No surprise to anyone, it’s been a volatile year in cryptocurrencies. But it may come as a surprise that it’s also been a volatile year for so-called ‘stablecoins’. A stablecoin is a cryptocurrency with an economic structure built on top of blockchain that aims to stabilize the trading price of the coin. A true stablecoin would offer the benefits of cryptocurrencies (cryptographic security, privacy, incentive alignment, digital usability, and accessibility to the unbanked) without the unusable volatility. It could form the basis for a robust decentralized financial system and facilitate economic adoption of cryptocurrencies.
The volatility in stablecoins is not reported very much and seems to be overshadowed by overall crypto volatility. However, stablecoin volatility is the more interesting news. Crypto volatility is nothing new. I’ve talked before about how the decentralized stablecoins are badly designed. This year is demonstrating these problems.
Recap on Stablecoin Problems
Existing stablecoin designs are fragile. They bootstrap ideas from traditional finance into a blockchain setting, introducing unaddressed failure points:
- Centralized designs rely on trusted institutions to hold reserve assets off-chain and/or report off-chain market data to the blockchain (called an oracle, which might report e.g. USD prices). These institutions are susceptible to failure, hacking, and fraud.
- Decentralized designs create complicated and poorly designed systems of financial contracts backed by volatile cryptocurrencies. These systems are prone to liquidity crises, in which speculative agents no longer take opposing sides of stabilizing trades, which cause these systems to fail.
NuBits is one of the first stablecoins dating back to 2014. If NuBits becomes ‘undervalued’, it relies on (1) people being willing to go long the currency to ‘arbitrage’, (2) people being willing to lock up their NuBits in a bond-like instrument that promises future returns when the currency becomes overvalued next (should this even happen), and/or (3) being able to sell new NuShares (representing some sort of interest in the system that becomes profitable if the coin becomes overvalued again) to in essence recapitalize the system. Obviously, people may not be willing to take these positions if their faith in the future value of NuBits is shaken, as NuBits experienced in 2016 and is ongoing currently in 2018. To date, it is down to pennies on the dollar.
The failure of NuBits has implications for other algorithmic stablecoin models, which are substantially similar. I’ve discussed this previously here with a focus on Basis. In fact, Basis just recently decided to close down and return $133M back to investors. This is attributed to regulatory concerns but may also be tied to the failure of NuBits.
bitUSD and Steem Dollars are Broken
I actually haven’t seen this reported anywhere, but bitUSD and Steem Dollars have both definitively broken their USD pegs this month. For some, this may be easy to overlook because of their small market caps. But together, this adds to nearly $20M (in pegged value), which is not insignificant compared to other decentralized stablecoins.
BitUSD is a crypto-collateralized stablecoin. That is, bitUSD coins are contracts that promise to pay $1 worth of crypto in the future, which are collateralized by crypto in the present. Crypto-collateralized stablecoins pair stablecoin holders, who seek stability, with speculators, who take the other side of the bet. Speculators lock cryptocurrency collateral as insurance on the stablecoin value. Based on their collateral value, speculators can create new stablecoins against that collateral up to a limit. In return the speculator attains a leveraged crypto position, which is valuable if crypto prices increase. For instance, if a speculator starts with 1 ETH in collateral (actually BTS is used as collateral for bitUSD, but ETH is more recognizable as an example), he/she can create and sell 0.5 ETH worth of new stablecoins, which leaves the speculator with a position of 1.5 ETH (but also a liability of the stablecoins created).
Steem Dollars work a little differently. All Steem holders essentially back Steem Dollars, which are payed as rewards on the Steem platform. Steem dollars can be redeemed for $1 worth of newly minted Steem, and so redemptions affect all Steem holders via inflation.
The precise reason bitUSD and Steem Dollars have broken their pegs (and at different times) probably has to do with liquidity in their settlement algorithms, through which stablecoin holders can redeem coins for underlying collateral. For instance, bitUSD has a 24 hour settlement process whereas Steem Dollars has a 1 week settlement process. The breaking of these crypto-collateralized coins has implications for other similar models, such as MakerDAO’s Dai. The algorithms are different (and, in some cases, manual as opposed to automated) but the designs are subject to the same fundamental risks.
Dai May Enter a Feedback Loop
Dai is another crypto-collateralized stablecoin. It follows the same model as bitUSD with some additions, such as a mechanism for emergency sales of Maker tokens to re-collateralize the system in a downturn. On first glance, Dai looks to be holding strong. The collateralization ratio remains high (although it’s seen worrying dips as the ETH price declines), and recently the price of Dai has appreciated to $1.02. To the casual observer, this seems to indicate strength. However, the decentralized stablecoins are complicated systems and, as I’ve argued before, their failure points are not well understood.
If we dig into the data a bit more, we see the recent price appreciation is tied to a decrease in the Dai supply. This suggests that speculators, who are providing collateral to the Dai system, are decreasing their leverage (i.e., how much Dai they are liable for with their collateral) by buying back Dai. In turn, this pushes up the price if the demand for Dai from stablecoin holders remains the same. It then becomes harder for speculators to lower leverage again or for liquidations to happen in the future with the given collateral. Essentially the buyback would have to happen at a higher price and only becomes harder in subsequent rounds of leverage decrease.
This presents a feedback loop. If the ETH price continues to go down, the feedback loop is only fixed if (1) more money comes into the collateral pool to create more Dai (this could be through new speculators entering the system or emergency sales of Maker tokens), or (2) people lose faith in Dai and no longer want to hold it, which means the system fails. There is no guarantee that (1) always happens. Note that emergency sales of Maker tokens doesn’t solve this: liquidity in these tokens could dry up similarly to how the speculator market can dry up.
Centralized stablecoins continue to grow and weather the crypto volatility well — as expected since they are collateralized by fiat currencies. They of course have their own problems of counterparty risk — essentially, we must trust bank-like entities to hold the fiat currencies in off-chain reserves. This trust was a problem blockchain was supposed to solve. These risks materialized this year with an incident in which Tether decreased in price to $0.85 (it has since recovered) and ~3% premiums on BTC/Tether markets vs. BTC/fiat markets, as shown below.
Stablecoins Going Forward
Current decentralized stablecoins have serious problems that are now materializing. However, their risks remain poorly understood. I’m not aware of any rigorous work on modeling these systems and better understanding their failure points. This presents a challenge to designing better systems. I’m working to address this in my research and a new stablecoin project, which I plan to write more about soon. Open to all ideas on how to move this forward.