I guess it is time to talk about algorithmic stablecoins again. Terra, or UST, is an algorithmic stablecoin whose price is maintained by an arbitrage relationship with another cryptocurrency, Luna. One UST is supposed to be worth one US dollar, and one UST can always be exchanged for a floating quantity of Luna with a market value of $1. If a UST is trading at $0.99, you can buy it for $0.99 and then exchange it for $1 worth of Luna, making an instant profit. If it is trading at $1.01, you can buy $1 worth of Luna (for $1) and use it to buy a UST worth $1.01, making an instant profit. Because of this arbitrage relationship, while the price of Luna can fluctuate, the price of Terra should always be $1: If it trades above or below $1, people will exchange Terra for Luna or Luna for Terra until the price of Terra gets back to $1.
On first principles this is insane. It relies on [Luna] always being worth something. If [Luna] trades at $0.01, you can print 10 million of them and buy 100,000 [Terra] and push the price up. But if [Luna] trades at $0.00, you can print infinity quadrillion of them and you’re still not gonna be able to push up the price of [Terra]. If [Luna] is worthless, it cannot be used to support the price of [Terra]. And because you just made it up, there is no particular reason for [Luna] to be worth anything, so there is no particular reason for [Terra] to be worth a dollar. If I made up [Luna] and [Terra] on my computer and said to you “I will give you the number 10 billion in this Excel spreadsheet if you give me 1 million U.S. dollars,” you would say no, and if I raised my offer to 400 quadrillion you would not change your mind.
Nonetheless! It works? The rough intuition here is that there is a lot of demand for stablecoins; there is particularly a lot of demand for Terra because Terraform Labs, the entity that created Luna and Terra, essentially pays 19.5% promotional interest on UST deposits. People want a stablecoin that is worth a dollar, so they are inclined to treat Terra as though it’s worth a dollar, which makes it worth a dollar. They buy lots of Luna to turn into Terra, which means that the price of Luna goes up, which means that there is plenty of valuable Luna to support the price of Terra, which means that Terra is robustly worth a dollar.
Ha, well, oops:
Rather than trading at $1, as designed, the TerraUSD coin, or UST, slipped over the weekend to around 99 cents. By Monday evening in New York, it had plunged to 60 cents, obliterating its previous low of 92 cents in May 2021. It clawed back losses on Tuesday and is fluctuating between around 90 cents and $1 — a sign of trouble. …
There are around 18.5 billion of UST in circulation, according to CoinMarketCap, a big enough presence that its swings could have systemic implications for other coins and protocols. And Do Kwon, the crypto upstart behind UST, has previously committed to buying as much as $10 billion worth of Bitcoin as part of his support of the coin, further entwining the project with the core of the digital-asset market. …
That led to a series of crypto market interventions from Kwon and the so-called council of the Luna Foundation Guard (LFG), a consortium of crypto players that includes Kanav Kariya of Jump Crypto. Jump Crypto declined to comment. Near midnight New York time on Monday, UST remained under pressure. Luna was trading around $29, down 52% from a day earlier, according to CoinMarketCap.
The declines spurred backers led by Do Kwon, the founder of Terraform Labs — which powers the Terra blockchain — to issue $1.5 billion in loans denominated in both UST and Bitcoin to help support the digital currency.
The basic thing that makes Terra valuable is confidence in it. The essential source of this confidence is … just sort of recursive social belief? If you think that everyone else will treat Terra as worth a dollar, then you will treat it as worth a dollar, and you won’t sell it for $0.90 in a panic, which means that it won’t go down to $0.90, etc. But if you think that everyone else will treat Terra as worth zero, then you will dump it as fast as you can at whatever price you can get, which means that it will go down below $0.90, etc.
Also there is an algorithm, but it is a complicated cloak thrown over these basic social facts. If one Terra goes down to $0.90, the arbitrage mechanism — you exchange one Terra for $1 worth of Luna, and then sell your Luna into the market for $1 — just doesn’t work. You exchange one Terra for $1 of Luna, but confidence in Luna is also falling, and the market is being flooded with Luna as people try to do this arbitrage. So the $1 worth of Luna you received is no longer worth $1, and then the next person who redeems Terra gets even more Luna and has to sell even more of them, which drives the price down more, which increases the amount of Luna being issued, etc., in a “death spiral.” There is no particular floor on this process, and it can go until everything is worth zero.
However! Kwon and the Luna Foundation Guard did a smart thing. During the virtuous cycle of Terra’s existence, as its market capitalization grew and as Luna became more valuable, they used their valuable Luna to buy a bunch of Bitcoins. Luna is a creature of Terraform: If you lose confidence in Terra you will simultaneously lose confidence in Luna, and being able to exchange one Terra for infinity bazillion Luna will not do anything to prop up the price of Terra. But Bitcoin is an entirely separate thing. Terraform made up Terra and Luna, but somebody else made up Bitcoin. If people lose confidence in Luna and Terra, Bitcoin will still be valuable.
And so the LFG bought a bunch of Bitcoin and promised to use it to defend Terra’s peg to the dollar. If one Terra goes down to $0.90, instead of turning Terra into Luna and selling them in a death spiral, the LFG can buy Terra for $0.90 and pay for it in Bitcoin. If the LFG has enough Bitcoin, and if Bitcoin’s price holds up, then it can defend the peg and keep the price of Terra close to $1.
The point is that you print a lot of Luna when Luna prices are high and exchange them for Bitcoin. And then if Luna prices fall, you can use the Bitcoin to buy Terra and keep it at $1, avoiding a death spiral. As I wrote last month: “The basic structure of the trade is (1) Ponzi, (2) acceptance, (3) diversification, (4) permanence.”
I suppose Kwon would have liked a few more months of widespread acceptance of Terra, so he could build up the Bitcoin reserves. (“We will keep growing reserves until it becomes mathematically impossible for idiots to claim depeg risk for $UST,” Kwon tweeted in March; the target was $10 billion of Bitcoin.) But he bought a lot of Bitcoin, anyway, which gave him a lot of ammunition to use to defend Terra’s peg when it broke over the last few days.
So far it is … working … okay? As of noon today, CoinMarketCap tells me that Terra is at about $0.91, which is simultaneously (1) quite a lot lower than $1 but also (2) quite a lot higher than $0. Those are the two stable equilibria of an algorithmic stablecoin: In the long run, either it’s worth a dollar or it’s worthless. Terra is still fighting it out: It broke the buck (bad), but it does not seem to be in a death spiral (good). Either Kwon, LFG and other Terra bulls will defend the peg and wrestle it back to $1, or, uh, well, or they won’t and it will go to the other place.
As of right now I don’t know what the answer is and wouldn’t want to hazard a guess, though I will say that $0.91 is much closer to $1 than it is to $0. “Deploying more capital – steady lads,” Kwon tweeted yesterday. I bet he’s having fun. “Close to announcing a recovery plan for $UST. Hang tight,” he tweeted this morning.
People often talk about “bank runs” on stablecoins. Here, for instance, is the Federal Reserve’s annual Financial Stability Report, released yesterday, which contains some worrying about run risk on stablecoins:
Stablecoins typically aim to be convertible, at par, to dollars, but they are backed by assets that may lose value or become illiquid during stress; hence, they face redemption risks similar to those of prime and tax-exempt [money market funds]. These vulnerabilities may be exacerbated by a lack of transparency regarding the riskiness and liquidity of assets backing stablecoins. Additionally, the increasing use of stablecoins to meet margin requirements for levered trading in other cryptocurrencies may amplify volatility in demand for stablecoins and heighten redemption risks.
The idea of a bank run is that you have a backed stablecoin, like Tether, in which each $1 stablecoin is in theory backed by at least $1 worth of dollar-denominated assets. The worry in a bank run is that people all decide to withdraw their money from the stablecoin at once, perhaps because they worry about its solvency (i.e., they worry that it is only backed by $0.99 of assets) or for some other reason. To meet withdrawals, the stablecoin has to sell assets, which drives down their price, which leaves the coin backed only by, say, $0.95 worth of assets. Panicky holders withdraw even more money, pushing the assets down more. And since holders get paid out at $1, they drain value out of the pool; people who don’t withdraw are left with claims on the bad illiquid assets, which are worth even less.
That’s not great, but what I want to emphasize is that this is much worse. A bank run is a fairly mild event compared to a death spiral. In a bank run, your $1 stablecoin is backed by some pool of assets that are worth something; a rush to withdraw money leads to fire sales that depress their value, but their value derives from something other than confidence in the stablecoin. In a death spiral, the whole system is built on confidence in the stablecoin; if that confidence evaporates then there is no floor at all on the value.
To put it another way: In the 2008 financial crisis, a money market fund called the Reserve Primary Fund “broke the buck”; it had exposure to Lehman Brothers debt, which defaulted, leading to a run on the fund and ultimately its liquidation. This was huge news at the time, a seminal event of the global financial crisis. Also Reserve Primary Fund investors got back about 99 cents on the dollar. The potential outcomes for a money market fund, or a fully backed stablecoin, are like (1) $1 or (2) slightly less than $1. That is not how algorithmic stablecoins work.
Meanwhile, the mechanism that LFG is using to defend the peg is interesting. It has loaned Bitcoin to market makers to use to protect the UST peg to the dollar. For one thing, this means that LFG (or, rather, the market makers) is effectively selling Bitcoin to buy UST, putting pressure on the price of Bitcoin: If you are dumping Bitcoin to prop up the price of Terra, the price of Bitcoin will go down.
If you are a market maker, presumably your bet here is that you can buy Terra at $0.91, it will stabilize, and you will end up being able to sell it in a few days for $1, making a quick 10% profit. Then I guess you buy back the Bitcoins you used and deliver them back to LFG? You have some Bitcoin/dollar price risk; I suppose you can hedge that. The trade here is that $1 is a stable equilibrium for Terra, and if you can push it back to that equilibrium you’ll make a nice return on your efforts. You just need to make sure that you have enough capital to push it back to $1.
Because the bad outcome is that this does not work, there is a death spiral, and Terra ends up worthless. Then you wasted like a billion dollars of Bitcoin buying something that was going to zero. That would be bad! Arguably LFG’s decision to deploy a lot of Bitcoin is a sign of confidence that it will work. Otherwise they’d keep the Bitcoins.