Tuesday, July 18, 2023

NY Fed: Metastablecoins

Stablecoin circulation
Last May, in the wake of the Terra/Luna crash, I wrote Metastablecoins, describing how coins like USDT intended to be stabilized via arbitrage were always at risk of the market over-running the arbitrageurs firepower, and More Metastablecoins, quoting from Bryce Elder's report on Barclay's analysis of "asset-backed" coins such as USDT that, while they claim to be fully reserved, do not offer immediate unrestricted liquidity:
We think that willingness to absorb losses, even though USDT is fully collateralized and has an overnight liquidity buffer that exceeds most prime funds, suggests the token might be prone pre-emptive runs. Holders with immediate liquidity demands have an incentive (or first-mover advantage) to rush to sell in the secondary market before the supply of tokens from other liquidity-seekers picks up. The fear that USDT might not be able to maintain the peg may drive runs regardless of its actual capacity to support redemptions based on the liquidity of its collateral.
Now Kenechukwu Anadu et al of the NY Federal Reserve have a deep dive into the events of May 2022 entitled Runs on Stablecoins. Below the fold I have some comments.

Anadu et al's abstract reads:
Stablecoins are digital assets whose value is pegged to that of fiat currencies, usually the U.S. dollar, with a typical exchange rate of one dollar per unit. Their market capitalization has grown exponentially over the last couple of years, from $5 billion in 2019 to around $180 billion in 2022. Notwithstanding their name, however, stablecoins can be very unstable: between May 1 and May 16, 2022, there was a run on stablecoins, with their circulation decreasing by 15.58 billion and their market capitalization dropping by $25.63 billion (see charts below.) In this post, we describe the different types of stablecoins and how they keep their peg, compare them with money market funds—a similar but much older and more regulated financial product, and discuss the stablecoin run of May 2022.
Their three types of stablecoins are:
  • Asset-backed coins, which they describe thus:
    Stablecoins backed by traditional financial assets resemble the structure of money market mutual funds (MMFs). Users can mint new coins by depositing dollars with the issuer. When users want to withdraw their dollars, they send their stablecoins back to the issuer, who returns dollars to users’ bank accounts.
    At least in the case of USDT, this isn't actually true, as Bryce Elder noted:
    Tether’s closed-shop redemption mechanism means it cannot be viewed like a money-market fund. Processing delays can happen without explanation, there’s a 0.1 per cent conversion fee, and the facility is only available to verified customers cashing out at least $100,000.
    "Processing delays can happen without explanation" may be an understatement. IIRC Tether's terms of service mean they don't actually have to permit withdrawals at all.

    The need for a custodian to hold the reserve assets backing these coins means that they cannot be decentralized.
  • Crypto-backed coins such as DAI, which they describe thus:
    DAI, for instance, can be backed by Ether, the native cryptocurrency of the decentralized, open-source blockchain Ethereum. In order for the peg to be credible, every dollar of DAI is backed by more than one dollar worth of Ether ... In order to mint DAI, an investor deposits the collateral in the smart contract; in order to redeem DAI, the investor deposits the DAI in the smart contract and receives back the collateral. Importantly, if the value of the collateral drops below the minimum collateralization level required by the contract, any user can call a function on the contract to liquidate the collateral through an auction and receive a percentage of the collateral as a reward. ... DAI can be backed by other cryptoassets including Bitcoin, USDC, and, more recently, tokenized mortgages.
    Stakes 1/19/23
    DAI is claimed to be "decentralized" in that it runs as a "smart contract on Ethereum, but note that Ethereum isn't very decentralized. I haven't seen an analysis of the potential for a run on DAI, but it is definitely exposed to contagion from projects that use it.
  • Algorithmic coins like Terra, which they describe thus:
    The developers of Terra created two cryptoassets: TerraUSD (UST)—designed to be stable—and Luna—designed to fluctuate over time, similarly to Bitcoin. Any investor in Terra had access to a smart contract that allowed them to create or redeem one unit of UST for one dollar worth of Luna. For instance, if the price of Luna was $10, the smart contract would exchange one Terra for 0.1 units of Luna. Therefore, irrespective of the price of Luna—as long as it was greater than zero—the value of UST should have been $1 due to an arbitrage opportunity.
Anadu et al's focus is on runs in algorithmic coins:
Terra redemptions increased the supply of Luna. As the price of Luna dropped due to the increase in its supply, each dollar redeemed out of Terra triggered an even greater increase of the supply of Luna. For instance, if Luna traded at $0.1, redeeming one unit of Terra would create ten units of Luna; but if Luna dropped to $0.01, redeeming one unit of Terra would create 100 units of Luna. At some point, investors would be unwilling to buy Terra even if its price dropped below $1 because it would be backed by an asset, Luna, whose price was rapidly declining.
Source
Exactly that is what happened:
As Terra suffered heavy redemptions, the supply of Luna increased from 365 million units on May 9, to more than 6 trillion units by May 13. Between May 7 and May 8, the algorithmic mechanism broke, and Terra broke the peg, with its price dropping from $0.9964 to $0.7934.
Just as with runs in traditional finance, such as recently happened at Silicon Valley Bank, the result was a flight to quality:
In about a week, between May 7 and May 16, the crash wiped out $17.17 billion in Terra’s market value and $20.77 billion in Luna’s market value. The run quickly propagated to other stablecoins, including USDT and DAI. U.S.-based stablecoins backed by traditional safe financial assets, however, such as USDC and BUSD, received significant inflows during the same period as investors moved from riskier stablecoins to less risky ones. In particular, circulation dropped by 8.70 billion units for algorithmic stablecoins and by 2.25 billion units for crypto-collateralized stablecoins; in contrast, the circulation of U.S.-based stablecoins increased by 3.88 billion units.
Note in particular that traders don't actually believe that USDT is safe, it is just that its size makes it convenient for traders to use USDT unless, like Wile E. Coyote, they look down at it as they did last May.

Again as in traditional finance, "bank" runs cause panic elsewhere:
the May 2022 stablecoin run, for instance, affected the broader crypto market, with approximately $200 billion in crypto market value (beyond stablecoins) being wiped out over eight days.
BTC "price" 2022
Anadu et al's analysis agrees with mine but is, of course, more expert. I have the following observations:
  • Looking at their first graph, it is striking that this major event only reduced the "market cap" of stablecoins by less than 10%.
  • Looking at their second graph, showing data just for the first half of 2022, it is striking that this major event only wiped out six months' gains.
  • Looking at BTC's "price" history for 2022, it started the year about 30% down from its peak the previous November and before the Terra/Luna news broke it was around $40K, down around 40%. The news took it down to around $30K or 55% down from the peak. So an event that hit stablecoins by 10% hit BTC by 25%. Other events continued to weigh on it the rest of the year, and 2023's desperate pumping hasn't managed to get it back over its "price" after the Terra/Luna collapse
  • The Terra/Luna collapse was a major factor in the subsequent collapse of Three Arrows Capital:
    The firm invested approximately $200 million in LUNA tokens in February 2022.
    The contagion from the collapse of 3AC spread widely, so while the run on Terra was not the start of the avalanche, it significantly accelerated it.
The New York Fed's analysis of runs on (meta)stablecoins should be compared with:

1 comment:

David. said...

On the subject of bank runs, see also Alexandra Scaggs' More on uninsured bank deposits:

"The banks that failed, particularly Silicon Valley Bank and First Republic, weren’t holding on to uninsured deposits in concentrated sectors because they were dumb. It was part of their business strategy. And until recently they were rewarded for doing it."