There are two major categories of ERC20 stablecoins: centralized or decentralized. A centralized stablecoin is issued by a single authority. The authority could mint any supply of the coins. A centralized coin holder has to trust the central authority for the pegged value to hold. A decentralized stablecoin implements a mechanism for minting, burning, and market operations to maintain a desired trading value.
The overall crypto market swing wildly. It is easy to maintain a peg when the market is under price fluctuations of 10-30%. However, it is not unreasonable to assume that the overall market could contract 95%. One of the primary uses of stablecoins is have a mechanism to protect one’s portfolio against such an event. If a stablecoin cannot reasonably survive such a contraction, it is not a viable stablecoin. One has to use a different strategy in case of market crashes, which is a probable event within the horizon of a few years at this junction of the crypto ecosystem.
As of the writing of this blog, the circulating volume of centralized stablecoins is about 10x of that of decentralized stablecoins. The volume of the top three centralized stablecoins (Tether, USDC, and Binance USD) is about $150 billion USD. The top 3 decentralized stablecoins (TerraUSD, DAI, and Fei) is about $25 billion USD.
Centralized
The most important thing to evaluate a centralized stablecoin is to evaluate the trustworthiness of the issuing organization. This is similar to how fiat currency holders evaluate a currency and its issuing nation state central bank. The key difference is that a central bank is backed by a monopoly of violence, and the authority of a stablecoin minter is the combination of its trust in the community, fund raising history, and legal obligation to its stated intents. That being said, I would probably trust the stability of a properly audited, sufficiently collateralized, and US-based1 issuing company over more than 95% of the central banks authorities in the world. In the other words, holding a properly issued centralized stablecoin is probably safer than holding but a handful of real world fiat currencies.
USDC
The issuing organization is founded by Circle and Coinbase. To the best of my knowledge, both of these companies have the power to issue USDC. The guarantee is that for each USDC coin issued, there is a corresponding amount of fiat USD deposit in a regulated bank. I would suppose that Circle and Coinbase do not keep a fixed 1 to 1 ratio. Their business model takes advantage of fractional reserve to invest the collateral, similar to how retail banks makes money from retail deposits.
Coinbase is a large public company. It has every incentive to comply with all of their stated intentions and how they are going to have sufficient funds to guarantee USDC withdraws. Circle is a well funded startup, and I would have to imagine that they have the same objective. I would feel comfortable holding a large chunk of my investment in USDC and believe that stablecoin value could be maintained under severe market turmoils.
USDT (Tether USD)
Tether had been the most circulated stablecoin for a long time. It is also the most controversial. One has to do its own research, e.g. link1, link2, link3, to make up their mind on how much to trust this stablecoin. I am comfortable accepting tethers in my defi transaction, even in large amount. However, I would not hold tethers long term.
TUSD (TrueUSD)
TUSD is issued by the company TrustToken. The company contracts a third party legal law firm to independently verify its digital assets to ensure that the issued TUSD could be claimed 1-for-1 in US dollars. It is similar to USDC. The key difference is how much I would trust TrustToken vs. Circle/Coinbase. I trust USDC more than TUSD. I trust TUSD more than I do Tether. I would not use TUSD as my primary stablecoin.
BUSD (Binance USD)
Binance is an interesting company. It is the biggest crypto exchange in the world. It also has questionable legal structures. It is operating somewhat in the gray area in the United States. Because of its market dominance, it gets a lot of usage in many parts of the world for many of its product offerings. Stablecoin is one of its popular products, and its centralized smart contract, EVM chain is wildly popular. However, I consider Binance a temporary phenomenon because it primarily takes a centralized approach to the decentralized ecosystem. I am concerned about its legal status in the US. I have seen similar oversea companies that had to shut down their US operations with just a single regulation change2. I would not hold any significant amount in BUSD.
Decentralized, Algorithmic Stablecoins
Decentralized stablecoins rely on at least some decentralized algorithms to maintain their peg value. There are different ways to categorize these coins based on the implemented stabilizing mechanisms, e.g. algorithmic vs. asset-back, over-collateralized vs. fractional reserve, etc. I will not attempt further categorizations here. Some of the mechanisms might even be administered by a centralized operator, but overall, the aim is to keep the key operations as transparent and as automated as possible.
The single most important criteria to determine the quality of any stablecoin is how much faith do I have on its pegged value under extreme market conditions. The only true test is that a stablecoin were to survive such an event. Unfortunately, the market swings in the past 3 years were not sufficient to convince me that any of the stablecoins discussed below would survive an extreme market contraction. I would not feel comfortable holding all of my crypto funds in any of these decentralized stablecoins and believe that they will hold the USD peg when the overall market crashes similar how ETH crashed in 2018. ETH went from $1200 to $100.
DAI
DAI was the first decentralized stablecoin. It seems that it takes a rather conservative approach. Every DAI that is minted, it needs to over-collateralized. For example, for every $100 of newly minted DAI, the protocol requires $150 worth of Eth deposited. The over-collateralized requirement might appear to provide a high guarantee on the underlying value of DAI. That is only true if the requirement to be 50% over is a sufficient buffer. However, ETH losing 50% of its value should not be considered a black swan event. We have seen ETH losing more than 90% of its value. The high collateral limit would not be sufficient.
DAI is minted when a user creates a VAULT and deposits cryptos. Every DAI in circulation is directly backed by excess collateral. The protocol uses an oracle, i.e. Chainlink, to make sure that each vault has sufficient collateral. When the collateral falls under a threshold, the protocol sells the cryptos to cover its DAI position. If there are excess DAIs from the liquidation process, DAI is refunded to the vault holder. If liquidation is not sufficient, the protocol incurs a protocol debt. The protocol maintains a maker buffer. This buffer could be used to cover the protocol debt. This buffer could be increased by a debt auction. The protocol mints MKR and sells those to increase the buffer. Minting MKR is inflationary, and all MKR tokens holders lose equally. If the buffer gets too large, the protocol would try to buy back more MKR and burn them. This is a deflationary action and all MKR holders gain equally.
MKR is the DAO token of the protocol. DAO could vote to update the protocol. DAO determines what cryptocurrencies are accepted as collateral.
DAI is not capital efficient. The capital that are committed to mint DAI is locked up. Worst, the collateral level is above 100%. Centralized stablecoins could use fractional reserve.
Even the high over-collateral level might not be enough if market falls precipitously. If Eth goes down 70%, almost all vaults have to be liquidated and still incurs protocol debt. MKR price could go down concurrently with ETH even before the all the liquidation take place. The remaining market capitalization of MKR might not sufficient to cover the protocol debt. At that point, there is no amount of minting of MKR is going to be able to cover the rest of the open DAI positions. DAI price would have to go below the peg.
The best thing about DAI is that it is highly decentralized and the algorithm is simple. As long as the number of circulated DAI is not large relative to the market capitalization of MKR, I would not mind conducting most of my defi transactions in DAI. However, I would not uses DAI as my cash reserve when I am expecting market turmoil. DAI was an amazing innovation. It took the first step toward a viable decentralized stablecoin. But at the same time, it is an unsatisfactory long term answer.
Fei
FEI aims to be more capital efficient than DAI. FEI is minted when someone buys FEI against a bonding curve set by the protocol. The capital accrued by the protocol is used to conduct market operations to maintain the peg. Initially, the only bonding curve is denominated in ETH. All of the ETH is put into a uniswap liquidity pool. The amount of total liquidity roughly matches the total FEI in circulation.
The stabilizing mechanism is mostly automatic. The incentive to push down the FEI price is simple to understand. If FEI is selling above $1, say $1.05. Someone could mint the token at $1 from the bonding curve, and then sells the token at normal exchanges for a $0.05 profit. When FEI is selling below the peg, the protocol controller has to buy FEI. I believe that this part is not fully automatic in the smart contracts yet and requires manual intervention. When the peg is below target for an extended period of time, the controller took out the uniswap pool and use the fund to buy FEI until FEI is back to $1. The remaining money will be put back into the uniswap liquidity pool. Hence, when FEI is selling at $0.95, a person has incentive to buy from an exchange (e.g. uniswap) because if the prices does not get back to $1, the protocol controller will buy it at $1. Oftentimes, the belief that someone will buy it at $1 is sufficient for the peg to be remained at $1. The person profits $0.05 one way or the other.
FEI’s market mechanisms are clever. It allows for the protocol to bootstrap itself. It allows for redeemability at all times. It does not require capital lockup from the user’s perspective even though it effectively creates a fractional reserve from the bonding curve. The incentive mechanisms are simple and direct. FEI holders should have a reasonable belief that the peg value holds under most conditions.
The key question is always the same: will the peg survive a 95% market contraction. I am going to say no. But unlike DAI, I could see FEI surviving the initial shock. If the contraction continues, I would predict that FEI will lose values once most stablecoin holders want to exit the crypto market all together. During the initial shock, the protocol controlled value is less than 5% of the value of the circulated FEI. If there is a run at the bank, the protocol controlled value evaporated quickly and FEI falls to zero. However, during market turmoil, most stablecoin holders are not in a hurry to convert FEI to more ETH. In fact, there might be more ETH holders who are looking to buy more FEI, therefore increasing the protocol controlled value in relative terms. If the bear market continues and FEI holders want to leave the crypto ecosystem, it will become an effective run at the bank, albeit at a slow pace. If the volume of sell outpaces buy even by a little, the FEI market would collapses because of the large value of outstanding FEI.
UST (TerraUSD)
TerraUSD (UST) is a stablecoin on the terra blockchain. Terra is a layer 1 blockchain that focuses on implementing stablecoin mechanisms. TerraUSD (UST) is the stablecoin that is pegged to USD. Terra chain supports stablecoins that are pegged to other fiat currencies or currency basket as well. LUNA is the native token of the terra chain. LUNA is the staking token used to secure the chain. Stakers validate transactions and earn rewards from transaction fees. The UST could be bridged to Etheruem as a ERC20 token.
UST’s peg value is obtained by balancing the supplies of LUNA and UST. Its incentive mechanism crucially depends on the assumption that UST price could be sufficiently influenced based on expanding and contracting the total circulated UST. When the price of UST is above $1, the protocol incentives users to burn UST to mint LUNA. When the price of UST is below $1, the protocol incentives users to buy UST with LUNA.
Similar to DAI and FEI, the mechanisms should work sufficiently well under normal market conditions. When LUNA and the overall crypto market crashes, the incentives might not be strong enough for UST to maintain the desired pegged value. LUNA could lose so much value that the outstanding value is only a small fraction, say 5%, of the total outstanding UST. At that point, even 90% of LUNA being burned to buy UST might not be sufficient to counter the downward price pressure. I suspect that its survivability is similar to FEI. The primary difference would be on the strength of the based assets: LUNA vs. ETH. If both of those assets fall equally in percentage, Terra and FEI’s market mechanisms probably perform similarly. While it is not a guarantee that the peg is not maintained, it is also not a guarantee that the peg is maintained, neither. It all depends on if there is a genuine panic sell on UST. The panic could be entirely avoided only if there is a guarantee that even in the event of a panic, peg value holds. That is one of the key insights of modern microeconomic theories.
Final Thoughts
I like decentralized stablecoins. I primarily use decentralized stablecoins in defi transactions. I want to use decentralized stablecoins only. However, I do not have a firm belief that any of the existing decentralized stablecoins could survive extreme market movements. Cryptocurrencies are known to be volatile, and I have seen huge crashes in the recent past. The belief that a decentralized stablecoin might not maintain value under a panic sell is the reason why a panic sell will happen. As much as I dislike centralized solutions, I expect a well collateralized centralized solution with a diversified asset portfolio to hold its peg even when the issuing company’s stocks crashes along with the overall crypto market. It would at least hold for as much as the underlying assets, which correlates with the asset prices outside of the crypto market. The biggest risk with centralized stablecoins is regulatory. The U.S. government has an outsized and unilateral power to completely freezes assets of the issuing authority, which would effectively kill the corresponding stablecoin. However, that risk is minimum compared to other market risks as of now. If my holdings were already in an centralized exchange (e.g. Coinbase), I would be indifferent about holding USD or USDC when I sell my portfolio in anticipation of a crash or in the middle of crash. If my holding were to be in decentralized wallets, I would only mostly hold centralized stablecoins and only hold small amount of decentralized stablecoins in a stressed market.
Footnotes
- The U.S. effectively has a monopoly in the world’s financial systems. If a company does not have serious operations in the US and its issued stablecoin exceeds in the 10s of billions of dollars, the US congress or even a simple executive order could completely freeze that company’s financial assets even when those assets are not held by US institutions. Financial institutions around the world cannot defy US rules and regulations in the slightest, or they themselves could be blacklisted. ↩
- Pokerstar was one such example. ↩