Athey et al. (2016): Bitcoin Pricing, Adoption, and Usage: Theory and Evidence | Article Review
This is a non-technical review of the paper "Bitcoin Pricing, Adoption, and Usage: Theory and Evidence" by Susan Athey, Ivo Parashkevov, Vishnu Sarukkai, and Jing Xia. The paper is a working paper available on SSRN, but to my knowledge it hasn't yet been published in a peer-reviewed academic journal. (SSRN is a distribution network and not a peer-reviewed journal.) However, Susan Athey is a well known economist from Stanford with serious technical chops, and her work is considered highly credible.
Overview
This is a "theory and evidence" type paper with the goal of understanding what gives Bitcoin "value", and what drives the exchange rate between Bitcoin and fiat (e.g. the "price.") The authors first develop a very simple model with two main forces:
- The force which motivates Bitcoin adoption is that it can be used for cross-border payments more cheaply than the traditional financial system.
- The force which motivates against Bitcoin adoption is the risk of network failure. You can think of this as a major problem being discovered in the protocol, an effective ban by the government, or any other external event which would kill the use of the network.
Expectations about the future exchange rate (e.g. price) is also a motivating factor in adoption. But the exchange rate is "endogenous"--meaning it's determined by the model. The fundamental forces are Bitcoin's utility in cross-border payments and the risk of technological failure.
The main result of the paper is that the steady state (i.e. long run) price of Bitcoin is determined by the ratio of its transaction volume to its supply. In other words, if long-run usage is higher then long-run price will be higher, which is not surprising. The price is also expected to increase over time as beliefs about the viability of the technology evolve. This seems consistent with the historical record.
It should be noted that the long-run price dynamics in this paper are not driven by a decline in the real worth of fiat. It is assumed in the paper that the real value of fiat remains constant. Thus, if we assume that the real worth of fiat goes down over time we should expect an even faster growth rate of the Bitcoin price than the adoption rate described by this paper's model.
After describing the model, the paper goes on to do two empirical exercises (i.e. data analyses):
- Is the Bitcoin price correlated with transaction volume as the model predicts?
- What share of Bitcoin activity appears to be from investors/speculators, what share from illicit activity, and what share from international payments?
Unsurprisingly, they find that the Bitcoin price is correlated with transaction volumes, except in brief periods of extreme price action. Also unsurprisingly, they find that most Bitcoin owners are hodlers who do not spend their Bitcoin. They thus conclude (accurately, in my opinion) that the primary use-case for Bitcoin currently still appears to be as a store of value / investment vehicle. However, they do show evidence that Bitcoin is being used for cross-border payments, and they also show that illicit activity is only a small fraction of the network activity.
Do the authors understand Bitcoin?
One of the first questions Bitcoiners have in their heads when they read an article about Bitcoin is whether or not the author seems to understand Bitcoin. I think these authors do. Here are some quotes from the article that show a solid understanding:
Even defining Bitcoin is complex: it can be described as a protocol, a currency, a payment system, and a technology platform.
--
Unlike a bank balance that can be viewed or manipulated digitally, an individual's Bitcoin is not an IOU or a promise to provide funds on demand; the individual with the passcode associated with an address has full control over its disposition; and that Bitcoin balance is not linked to anything else.
--
One individual can transfer Bitcoins to another anywhere in the world, without relying on counterparties or trust relationships beyond the trust in the software, and indeed without getting authorization to do so from any company or government. All that is ncessary to transfer value elctronically is the passcode to authorize a ledger entry to move the Bitcoins to another address.
--
On the other hand, if the recipient wants to obtain value from the Bitcoins in the short term, she must either find a merchant who will accept them, find an individual buyer for the Bitcoins, or sell them on an exchange. This creates frictions and risks in using Bitcoin.
--
[Studying individual behavior on Bitcoin] is posible because the ledger is public, although Bitcoin is "pseudonymous," meaning that identity information is not attached to addresses, addresses can be followed over time. However, since individuals typically have multiple addresses (due to some details of how addresses are managed, in "wallets"), it is not trivial to identify which collections of addresses are individual users or Bitcoin businesses.
Overall, I'd say the authors have a pretty good understanding of Bitcoin.
Main takeaways for Bitcoiners
This article provides some useful takeaways for Bitcoiners.
-
First, they could only detect 1% of the transactions as being related to contraband or other illicit activity. Assuming their methodology is reasonably accurate (which I don't have the bandwidth to check right now), this is a good counter to the argument that Bitcoin is primarily used by criminals.
-
Second, they show that (as of 2016), the share of the network being used for payments isn't growing over time. This suggests that in 2016, Bitcoin was still in a phase where price dynamics are driven primarily by evolving beliefs, investor sentiment, and speculators, and less so by usage as a medium of exchange (MOE). That's still likely the case today, I'd wager.
-
Third, they showed evidence that Bitcoin is being used for cross-border payments. However, I found this part fairly hard to understand. As a referee, I would have asked them to explain it more clearly.
-
Finally, they came with an algorithm to try and detect business users, illicit users, illicit payments, and cross-border payments on-chain. I'm sure there's much to criticize here, but people interested in chain analytics may be interested in checking out their approach.
Keep in mind that this article was written in 2016. Much has changed since then, including the growth of the Lightning Network. I suspect that if they were to repeat their analysis today, including both on-chain and lightning transactions, they sould find that Bitcoin is being used more often now for actual e-commerce (though still likely a minority use case.)
Concluding thoughts
I liked the paper. I think it gave a fair and accurate description of Bitcoin. I think the paper was correct to highlight Bitcoin's utility in cross-border payments as a primary source of its value. One could also envision a model in which government confiscation or money printing is a risk, which would create another source of Bitcoin value. The empirical work is interesting, though one would have to believe in their chain analysis algorithm to truly believe the results.
Whether from these authors or someone else, I'd love to see follow up work looking at the Lightning Network has changed some of these adoption metrics and use cases.
Some interesting charts
Below I present some interesting charts from the paper. I present them without comment, as you'll have to read the paper to really understand how to interpret them.