Friday, May 26, 2017

The economist's case for at least agnosticism about Bitcoin

As far as I can tell, the primary problem with Bitcoin is that after you've bought it, you become medically incapable of shutting the @#$% up about Bitcoin. So it goes.

It took me a long time to buy any bitcoin, but I should have done it about three years ago. This isn't cheap talk, by the way. I know exactly why I should have bought it back then, based on the knowledge I had at the time (which is the only criterion by which you ought to regret any decision). To wit: I considered myself a Bitcoin agnostic. This made me more optimistic than perhaps 99% of finance people I spoke to. But then again, 99% of finance people I spoke to also couldn't easily explain why Bitcoin existed in the first place. 3 years later, all of the above is still true, but I finally got off my butt and did something about it, albeit after an enormously costly delay.

The standard economics textbook description of money says that it tends to arise because it helps facilitate exchange. If we need to barter goods with each other, it's hard for me as a blacksmith to obtain wheat unless I can find a wheat farmer who also coincidentally needs blacksmithing services. But if I can exchange my blacksmithing services for some asset money (as yet undefined as to what that is) and then turn the money into wheat down the line, this greatly allows us to trade more and grow the economy. So far so good. But what exactly makes something money?

The standard economics textbook definition of money says that it has to fulfill 3 purposes, namely
#1. It has to be a unit of account - a way of measuring how much of something you have
#2. It has to be a medium of exchange - a means for people to transact amongst each other and exchange goods and services indirectly, rather than directly through barter
#3. It has to be a store of value - that is, have some worth derived from an alternative use other than the monetary aspect itself, to ensure that people will be willing to hold it.

Under this standard narrative, bitcoin fails. #1 is fairly easy to meet, but bitcoin's big strength is in #2, which it passes with flying colours. Importantly, however, it fails badly on #3. Digital bits have no inherent value, and no external use to make them a store of value. Ergo, it should have no value above zero, and anything else is a bubble about to collapse. So goes the standard story.

Gold originally fulfilled all three purposes. You could weigh it, and trade accordingly. Turning gold into gold coins helped with 1 and 2, so drove out raw gold. It was easier to transact and measure in coins. Of course, the problem with coins is that they could get filed away at the edges to steal some of the gold, but still be worth approximately one coin. So eventually the coins got replaced with pieces of paper that were claims on gold in a government vault. At the start, you could actually make the conversion. Then conversion became increasingly a fiction, before FDR decided to do away with the pretense of convertibility, suspending the conversion and limiting the ability of private individuals to hold gold.

 At this point, you may be wondering how US dollars continue to meet the 3 definitions above. After all, they kept being used as money, and economists didn't all seem in a raging hurry to update their definitions. So the standard answer is that the 'store of value' aspect is that the government, who has guns, will accept USD as a means of paying taxes (and indeed demands that form). Because they have a guaranteed value for that, they have a guaranteed value for everything else.

To me, this seems to have a definite flavour of ex-post rationalisation. My hunch is that if you asked people 100 years ago whether they would still be equally willing to hold dollars if they were backed by nothing at all, they would have answered strongly in the negative. In the end, they were prohibited from switching into gold at the time, so it was a moot point. But what about now, when they could switch? I highly doubt that many people today would explicitly state that they're willing to hold dollars because they can pay their taxes in them. 100 years ago, they probably would have laughed you out of the room.

The economists are right in a narrow sense, of course (as they often are). Bitcoin does indeed fail as a store of value, and, technically, the dollar does not.

And yet, here is some evidence that ought give you pause, assuming you're not an economist in the midst of full-blown cognitive-dissonance-induced denial:

This is Bitcoin today, stubbornly refusing to prove economists right by ceasing to exist. As a matter of fact, since the coinbase time series starts in January 2013, it's up some 19,000% or so.

Don't look at the graph and ask if you think it's about to drop. Look at the graph and ask how much it would have to drop to get to where it was in 2013 (let alone 2009).

At a certain point, it seems prudent to at least consider the possibility that there might be something going on here, but you don't know what it is, do you Mr Jones? Or at a minimum, ask the following question - what level of future growth would you need to see to change your mind? Another 100%? Another 1000%? Can we agree on it now, so that if it eventually happens, you might reconsider the question?

The null hypothesis here is not in much doubt. Bitcoin is a bubble, and will eventually collapse.

Actually, the true null hypothesis is a little more specific, at least if you believe standard economics. Bitcoin should have a price of zero. It has no value except as a currency, and it is worthless as a currency.

So what is the alternative hypothesis?

The alternative hypothesis is that Bitcoin is likely to stay at a non-zero value for quite a long time, if not indefinitely, and moreover may end up being worth a lot. That may sound woolly and hand-wavy, so let me explain.

First off, how many things can you name that truly have a value of zero?

It's surprisingly hard. If you don't believe me, here's a photo of cans of air from the top of Mount Fuji selling for ¥500


And things like rubbish or nuclear waste have definitively negative prices - you have to pay to get rid of them. They're still not exactly zero.

The point is that "fundamental value" is a concept that, in my opinion, creates at least as much confusion as it dispels.

The primary value of an asset today is what you think someone will pay for it tomorrow. If they can use the asset for some external purpose, and you have a guide to what that external purpose is likely to be worth to them, you have a guide to what it will be trading at tomorrow. But that's all it is. If you have some other way of estimating what people will pay for the asset tomorrow, then you don't need the intermediate heuristic of fundamental value. (This is especially true for assets which don't directly produce cash flows - for ones that do, there's a better case that you should just value the cash flows, but even then, you still need to know tomorrow's willingness to pay unless you're able to hold the asset to infinity to collect all the future cash flows).

So in that case, what should Bitcoin be worth? Whatever people are willing to buy it for tomorrow. And what number is that? Well, that's the rub. But at least we know the right question to ask now.

As a consequence, we can begin to formulate an alternative definition of requirement #3 for money that we started with. Specifically:
#3A - If you accept the asset today in exchange for giving up valuable goods or services, you have to have a very strong belief that you will be able to exchange said asset tomorrow for someone else's goods and services, and receive approximately the same value as what you exchanged today.
Viewed from this angle, we can see that requirement #3A is at heart a co-ordination problem. Once we all agree on something being money, it becomes money. More importantly, we can see why people mistakenly viewed #3 as being the requirement. In essence, being a store of value is one way of solving the co-ordination problem. If it's common knowledge that some people will be willing to accept gold because it's useful for jewellery, most people who don't value it for jewellery are nonetheless willing to hold it.

But this isn't a strict requirement. Once the belief is established, it becomes self-fulfilling. When you accept US dollars, you aren't doing the iterations and thinking that it will eventually be exchangeable for taxes. You're just accepting it because you can buy your groceries with it tomorrow. Now, in the long run, it's true that if the US government collapses, you don't want to be holding US dollars, so in that sense the economists are right. But this is a long way from most people's actual calculation.

In the case of Bitcoin, a belief that Bitcoin will retain some value tomorrow can justifiably be sustained as long as I know that there's a decent number of drug dealers and corrupt Chinese officials who want to hold Bitcoin because it's (sort-of) anonymous and can be easily taken out of the country when the porridge hits the propeller. But in the short run, I hold Bitcoin because I think that people tomorrow will hold Bitcoin.

In fact, it's stronger than that. Like a classic bubble, people actually believe that more people will want to purchase bitcoin tomorrow, and at higher prices. In other words, the supply is fixed, and the more the price goes up, the more people begin to think "Huh, maybe I should hold at least a few grand worth of Bitcoin, just in case." If more people begin to think that, the price will indeed keep rising. Of course it can't rise like that forever.

But even if you think of Bitcoin as a bubble, it behooves you to notice something rather different about it from most bubbles, like the tech boom. In the case of Bitcoin, it seems to me from anecdotal experience that many, if not most, of the people buying bitcoin today are planning to hold it for a long time, if not forever. And this is definitively not true for most bubbles. People generally ride bubbles planning to get out once it's gone up enough, then go back to holding cash, or houses, or whatever. If that's what most people are thinking, the belief structure becomes very unstable, as any dip in price suddenly might cause a lot of people to switch to selling. Even if Bitcoin is a bubble, if most of its adherents plan to hold onto it for a long time, regardless of current price levels, then this reduces the likelihood of a complete collapse when everyone rushes for the exits.

In other words, even if this is a bubble, it may be a surprisingly durable one.

And the reason that "bubble" here is not necessarily a pejorative term is a point made by Moldbug - that money is the bubble that doesn't have to pop. In other words, there will be at least one good that is held in excess of its demand for other uses, because of its use for transactional purposes.

It may seem strange to reference Moldbug, since he comes out as a skeptic, based on his guess that the government will outlaw it.

But there is a counter-argument to that - the Uber problem. Namely, the government has a limited amount of time in which it can easily ban Bitcoin. The reason is that as the price gets high enough, enough people have enough to lose that it becomes politically costly to ban it. And so at some point, you get a compromise answer, like Coinbase seems to have done - you have to submit ID, it's linked to your bank account, and you have to give a social security number. The US Government levels capital gains taxes, everyone is happy. Why ban something if you can make more money by taxing it instead?

Because there is one rhetorical claim about Bitcoin made by its proponents that I think has caused more confusion than any other. It was this realisation that made me change my mind and invest in it. (Which, to emphasise, I'm not encouraging you to do. I'm some stranger talking smack on the internet, and this is not financial advice. But still)

It is this:

Bitcoin is not going to be a substitute for the dollar.

Bitcoin is going to be a substitute for gold.

Which is to say, the reserve asset that you hold in some amount as a hedge against the @#$% hitting the fan. This is of course, mid-level @#% hitting the fan, such as large-scale financial instability - if things really get hairy, the only worthwhile assets will be guns, ammunition, antibiotics, water purification tablets, and that kind of thing. But again, the same holds true for gold. If you honestly think that in a post apocalyptic New York there's going to be a vibrant demand for gold for jewellery purposes, perhaps you would do better investing your savings in shares in the Brooklyn Bridge.

Put another way, the case for Bitcoin in concise terms is that Bitcoin is to gold what neocameralism is to monarchy.

That is to say, it's what you get if you took an old but existing arrangement, and instead of trying to mimic it exactly, you thought about how you would design a modern version of it that a) retained the essential strengths while b) utilising technological innovation since the early form to overcome its weaknesses. (Some thoughts of mine on the neocameralism vs monarchy comparison are here).

In the view of Bitcoin, the essential aspect of gold is its relatively fixed supply. So let's go one better, and make a mathematically fixed supply. Rather than gold coins, let's create highly divisible bitcoins that can be traded across borders costlessly. Rather than measure purity over and over, let's create a blockchain to solve the problem of double-spending and transactions between mutually suspicious parties. Meanwhile, the fact that it can be mined by anyone easily at first, but only with more difficulty later, encouraged people to get in on the action early.

If you thought an essential aspect of gold was its value in jewellery, then you'd be a skeptic.

Rather, the other essential aspect of bitcoin was its first-mover advantage. Sure, someone else might invent other coins (and they have), but because Bitcoin was the first to market, it already has the advantage of incumbency. And in a co-ordination game, that's a huge deal.

And I think phrasing the question this way to economists helps to clarify the issue. In other words, if you're a Bitcoin skeptic and think its a bubble that's inevitably going to burst, I would ask you: is gold a bubble? This is harder to prove than in the case of Bitcoin, because it does have a fundamental value from other uses, so its value shouldn't go to zero. As a consequence, evaluating whether it's a bubble is much more thorny and more subjective. But it seems pretty clear to me that central banks aren't holding gold because they're about to turn their bullion into wedding rings. As Moldbug points out, in 2011 gold reserves were 50 times annual production. For silver, they were twice annual production. Assuredly there is something unusual about people's desire to invest in gold. So if you think that this is creating excess demand, surely this is pushing up the price, no? Supply is pretty fixed in at least the short term, if not the medium term too. And isn't excess demand pushing up prices the definition of a bubble? The point, of course, is that with gold this state of affairs has persisted for an extraordinarily long time. Is there any particular reason to assume that gold is about to disappear as a hedge asset? Not to me.

But I know my economist friends well, and I know their objection to the above reasoning, which makes Bitcoin different from gold. Which is to say, without a fundamental source of value other than as a money-like good, isn't the whole thing liable to unravel really quickly? Put differently, you and I might be willing to hold Bitcoin because we assume that there's reserve demand from Chinese officials and drug dealers, but why are the Chinese officials and drug dealers themselves willing to hold it?

This is another way of saying, why don't we all iterate backwards and realise that without an ultimate holder of the good from some other source, the value to everyone should be zero? Suppose we have a game where if we co-ordinate on a good being money, it gives value to both of us, but in the final round whoever is holding it ends up with a worthless asset.

If the game is finitely repeated, the economists are absolutely right. If everyone correctly performs backward induction, you'd predict a) Bitcoin should never have a positive value to begin with, and b) even if it does, this should be rather fragile. If it's an infinitely repeated game, then the Nash Equilibrium has more possibilities, as it usually does. In this case, if there is no final period, then it seems more like a straight one-shot co-ordination game where if we both agree, we both benefit. But let's take the finitely repeated version with a penalty for holding in the last round, as the logic is stricter there. And the logic dictates that since no-one is willing to hold in the last round, they don't want to hold in the second last either, and so on.

But here is the trillion dollar question - how much do people actually perform backward induction? And if they don't, how should you act in response?

The classic version of this is the iterated Prisoner's dilemma. Suppose two people are playing against each other 100 times in a row. The economist's answer is that if we're only playing a finite number of times, there's only one Nash Equilibrium to this game. We both defect in the last round. Knowing this, we both defect in the second last round, and the third last round, and thus in all rounds.

And yet... people don't. They routinely co-operate, and only begin defecting towards the end. This is why tit-for-tat works so well in practice. Because most people don't actually do backwards induction for more than a few iterations. This is why they don't start defecting until close to the end.

And bear in mind that, unlike Prisoner's Dilemma, Bitcoin is a co-ordination game, meaning it actually is a Nash Equilibrium for us all to believe in Bitcoin, at least in the one shot version. In the case of Prisoner's Dilemma, you can mathematically prove that people aren't acting rationally, and yet they still do it just the same. Here, it at least can be rational.

Now, bear in mind, the economists aren't wrong on the bigger picture - it still might collapse, for all the reasons they say. But that's not the same as saying that it has to collapse. I would guess, rather, that the opposite is likely to be true. The longer it goes without collapsing, the stronger the self-fulfilling aspect of the belief becomes, and the more stable it becomes.

Mainstream economists and finance types are looking at Bitcoin continuing to rise in price, yelling that this is a stupid and unstable equilibrium and that people should all start defecting immediately.

This is just like the economist watching two people play prisoner's dilemma and continue to co-operate round after round. You can laugh and call them morons, but a betting market just opened up. It's round 43 of 100. They both co-operated last round. Rubber to the road, what would you bet they're going to do this time?

After eight years of people continuing to not defect in Bitcoin, perhaps, dear economist, it's time to re-examine your assumptions.

Updated: On the other hand, if you wanted to make a concise case for a bubble, just check out some of the bizarre creations down at the lower market cap end of the cryptocurrency list. $10 million of FedoraCoin, you say? It's woefully underperforming PepeCash at $13 million. Hmm.


  1. Shouldn't we distinguish between "store of value" in the strict sense (like the chemical properties that make gold a better "store of value" than ice cream) and the stability of supply and demand for the money-commodity (whether extrinsic factors like demand for gold from jewelers or "intrinsic" factors like demand for USD from the IRS and the NY Fed)? That latter kind of stability is what makes the difference between a possible medium of exchange and a *liquid* medium of exchange (one that isn't vitiated by volatility, by sudden bubbles and crashes in the value of the currency) and a *reliable* unit of account (one that is not only easily quantified and measured out, but which can be used to denominate contracts and leases over a period of years without requiring some form of indexing inside the contract itself).

    I imagine the relevance of this q. to BTC is obvious to you, but to be pedantic and spell it out; what if the problem with BTC as a currency isn't backwards induction and isn't a total lack of any extrinsic value, but is that it will always be too volatile to function as actual *money*? It can have novelty value; it can be a convenient medium of exchange for criminals (like cigs inside a prison), which creates demand and raises the novelty value further; uncertainty about how high the price of BTC will go can create speculative bubbles; but ultimately the unpredictability of black-market activity and the bubbles to which it gives rise make BTC volatile, so it will never be monetized.

    1. Interesting point, I hadn't really thought about that. That probably actually is the biggest obstacle to using it for day to day money at the present time. But in some sense, this would go away if everything were denominated in Bitcoin, as the Bitcoin then mostly becomes invisible. Which is to say, we do think much about the price of the dollar, we just think about the price of bananas or iPhones IN dollars. The only time you notice the dollar itself is in periods of hyperinflation. There have been plenty of hyperinflationary currencies, but they existed only by fiat (in the true sense) - one suspects that in Zimbabwe or Weimar Germany they would have switched to USD if allowed and practical.

      The biggest menu cost aspect to Bitcoin as money is that it's deflationary - you need to keep adjusting the price of goods downward. Which would be somewhat odd, but I imagine you could get used to. Deflation used to be a lot more common under the gold standard.

      But it's harder to know what to make of it if you think of Bitcoin as gold. It's definitely true that massive volatility is undesirable in a hedge asset. Though if it lasts, I suspect the volatility will settle down.

  2. E-coins are uniquely capable of being used inside smart contracts, i.e. self-monitoring, self-transacting, self-settling dutch auctions between computers (e.g. between SatNavs and Road pricing sensors, between music files and streaming services etc etc). That is the Web 3.0 and it will be several orders of magnitude bigger than what e-coins are used for today