These Three Graphs Prove That Bitcoin Is a Speculative Bubble
Bitcoin is a speculative bubble, and I have the graphs to prove it. Surprisingly, I just need three of them.
When compared to the regular economy, bitcoin’s macroeconomic figures are vastly more accessible and reliable: all bitcoin transactions are publicly registered as to quantity and time, and that’s a standard that the U.S. dollar can’t hope to meet. In bitcoin, all your transactions are anonymous, but they’re also made in public. And that lets us do a good deal of macroeconomic analysis without the hedging that it usually requires.
My first graph shows the daily number of bitcoin transactions per U.S. dollar of bitcoin’s total market capitalization:
This chart shows a dramatic reduction in the total number of transactions, irrespective of size, per dollar of bitcoin’s market cap, from December 2012 – December 2013. In absolute terms, market cap has generally gone up, and the number of transactions has mostly just bounced around a lot. The total value of bitcoin is going up, but it’s mostly getting parked rather than being put to work. Apparently there just aren’t a lot of appealing ways to spend bitcoin, anecdotal news stories to the contrary notwithstanding.
Instead, an increasing amount of bitcoin’s putative value (as measured in USD) is being squirreled away by larger and larger miner-investors. It’s not fueling a diversifying, all-bitcoin economy: if it were, transactions would be keeping up with or even outpacing market cap, particularly if bitcoiners came to rely increasingly on bitcoins and decreasingly on dollars for day-to-day purchases. That’s very clearly not happening.
Instead, people are mining additional bitcoin, and speculators are buying in – and thus both of them are growing bitcoin’s notional market cap – but these folks simply aren’t adding all that much to the number of daily transactions. Even a bitcoin mining pool would only disburse its proceeds once, and if the individuals in the pool mostly just held their bitcoin, you’d get a graph a lot like the one above.
This is not a growing economy. It’s a hope for a growing economy — despite strong evidence to the contrary. And there’s a word for that.
My second graph shows the daily total transaction value of the bitcoin economy, denominated in U.S. dollars, divided by the total market capitalization of the bitcoin economy on that day, again denominated in U.S. dollars:
This is an attempt to look at the velocity of bitcoin. The velocity of money is an important number in macroeconomics, at least in the real world. Allowing for some expected (and expectedly large) daily fluctuations, the velocity of the bitcoin economy has stayed fairly stable. It doesn’t seem to be trending upward or downward over the last year.
But how fast is the bitcoin economy, really, and how does it compare to the U.S. dollar?
In the standard procedure for finding the velocity of money, we take the total transaction value for the time period and divide it by the average amount of money in circulation.
Deep breath: The average number of bitcoins in circulation during the year was 11,275,182. And the total value of bitcoin transactions, denominated in bitcoin, was 76,390,490. Using figures given in bitcoin, the velocity of money was thus 6.775.
That’s actually fairly close to the U.S. dollar’s M1 velocity over historical timescales. It’s also fairly close to the figure for last year.
But there are at least couple of very serious methodological problems here.
First, whether we like it or not, we need to reckon the velocity of the bitcoin economy in U.S. dollars and not in bitcoins. That’s because a great deal of evidence suggests that bitcoin is being used as a method of payment, but not as a unit of account.
No one ever keeps their books in bitcoins. Vendors can make bitcoin available as a way to pay for things, but they always, always, always measure bitcoin’s value in dollars, and they adjust bitcoin-denominated prices continuously so that the sticker price is the same in U.S. dollar equivalents. Bitcoins are simply too volatile; their price fluctuates too wildly to do it any other way. Shopkeepers don’t usually care to be currency speculators.
So we need to recalculate: Bitcoin’s dollar-denominated total transaction value for the year was $13,029,440,205.00. And bitcoin’s dollar-denominated average market capitalization for the year was $1,866,267,446.79. That yields a velocity of 6.98.
That’s again very close to the estimated velocity of money of the U.S. dollar.
But now we come to the second methodological difficulty: The above similarities are probably an illusion. The entire bitcoin money supply consists of something a lot like M1-type money: that is, it’s essentially all either circulating cash or (effectively) demand deposits, like checking accounts. But the entire dollar money supply is much larger, and those other parts have their own velocities and particular properties, complicating the picture in ways that I don’t quite feel anyone should speculate on. Bitcoin has security and privacy features that differ from M1, and bitcoin M2 will be another creature entirely, if and when it exists.
The key here is that nothing seems to be happening all that dramatically in bitcoin’s velocity of money over time. It’s not circulating more rapidly over time, which is what one should expect if it were taking off as a currency, and if more and more transactions were of the form of people passing bitcoins around for stuff. Instead, most transactions (that is, most that don’t go dollar-to-bitcoin-and-then-stop) are likely to be money-to-bitcoin-to-stuff, after which the merchant reverts to the dollar as soon as possible. If the bitcoin economy were becoming independent, we might expect a takeoff in the velocity of money, but we’re definitely not seeing it yet.
And now we’re ready for the third graph, which is the real smoking gun. The blue line is the average value of all bitcoin transactions for the day, in dollars. The orange line is the dollar-denominated price of one bitcoin multiplied by five:
The conclusion is clear: the mode bitcoin transaction likely consists of a fairly rich American using disposable wealth to buy an arbitrarily chosen quantity of bitcoin for speculative purposes.
In layman’s terms, the typical bitcoin buyer likely thinks as follows:
“I’d like to get me some bitcoin. And I think… oh… maybe five of them will do. Now — how much do they cost again?”
How much does bitcoin cost? These folks don’t even care – they just want some. Thus we can conclude:
The mode bitcoin is probably mined, disbursed, and never goes anywhere thereafter. The mode transaction is someone buying an arbitrarily chosen amount of bitcoin and then sitting on it forever. Consumers using bitcoin to buy stuff (other than dollars) appear to be few and far between. Bitcoins circulating without immediate reconversion to the dollar are likely very few. And all of this has been true for at least a year.
This trend cannot continue, because disposable income – and enthusiasm for a fad – are both finite. The value of one bitcoin can continue to explain bitcoin’s average transaction value for only so long, and situations in which current face values of an asset determine how much people will buy are not just evidence of a speculative bubble. They are the very definition of a speculative bubble.
My work is available on request, and I will be happy to take questions. I am an economics autodidact, and I freely admit that macro isn’t my strong suit. As a result, I’ve probably made some mistakes, and I welcome corrections. Lastly, I have to say that all of this seems clearly true to me — so far. Tomorrow could easily prove me wrong, but I’m pretty sure it would need new data to do so.
Update: Commenter Nick asked for an x y scatterplot of the correlation between bitcoin’s price and the average daily size of bitcoin transactions. That’s a fair request, and here it is, with logarithmic scales on both axes to preserve the movements at the lower price ranges. The correlation coefficient is .92, which is pretty darn high: