchange. “I’ll give you two slices of pizza for a bitcoin” just because I want to have a bitcoin, is a barter transaction, not an indirect-exchange transaction. And indeed, it appears that perhaps the first “real” bitcoin transaction was a somewhat legendary swap of a large block of coins for a pizza. According to the Bitcoin Wiki’s history entry for 21 May 2010, “Laszlo first to buy pizza with Bitcoins agreeing upon paying 10,000 BTC [currently exchangeable for about $310,000] for ~$25 worth of pizza courtesy of jercos.”
I might also want a bitcoin for any reason I feel like having one. I might want to just study it and see how it works or collect it as a virtual souvenir or trophy. I might want to use some of its code string as T-shirt art. I might want to stay up nights trying to crack the system because it’s there, like the proverbial unclimbed mountain. I may just want to feel cool and smart by having a bitcoin and telling friends about it. None of these purposes constitutes an indirect-exchange purpose. These are all direct uses.
But this is already somewhat more than needs to be shown. The regression theorem concerns the emergence of indirect exchange characteristics on top of previous direct-use and direct-exchange characteristics. Yet the only one who knows the difference is the person using it. The only way to find out whether little Timmy just bought a bitcoin because he thought it was cool to have one or because he thought he could later buy other stuff with it is to ask little Timmy, and then we are still not sure if he is telling the truth. These are empirical questions.
Nor does any direct-use value have to persist once indirect-exchange value has emerged. The regression theorem is only a temporal-sequential explanation of the initial emergence of indirect-exchange value. After that, the initial direct-use value is no longer required: the emergence of indirect-exchange value has both already happened and already been explained.
All that is required for a transition from direct to indirect exchange is an increasing number of people wanting to have a good just because they want it—for any reason, which in turn can give rise to some people realizing that they might want to obtain that good because they know of other people who want it—also for any reason. Actors therefore begin to expect that indirect exchanges will start becoming more widely possible with this good. The indirect exchange component thereby begins to grow relative to the direct-use and direct-exchange components. Even if the value components other than indirect-exchange value have fallen away completely (which I don’t think they have, although risks are present, as we soon discuss), this does not impact the regression theorem, which has already done its work right at the beginning and is now free to return to its Viennese study and resume smoking its pipe undisturbed.
The foregoing may still not seem to some observers like much in the way of direct-use value components for bitcoins, but they are something, and something is all that is needed to interpret the case in terms of the regression theorem. The economist’s own value judgments should not play into assessing whether objects are “really” valuable or not. The actions of those persons actually acquiring and using these objects in these ways are already enough to tell us otherwise. Whatever the observed actors were thinking, no matter how crazy it may seem to some observers, economists just have to deal with it.
A house of virtual cards?
Admittedly, the expected indirect-exchange-value component now appears to reflect a very large share of the value of bitcoins. The other value components we have sought to identify above, to the extent they persist, also appear highly dependent on this component. A hypothetical total evaporation of the exchange value (through some successful crackdown or “impossible” system failure) would also diminish most of the aforementioned geek coolness, curiosity, collector, and social signaling values right along with it. The scale and independence of the direct-use values are tenuous and apparently tied tightly to the indirect-exchange value. This may be why some observers have missed these components altogether or just do not trust them. If indirect-exchange value is really all that remains, a crash could well indeed take the money down to zero exchange value and extinction as a money.
Such lack of trust might arguably be wise from a historical, ethical, and financial planning point of view, and this is a reasonable subject for vigorous debate. However, this is not a problem for the praxeological regression theorem. It does not concern itself with such forward-looking ethical evaluations based on a particular claim to wise judgment one way or the other.
I am not saying the values of bitcoins outside of their expected indirect-exchange values have ever been large or stable, only that such valuations were, have been, and still are present among the relevant actors to varying degrees. All one needs to show to erase the alleged regression-theorem/bitcoin paradox is that direct-use and direct-exchange values were present: 1) at all (no degree-of-presence judgment is relevant); 2) at the very beginning (not needed later), and 3) within the value scales of the actual persons involved in creating and dealing with the objects early on (not within the value scales of later users or later economists).
Based on the above considerations, it appears that the regression theorem and bitcoins need have no quarrel with one another. Just because the direct-consumption value components were (and are) psychological or sociological in the sense of pertaining to factors such as inherent geek appeal, professional challenge to specialists, curiosity, and membership signaling, does not mean they were therefore non-existent. And existent was all they ever had to be.
(Originally published at Konrad S. Graf: Investigations and Observations)