The Evolution of Algorithmic Stablecoins
The market has made its desire for stablecoins clear as more of the world’s value moves on-chain, and this momentum shows no signs of stopping.
While the boundaries are not always clear, stablecoins can largely be grouped into fiat-backed (USDT, USDC, BUSD, etc.), crypto-collateralized (DAI, sUSD, mStable, etc.), and algorithmic, the focus of this piece.
Algorithmic stablecoins (algostables) are tokens that deterministically adjust their supply using an algorithm with the goal of moving the token’s price toward a price target. Their main advantage over fiat-backed stablecoins is decentralization, and their main advantage over crypto-collateralized stablecoins is capital efficiency. While dwarfed by fiat-backed and still behind crypto-collateralized stablecoins in terms of market cap, experimentation among new algostable projects has grown alongside the increase in usage of stablecoins in general. As different models have been tested and refined, they have become increasingly complex and have shown greater promise.
We will be examining the following protocols (all of which are pegged to the US dollar):
- Seigniorage Shares
- Basis Cash
- Empty Set Dollar
In 2014, Robert Sams published a paper presenting a method of stabilizing the value of a cryptocurrency now widely known as seigniorage shares. It is based on the Quantity Theory of Money, which states that general price levels in an economy are directly proportional to the amount of money in circulation. A corollary to this assertion is that a currency can be stabilized by expanding its supply in proportion to how far the current exchange rate deviates from the desired peg when demand for it increases and it rises above its target price. Conversely, when demand decreases and it falls below the target price, the supply is contracted.
This is accomplished through a smart contract that acts as a central bank whose mandate is to issue a currency and control its monetary policy such that the currency trades at $1. Expansion is accomplished by issuing new tokens when demand is high, while contraction is accomplished by auctioning “shares” when demand is low, burning the acquired tokens in the process. Holders of these shares receive inflationary rewards in future phases of expansion.
While the seigniorage shares model never took shape in the precise form that was put forth in the paper, it has, to varying degrees, served as inspiration for many stablecoin projects that followed it. We now turn to a few of these projects.
Basis is an adaptation of the seigniorage shares model that pioneers a three token model: an elastic-supply stablecoin, a debt-financing bond token that is a claim on future issuance of the stablecoin, and a value-accruing share token that is akin to equity in the protocol.
Like seigniorage shares, when price is below the peg, Basis attempts to take coins out of circulation by creating and selling bond tokens in an open auction. These bonds can be redeemed at a future date when the protocol enters an expansionary phase, granted that they don’t go unredeemed for five years, at which time they expire. Speculators can earn a premium by purchasing bonds with the stablecoin when it is below the peg, receive stablecoins at a future date when the price is above the peg, and sell them on the market to pocket the difference. The potential gain from buying bonds is greater than if stablecoins are simply purchased at a discount in anticipation of its price eventually returning to its peg.
When price is above the peg, additional stablecoins are issued. Bonds have preferential access to this new issuance and are redeemed in a first-in-first-out fashion, meaning purchasers at steeper discounts will have to take on the risk of waiting longer until redemption. If any new issuance still remains after all outstanding bonds have been redeemed, it is distributed to share tokens holders on a pro-rata basis as seigniorage.
Citing regulatory concerns, the Basis team ultimately decided to forgo their launch in 2018.
Basis Cash is a modified version of Basis deployed on Ethereum. Like Basis, it has three tokens: Basis Cash (BAC, the stablecoin), Basis Bonds (BAB), and Basis Shares (BAS).
Like many other algostables, Basis Cash runs according to a unit of time called an epoch that determines the length of intervals between monetary adjustments. Each epoch, the protocol checks whether the BAC price has moved out of a parameterized range that indicates the price is sufficiently stable to determine whether a contraction or expansion has been triggered.
BAB are sold at a price of P BAC, where P is the price of BAC in DAI (making the effective price P² DAI), ensuring a premium upon redemption. Unlike the original Basis project, the bonds have no expiration date.
BAB can be redeemed for BAC if the price of BAC is above 1 DAI and the treasury contract has a positive balance of BAC. Redemptions are disallowed when the price of BAC is below 1 DAI. This policy is for the purpose of preventing bond holders from prematurely cutting their losses and creating unnecessary downward pressure on the price of BAC.
Basis Cash launched by incentivizing actions that were beneficial to the network through distribution of BAS. Users who deposited stablecoins into a smart contract received BAC. Using the deposits, a BAC/DAI pool and a BAC/BAS pool were spun up on Uniswap and liquidity provision was incentivized through distribution of BAS.
Empty Set Dollar
Empty Set Dollar is another variant of Basis on Ethereum. Despite having only two tokens — ESD (its stablecoin) and coupons (essentially the same as bonds) — it functions similarly to the three-token model of Basis and Basis Cash.
Instead of share tokens, ESD (its stablecoin) can be staked in a DAO to receive a pro-rata share of new issuance. Unstaking requires a five-day lockup period in which the tokens can neither partake in inflationary rewards nor be traded, thereby incentivizing token holders to keep their tokens staked. Staked ESD therefore serves as a kind of share token with added illiquidity due to the time lock.
Marrying the functions of stablecoins and share tokens in this way makes the protocol’s value accrual inherent to the stablecoin itself. This makes no significant difference in terms of incentives compared to the three-token model since those of stablecoin and share token holders are aligned in both cases; stablecoin holders desire stability, and stability leads to growth of the network and thus inflationary rewards for share holders. However, it may potentially provide more protection against downward price movement as stablecoins being locked up means there is a smaller liquid supply that can be sold on the market.
Coupons have expiration dates (like the bonds in Basis) that last 30 days. This ensures that the system’s debt doesn’t accumulate indefinitely and is guaranteed to diminish over time. The downside is that it disincentivizes purchasing coupons because now, in addition to the risk that the peg will never be restored, there is the risk of the peg not being restored fast enough.
Debt accumulation is further suppressed through the limit on coupons exceeding 20% of the ESD supply; the protocol is unable to contribute buying pressure beyond this threshold. The only ways in which a position of 20% debt could be resolved is through raw demand for ESD, causing the price to go above the peg and coupons to be redeemed, or the expiration of outstanding coupons.
Certain mechanics were put in place to distribute the initial supply of ESD. Epochs are advanced manually by calling a function on the protocol’s smart contract, and the first user to call this method each epoch was rewarded with 100 ESD. Further, ESD’s price oracle was fixed at 1.1 USDC for the first 30 days, causing a 10% supply expansion every epoch that was distributed to bonded ESD holders. Providing liquidity to the ESD/USDC pool was also incentivized through inflationary rewards of ESD.
Problems with Uncollateralized Models
While there are minor differences in aspects such as governance, bond pricing, timing of redemptions, expirations, and epoch length in the uncollateralized models we have looked at, their stabilization mechanisms, which are arguably the most critical component, are essentially the same.
Expanding the supply when demand is abundant is relatively easy; the protocol can costlessly mint additional tokens to meet excess demand, which will naturally lead to selling pressure and bring the price back down to its target. The difficulty is in bringing the stablecoin back to its peg when it falls below it. In this case, the supply needs to be contracted. However, in contrast to how an expansion in supply can be executed by the protocol at will, the stablecoins are in the possession of users rather than the protocol. Thus, a mechanism is needed to incentivize users to relinquish their stablecoins to the protocol to be burned, thereby contracting supply and leading to a price increase back to the peg.
As we have seen, uncollateralized models accomplish this through bonds (coupons in ESD). Bonds are essentially a call option on the restoration of the peg. A greater deviation from the peg offers a greater opportunity for profit, but it also points to the protocol’s failure to maintain the peg and can damage confidence in the peg’s eventual restoration. If the stablecoin is found to be useful, demand will continue to rise and new tokens will be minted accordingly. As long as people have faith in its system, bonds will continue to be purchased whenever the price falls below the peg. However, if enough people begin to doubt its efficacy or simply lose their desire to hold the stablecoin, downward price pressure could cause a death spiral.
The fundamental issue is that the task of maintaining the peg in a crisis is wholly dependent on individuals making bets in favor of the longevity of the protocol. In other words, when there is a lack of raw demand for the stablecoin, speculators willing to wager that demand will eventually emerge are required to maintain stability. The strategy can be thought of as using the potential future growth of the system as value in the present. In order to meet the premium that bonds command, the money supply must grow beyond the pre-crisis level for bets to be profitable. The continued health of the system thus presupposes limitless growth. Such an assumption is not necessarily unreasonable; the monetary base of most world currencies have had decades of uninterrupted growth. Nevertheless, it is still an assumption, and one on which the stability of the system is wholly reliant.
Even when speculators do support the peg, they exact a large toll in that the protocol must compensate them. They are mercenary capital that will come when times are good and leave as soon as a better opportunity arises. In the world of DeFi where extreme APYs can be achieved, incentives must be fairly powerful in order to encourage certain behaviors and discourage others. Thus, the compensation must be significant and paid continually in perpetuity.
Since newly issuing tokens is costless, the price of tokens never remains above $1 for long. It often does remain below $1 for extended periods, however, since each dip requires speculators to step in. The protocol reliably exerts downward price pressure, but support from speculators varies. Thus, the system of bonds can work well when issuance is low, but breaks down when there are no buyers of last resort large enough to stabilize the protocol. Speculators naturally abandon this bet when the required growth becomes incredible, leading to a collapse. Ultimately, there is significant tail risk in depending on the whims of speculators for stability, and it is difficult to view this type of system as anything other than a game of musical chairs.
Unlike collateralized stablecoins, where assets are first provided before new stablecoins are minted, uncollateralized algostables require a zero to one, where value is created out of thin air. Lacking collateral means that there is no guarantee that the stablecoin can be exchanged for an equivalent value of another asset. In other words, there is no hard floor to its price. This makes it particularly difficult to bootstrap and convince people that the algostable is worth $1 to begin with. The hope is that $1 acts as a Schelling point that causes people to purchase bonds. However, in practice Basis Cash and Empty Set Dollar spent a miniscule amount of time near $1 and have instead experienced a slow but sustained decline. Some clever ideas that improved on seigniorage shares were implemented, but they were inadequate to overcome the fragility inherent to being uncollateralized. The two projects are now trying to remedy these issues in subsequent versions of their protocols.
Collateralized stablecoins require upfront collateral to create, at times making it difficult to meet demand. On the other hand, unbacked algostables require no collateral and can costlessly issue new tokens. While the idea of the ultimate capital efficiency in which no collateral is required to produce new stablecoins is enticing, in terms of risk, holding an unbacked stablecoin as opposed to a collateralized stablecoin is strictly worse from the user’s perspective. The benefit of capital efficiency is only felt by the supply side (i.e. the protocol) when tokens are minted. What then would incentivize a person looking to hold a stablecoin to choose an unbacked one over a backed one?
This brings us to algostables that utilize collateral in some form.
Protocol Controlled Value
A new cohort of stablecoin projects have recently emerged with mechanisms that can be categorized under the bucket of protocol controlled value (PCV).
In most DeFi applications (MakerDAO, Compound, Uniswap, etc.), users deposit funds and in exchange receive tokens representing an IOU. In contrast, with PCV the deposited assets cannot necessarily be withdrawn on the same terms with which they were deposited. The protocol controls the means of redemption and adjusts their parameters to align with the protocol’s objectives, akin to how central banks use their reserve assets to achieve their monetary goals.
The methods by which the protocol uses its assets to manage its stablecoin are called algorithmic market operations (AMOs). For example, in contrast to uncollateralized models, models that adopt PCV are able to use their collateral to provide liquidity. This liquidity is limited and is only provided insofar as it aligns with the protocol’s aims. In other words, solvency does not necessarily imply liquidity.
There are a number of benefits to utilizing PCV:
- Protocols that fully own their assets can become stagnant but can never die
- It reduces reliance on speculators to manage the stablecoin, and consequently reduces the need to pay third-party actors
- The protocol can incrementally increase its reserves by utilizing idle collateral assets in activities such as liquidity provision and yield farming
We now turn to algostable models that utilize PCV.
The Fei Protocol is a collateralized stablecoin project on Ethereum that utilizes PCV. It has two tokens: FEI (its stablecoin) and TRIBE (its governance token). The protocol aims to maintain a liquid FEI/ETH market that trades near the USD/ETH price.
FEI has an elastic supply that tracks demand. It enters circulation through a buy-only bonding curve denominated in ETH whose price function starts at a discount to reward early adopters of FEI in exchange for supplying ETH as PCV and approaches and fixes at $1.01.
Rather than using PCV directly as collateral that can be withdrawn at a $1 exchange rate at any time, it is deployed as “liquidity collateral” in the FEI/ETH Uniswap pool to create a secondary market where users can sell FEI for ETH. For each addition of ETH to PCV from the bonding curve, an equivalent amount of FEI is minted based on the pool’s current FEI/ETH price and these assets are added to the pool’s liquidity.
When price is below $1, trading activity in this pool is incentivized with rewards and penalties that drive the price towards $1. Purchases are encouraged through a FEI mint reward while sales are discouraged through a FEI burn penalty. Rewards and penalties both grow in proportion to the distance from the peg. Rewards also grow in proportion to the time that price has remained stationary, while penalties ignore time but grow quadratically with distance rather than linearly. The penalty for selling is always higher than the reward for purchasing, ensuring that trading below the peg has a net deflationary effect on supply.
Trading incentives are unnecessary when price is above $1 due to the stability being maintained through an arbitrage between the bonding curve and the Uniswap pool. When the market price exceeds $1.01, arbitrageurs can purchase FEI from the bonding curve at $1.01 and sell on the market. If the price is always at or below the peg, every trade would end below the peg and incur a burn penalty. Therefore, the bonding curve’s $0.01 buffer is essential for enabling arbitragers to sell FEI and provide stability without being penalized.
If the trading incentives below the peg prove insufficient to motivate speculators to bring the price back to $1 for a certain length of time, the protocol restores the peg by using its PCV to purchase and burn FEI. It does this through the following process:
- Withdraw all of its liquidity from the FEI/ETH pool
- If the pool still has liquidity, purchase enough FEI to bring its price back to $1
- Deposit all ETH and the corresponding amount of FEI back into the pool
- Burn the remaining FEI
If even the PCV is insufficient to backstop the price of FEI, TRIBE tokens could be minted to purchase FEI on the market and burned in order to recapitalize.
The protocol’s dynamics are nicely summarized in the figure below.
A key aspect in understanding Fei’s current situation is its launch event.
A period of a few days were set aside during which anyone could opt in to purchase FEI with ETH through the bonding curve explained earlier. The price started at $0.50, reached $1.00 at the target of 100,000,000 FEI sold, and was capped at $1.01 if the supply extended beyond this target. Upon completion of the sale, all participants received a pro-rata share of FEI at the same average price. Participants also received 10% of the TRIBE supply on a pro-rata basis and were given the option of pre-swapping their purchased FEI for TRIBE before bootstrapping a FEI/TRIBE Uniswap pool with 20% of the TRIBE supply.
While the burn penalty associated with selling FEI was designed to disable users exiting the protocol with a profit, the involvement of TRIBE tokens in the launch event created a loophole that whale bots took advantage of:
- Deposit ETH into the smart contract
- Receive 1 FEI per $1.01 of ETH deposited and a proportional amount of 10% of the total TRIBE supply
- Post-launch, sell TRIBE for FEI and sell all FEI for ETH
This trade is profitable as long as the gain from selling TRIBE is greater than the losses from minting FEI at a premium and selling it below the peg. This depends on how quickly FEI and TRIBE are sold, and FEI in particular since the more users who sell before you, the further the deviation from the peg and thus a higher burn penalty.
The dumping of both FEI and TRIBE resulted in an unfavorable situation for the remaining holders of the tokens. While the protocol technically remains overcollateralized, FEI is below the peg, TRIBE is below the pre-swap rate, and users don’t have the ability to convert their assets back into the amount of ETH that they invested. Even if the protocol allowed such a conversion, it would not be able to make all users whole due to the ETH extracted from the protocol by the whale bots.
Yet there is clear demand to exit despite the penalty; the price of FEI has historically only momentarily been at $1 a couple times before immediately plummeting. One therefore has to wonder whether being “overcollateralized” in this way is actually meaningful or implies what the term normally does. Indeed, the point of collateralization is to be able to exchange a stablecoin back into the collateral. In practice, no one can redeem their FEI for $1 worth of collateral while the peg is at or below the peg. The protocol severely limits liquidity and forces users to wait until demand pushes the price above $1 to be able to receive full value for their FEI. This system is advantageous to the protocol; it gains stability when FEI is purchased and burn fees when it’s sold. But from the user’s perspective, they are extortionate.
This raises the question of why anyone would want to lock up their assets in a system that charges you to exit in the first place. Supposedly, users are incentivized to exchange a volatile asset for a stable asset. But given that there was no history of stability prior to inception, many users who participated in the launch event likely did so believing it would be a good investment. In other words, they didn’t participate for stability (for which many alternatives were already available) but because they believed that they would be able to exit at a profit. Given the frustration expressed by the community post-launch, it is likely that most participants did not fully understanding the mechanisms at play. Even if FEI was able to successfully stabilize, the protocol would likely have a hard time growing since there exist plenty of other stablecoins where exits are not penalized. Further, lacking the foundational ingredient of stability, it has failed to produce utility and find viable use cases. Fei may be good at locking users in, but for that very reason it has difficulty attracting new entrants.
Another reason people may be reluctant to exchange FEI for ETH can be seen in what happens when the value of ETH fluctuates. The protocol retains all the upside if ETH appreciates in value, but if it depreciates in value FEI holders get to avoid this negative price movement. This may look like a symmetric tradeoff at first glance, but the second part of the equation is only true if FEI maintains its peg. Although Fei started off overcollateralized and has remained so throughout its lifetime thus far, let us imagine a scenario in which the protocol becomes undercollateralized due to a drop in the price of ETH. In contrast to crypto-collateralized stablecoins, the protocol does not liquidate its collateral when the total value drops below the total value of user owned FEI in circulation. What happens when the protocol becomes undercollateralized? An opportunity arises to purchase FEI at a discount to $1, so supposedly traders would sell their ETH for FEI. But this assumes continued faith in the protocol and value of FEI and that the peg will be restored. But the fact that it is undercollateralized directly contributes to doubt in the system’s future recovery. Thus, like the uncollateralized models, the protocol ultimately becomes dependent on speculators.
As it turned out, much more than the approximately $223M of ETH required to reach the ceiling FEI price of $1.01 was committed at the launch event — so much so that FEI/ETH and FEI/TRIBE are still the two largest pools on Uniswap. Despite this popularity, the resulting massive TVL, and a few DeFi integrations, the protocol has failed to achieve its stated objective of a $1 FEI until recently. Further, reweights and sell/burn mechanisms had to be disabled due to technical issues. Fei originally had plans to expand its AMOs as well as the set of tokens it would accept as collateral, but it is currently busy trying to ameliorate the issues it has experienced.
Frax is a fractionally algorithmic stablecoin protocol. It is partially collateralized, and the unbacked portion is stabilized algorithmically. It aims to become the de facto stablecoin used in DeFi.
Frax v1 launched with two tokens, FRAX and FXS. FRAX is the stablecoin and FXS is the value-accruing governance token. USDC was initially chosen as the sole form of accepted collateral.
The motivating idea behind Frax is to let the market decide the appropriate collateralization ratio (CR), the percentage of the outstanding stablecoins that are backed with collateral. The CR, which can range from anything between 0% (fully uncollateralized) and 100% (essentially wrapped USDC), is programmatically adjusted to track the demand for FRAX. It changes according to the price of FRAX relative to a price band around $1:
- If the price of FRAX rises above the band, the CR decreases to enable additional FRAX to be minted with less collateral
- If the price of FRAX falls below the band, the CR increases to supplement collateral reserves and restore confidence in the system
- If the price of FRAX is within the band, the CR is adjusted based on the market cap of FRAX and the liquidity of FXS (since a larger FRAX market cap and less FXS liquidity imply greater fragility)
When the CR is changed, the protocol will execute a buyback or recollateralization to make the amount of collateral match the CR. When there is excess collateral, it is used to purchase and burn FXS on the market. When there is a collateral deficit, FXS is minted and sold to acquire additional collateral. The net effect is for the price of FXS to increase when there is excess collateral and decrease when there is a deficit.
The protocol allows users to mint and redeem FRAX for $1 worth of USDC and FXS at any time. Their ratio is determined by the current CR. For example, to mint $10 worth of FRAX when the CR is 90%, users would send $9 worth of USDC and $1 worth of FXS to the protocol (along with a minting fee). The collateral is added to the reserves and the FXS is burned. Redemptions work the same way but in reverse; in exchange for FRAX, users receive USDC from the reserves and newly minted FXS in proportion to the CR (minus a redemption fee).
Since the protocol is a counterparty always willing to trade at $1, users are incentivized to arbitrage FRAX when its price deviates from the peg. When the price of FRAX is above $1, users can mint FRAX through the protocol for $1 each and sell it on the market to make a profit. Users can likewise profit when the price of FRAX is below $1 by purchasing FRAX on the market and selling it to the protocol for $1 worth of collateral.
A protocol can never issue a stablecoin in exchange for only partial collateral since this would immediately create an arbitrage opportunity by using the stablecoin to buy more of the collateral than was provided, ultimately pushing down the price of the stablecoin until it reaches $1 multiplied by the % of collateral provided. The reserves, however, can be fractional. A fractional reserve creates a stable price floor that is higher than 0. The gap in the collateral is filled by people’s perceived utility of the system. This utility can be thought of as latent demand (i.e. people willing to buy if price goes below $1). In uncollateralized models, utility is the only thing backing the price. In partially collateralized models, you only need enough utility to cover the collateral gap.
Frax started out fully collateralized, solving the bootstrapping problem faced by unbacked algostables of infusing a token created out of thin air with fixed value. Liquidity provision for its tokens is incentivized, with longer lockup periods of LP tokens providing greater rewards. This attracts participants early on when the network is still small. As it grows, new users can be drawn in with smaller incentives because there is less perceived risk in the system and the CR can be lowered because of the stablecoin’s increased utility.
In addition to the stability mechanisms outlined above, Frax v2 introduced AMOs to utilize idle collateral as well as minted FRAX. Collateral is only used in applications that don’t require a lockup so that it can be withdrawn at short notice to service FRAX redemptions. A conservative upper limit is placed on the amount of FRAX that the protocol allows to be minted for use in AMOs. It is the amount such that, if it were all immediately sold on the mrket, it would bring the price of FRAX down to the edge of the price band. In other words, it is the maximum amount that is not enough to cause the CR to change. If there is ever a conflict between the base stability mechanism and an AMO, stability is prioritized and the AMO is paused.
Collateral and minted FRAX are deployed through AMOs to earn yield, transaction fees, and interest. This added value has the effect of increasing reserves. The protocol therefore mints new FRAX, expanding the FRAX supply and bringing the CR down to the level determined by the market. This FRAX is used to purchase and burn FXS on the market.
Here are some live as well as proposed AMOs:
- Collateral Investor — Invests idle collateral in DeFi protocols
- Curve — Operate a stableswap pool with liquidity controlled and owned predominantly by Frax to provide FRAX/USDC liquidity and earn fees and CRV rewards
- Lending — Directly lend minted FRAX and earn interest from borrowers through existing money markets
- Collateral Hedge — Perfectly hedge collateral with a short position to create a basis trade of net 0 volatility
- Tornado Cash — Mint FRAX directly into Tornado anonymity sets to expand privacy features
- Bonds — Issue tokens on an x*y=k curve and pay its holders an interest rate so they can redeem their bonds at higher prices. Alternatively, bake this functionality into staked FXS
Frax has started out by only accepting USDC to dampen volatility in the collateral so that it can smoothly transition to lower CRs. As the velocity of the system increases, it becomes safer to include volatile cryptocurrencies such as ETH and wrapped BTC. Using decentralized assets would eliminate the regulatory, censorship, and counterparty risks associated with USDC.
Since USDC is tightly pegged to $1, the small fluctuations in its value have essentially no effect on the CR or the value of FXS, and any change to the CR is largely a function of demand for FRAX. The situation changes with volatile collateral. If the price of the collateral rises, the excess value will be distributed to FXS holders. If the price of the collateral drops, FXS must be minted and sold to acquire more collateral and keep the CR at its target. This adds a new dimension to tracking the CR, as FXS is effectively tasked with absorbing not only the fluctuations in demand for FRAX but the volatility of the collateral as well in order to maintain FRAX’s peg at $1. FXS will thus experience greater volatility, and since it is the final backstop to prevent a collapse in the price of FRAX, having sufficient liquidity becomes crucial for the reliability of the system as a whole. This can be mitigated through the collateral hedge AMO mentioned above, but this relies on the existence of other well-functioning derivatives protocols and has not yet been implemented.
The advantage of Frax over overcollateralized models is that it can create more stablecoins with less capital. However, it is at an inherent disadvantage to USDC, layering smart contract risk and price risk on top of centralization risk. FRAX is more capital efficient than USDC from the perspective of the protocol since it does not have to be fully backed and it also has the ability to mint a limited amount of FRAX with no collateral for use in AMOs. But from the perspective of the users minting new tokens, there is no difference; both USDC and FRAX require a full dollar’s worth of collateral to be minted. Using a stablecoin to mint another stablecoin that is only partially collateralized and takes a fee in the process seems self-defeating. USDC is also more price-stable and has wider integrations with DeFi. For users looking to simply hold a stablecoin, USDC is clearly superior.
But the reality is that not all users acquire stablecoins merely for the prospect of price stability, and FRAX is a case in point: users are currently incentivized to mint FRAX and provide liquidity on FRAX pairs to receive inflationary FXS rewards. Yet each decrease in rewards will test whether holders still consider them enough to hold FRAX over USDC. Once rewards cease, users will only prefer FRAX if it has a better combination of stability, liquidity, use cases, and collateral. If it fails in this regard, holders will rush to swap their FRAX for USDC as there is only downside to holding wrapped USDC that is undercollateralized.
At first glance, it may seem like holders of FRAX always have a safe exit even if its price falls below $1 since the protocol is a counterparty always willing to trade FRAX for $1 worth of assets. However, one has to keep in mind that what is received in exchange is partially USDC and partially FXS. Assuming USDC has no issues, there must be willing buyers of FXS. But the protocol doesn’t take liquidity into account; it merely provides the appropriate amount of FXS at its current price. If there is no liquidity, FXS may technically be worth the price that the protocol redeemed FRAX for, but it cannot be converted into other assets without significant slippage, rendering its price nonmeaningful and the exchange with the protocol unfavorable.
The FXS liquidity currently on the market is precisely the limit to the amount that the protocol can increase its reserves. Therefore, purchasers of FXS are the lenders of last resort. If the market brings the CR back to 100%, the protocol may not be able to fully recollateralize; value taken out of the system by users selling FXS at a profit is permanent. In a crisis, the bag holders would be the FXS holders who didn’t sell above their purchase price and FRAX holders who didn’t sell while there was sufficient FXS liquidity. In other words, the reflexive dynamic of those who exit later being punished more severely would be at play in Frax as well.
As we have seen, FXS accrues fees, excess collateral value, and earnings from AMOs, and its value is highly correlated with demand for FRAX. It can therefore be thought of as storing the value of the protocol’s utility, and it’s market cap and liquidity serve as proxies for the protocol’s health. Unfortunately, the amount of value stored in FXS is opaque since it depends on a wide variety of variables such as minting and redemption of FRAX, AMO earnings, and inflationary rewards (which currently outpaces deflationary burnings). We would like to know whether the value coming into the system exceeds the value going out, but this cannot be calculated. As a result, unlike overcollateralized stablecoins you cannot analyze how much downward pressure FRAX can take before it reaches the point of no return.
In the final analysis, while Frax is much more resilient than uncollateralized algostables, in a crisis it still requires speculators to step in. Since the partial collateralization mitigates the potential downside for speculators, it is reasonable to expect them to provide support more readily. Nevertheless, the fundamental weakness of these systems are the same; it only takes a single event in which there are not enough speculators to maintain the peg to initiate an irreversible downward spiral.
Regardless, FRAX has been able to maintain impressive stability from its inception until today. It has been integrated with DeFi protocols such as yEarn, Aave, Compound, Cream, and Curve. Notably, FRAX has become the first algostable to make it onto Curve. The number of FRAX trading pairs has been increased by partnering with select projects that want to incentivize their stablecoin LP farms with FRAX. The project launched on Ethereum but has since bridged to multiple other chains. Earnings from AMOs have gradually expanded the FRAX supply. Thus far, its thesis appears to be playing out.
Humans are memetic creatures that tend to copy others, and this is just as true for market behavior as it is for anything else. Price movements in markets tend to have momentum, and providing incentives for agents to act counter to that momentum is not easy. Algostables can be highly reflexive, and much of their demand is driven by market sentiment. This can be a boon during a project’s early stages when it is looking to grow, but it can also be its undoing if instability gains a foothold. The community around a project can play a significant role in shaping sentiment and forming a collective conviction that tokens won’t be sold en masse.
Once an algostable is successfully established and adoption reaches a certain level, it may go unscathed from events that would have killed it earlier on in its lifetime. Tether was in a legal dispute with the NYAG, which in a public lawsuit claimed that USDT is only partially backed. But nothing happened to the price of USDT. USDT had tight integrations with the crypto community, being highly liquid and supporting many crypto pairs. Users considered it to be money, and that gave it a monetary premium over and beyond its supposed collateral deficit.
This demonstrates that utility is paramount and, at the end of the day, stability comes from adoption. The US dollar isn’t stable because it has the best monetary policy. Rather, it is stable because it is the most widely used fiat currency in the world.
The mechanism employed by an algostable is only half of the problem. Ultimately, the ingredients for stability require an incentive compatible mechanism, but it also requires an ecosystem of usage that creates organic demand for the stablecoin. A mechanism alone cannot bootstrap this, and it is therefore indispensable for projects to actively foster non-speculative use cases.
The algostable design space has only begun to be explored. The potential market for a decentralized, capital-efficient stablecoin is huge, and new experiments are bound to continue emerging to capture a piece of the ever-expanding stablecoin pie.
Written by @fractalfren