Why It's Hard To Create Cryptocurrency Value With Just An Algorithm

And, as society is learning the hard way, it is hard to create other types of value - and of values - with just an algorithm, as well. JL

Matt Levine reports in Bloomberg:

Bitcoin is valuable. Bitcoin’s value is incredibly volatile. To look at those two facts - a cryptocurrency that is both valuable and volatile - and say, what we need is monetary policy. The problem is that you can’t just apply a stabilizing monetary policy to Bitcoin because it’s not programmed to work that way. You can build another cryptocurrency that is programmed that should be stable. But there is no guarantee that it will be valuable. Whilst algorithmic (cryptos) eliminate need for trust in a third party, they (are) dependent on investor belief and confidence. As long as all users believe that the coin will be stable, their behaviour ensures it will be stable, but if some users start to lose confidence and sell, the coin risks falling into a downward spiral.
Here’s a trade:

1. I will issue some Mattcoins for $1 each.
2. I will use those dollars to buy something nice for myself; forget about the dollars.
3. The Mattcoins will then trade freely on a crypto exchange.
4. If the Mattcoins trade for more than $1, I will issue more Mattcoins, increasing supply and driving the price back down to $1.
5. If the Mattcoins trade for less than $1, I will buy back some Mattcoins to reduce supply and drive the price back up to $1.
6. Not with dollars, I mean; remember, the dollars are gone.
7. Instead, I will buy back the Mattcoins in exchange for a note promising to pay more Mattcoins in the future: If you give me one Mattcoin now, I might promise to give you back 1.1 Mattcoins at some point in the future. (For instance, if they trade above $1 and I need to issue more.) This reduces the current supply of Mattcoins, meaning that people who want them have to pay more for them, driving the price back up to $1.

Let me make three points about this mechanism:

1. It more or less describes the mechanism of Basis, an “algorithmic stablecoin” that got a lot of attention when it was proposed in 2017 but that ultimately did not launch due to securities-law concerns. This is not a coincidence; I have loosely modeled my little trade on Basis.
2. It also more or less describes the mechanism of modern developed-world monetary policy: When the dollar, for instance, starts to be worth less than a dollar (this is called “inflation”), the Fed will issue bonds (promises to pay more dollars) in exchange for dollars, reducing the money supply; when it starts to be worth more than a dollar (“deflation”), the Fed will buy back bonds to increase the money supply. (Do not write this on your macroeconomics exam or anything, but it’s close enough for our purposes.) This is also not a coincidence; Basis clearly modeled its mechanism on a stylized form of central-bank operations.
3. Mattcoins are obviously insane.

 The reason that Mattcoins are obviously insane has nothing really to do with the mechanism; the mechanism is good and fine and it’s how actual dollars work. The reason that Mattcoins are insane is just that the correct price you should pay for a Mattcoin is zero: No one uses them, no one wants them, there’s nowhere to spend them, I have probably programmed their blockchain wrong, they have my name on them, they’re a ridiculous dumb joke. If somehow I convince some people to buy Mattcoins for $1, and they start trading, they are not going to trade up, so my ability to expand the money supply is irrelevant; they are going to trade down, at which point I will buy Mattcoins back in exchange for a promise to pay more Mattcoins in the future, which is also worthless. There is a death-spirally element to trying to prop up the price of Mattcoin by promising more and more Mattcoins.So why does it work for the dollar and not for Mattcoin?

Well, I think there are two somewhat separate things going on here, value and stability. A dollar is clearly valuable, in that a large number of people already accept it in exchange for stuff, you can pay your taxes with it, etc. But the fact that it’s valuable doesn’t tell you what that value is; the fact that the only practical way to buy a sandwich in America is with dollars doesn’t specify how many dollars a sandwich costs. And in fact currencies’ values have a tendency to fluctuate, and a big purpose of central banking is to limit that fluctuation. But the tools that central banks have tend to rely on the fact that everyone agrees that a dollar is worth something; the central bank just tries to limit the volatility of what that something is.
Mattcoin’s mistake is applying that sort of stabilizing mechanism to thin air. You just can’t stabilize the value of Mattcoin at $1 if its value is obviously $0. The stabilizing mechanisms can work for something that everyone wants; they can’t work for something that no one wants.

I genuinely do not know what this means for Basis and other algorithmic stablecoins. I am not being sarcastic here; I genuinely do not know. Bitcoin, after all, was created from thin air not all that long ago, and now it seems almost inarguable that it is valuable. Bitcoin’s value is incredibly volatile, though. It is not unreasonable to look at those two facts — a cryptocurrency that is both very valuable and very volatile — and say, well, what we need here is monetary policy. The problem is that you can’t just go apply a stabilizing monetary policy to Bitcoin, not only for philosophical reasons but also because it’s not programmed to work that way. You can go build another cryptocurrency that is programmed to work that way. That other cryptocurrency should be stable, or stabilize-able anyway. But there is no guarantee that it will be valuable.
Or that is my intuition, here in 2019 after the crypto collapse. I think there was a time not that long ago — certainly in mid-2017, when Basis was proposed — when the default assumption was that every cryptocurrency would automatically be valuable. If that was your assumption, then you could sensibly talk about launching a cryptocurrency and using algorithmic monetary policy to stabilize its value.

Anyway here is an admirably clear discussion of algorithmic stablecoins (particularly Basis) by Ben Dyson, on the Bank of England’s Bank Underground blog. (An algorithmic stablecoin is a cryptocurrency that is meant to be worth a
dollar, or whatever, but that does not back that value by holding 1-for-1 reserves of dollars.) Dyson writes:

The problem is that the Basis white paper assumes that the algorithm can push up the price by simply reducing the total stock of Basis coins (by swapping them for bonds and destroying the coins). But in reality the price is determined not by the total stock of coins, but by the balance of sellers and buyers in the market. The algorithm needs to reduce the number of sellers, but it is these low-confidence sellers who are least likely to buy Basis bonds. Meanwhile the high-confidence coin-holders, who might be tempted to swap their coins for bonds, are the ones least likely to be sellers in the first place. Consequently Basis bonds are not an effective way to reduce selling pressure on the market.

Whilst algorithmic stablecoins like Basis manage to eliminate the need for trust in a third party, they instead end up being heavily dependent on investor belief and confidence. As long as all users believe that the coin will be stable, their behaviour ensures that it will be stable, but if some users start to lose confidence and sell, the coin risks falling into a downward spiral. 

The post is so compelling that it kind of makes you wonder what people were thinking with all of this algorithmic-stablecoin business. I think the answer is just that in mid-2017, investor belief and confidence in cryptocurrency generally was really high; you could take value as a given — for virtually any project, and certainly one backed by a good development team and fancy economic theory — and focus on stabilizing mechanisms. That no longer seems quite as compelling.