By: Joe Kendzicky

Twitter: @JKendzicky

Intro

Decentralized stables are currently one of the hottest segments of DeFi, with a collective TAM of ~$100B locked within USDC and USDT alone. Such a heavy reliance on centralized systems seems to contradict the ethos of the industry, creating a huge opportunity for protocols that can offer a truly decentralized solution. Not surprisingly, a flood of projects have emerged seeking to do just that. In this report we will highlight some of the frontrunners in the decentralized stablecoin space and unpack their design models. Additionally, it is our goal to demonstrate that, similar to the L1 ecosystem- where a fundamental trilemna exists between throughput, security and decentralization- there is no ‘perfect’ stablecoin construct.  This is not to disregard the fact that certain architecture models are objectively better than others, but at the end of the day protocols must incur trade offs across 4 key properties: stability, capital efficiency, decentralization and scalability.

MakerDAO (DAI Stablecoin)

One of the oldest DeFi projects to date, MakerDAO is a decentralized credit protocol that collateralizes various crypto assets and mints debt against it. This debt comes in the form of a stablecoin called DAI, and Maker utilizes a system of external market forces, incentive mechanisms, and policy tools to maintain a 1:1 peg of Dai against the dollar. Let’s unpack.

Similar to a bank, the user will bring some collateral to the table in order to facilitate a loan against. In a bank setting, the borrower could bring a house, a car, receivables etc to the table.

With a loan against a digital asset, the value of the loan must always be less than the value of the collateral, to protect the lender against default risk, or liquidation risk, depending on how liquid the collateral is. To protect against this risk the borrower has the option to deposit various collateral assets (ETH, UNI, LINK, YFI etc) in order to mint debt. Various assets are subject to various margin requirements based on risk. This is pretty intuitive: less liquid, higher volatility assets will be harder to liquidate at scale. Thus, a large buffer needs to be applied to facilitate a smooth unwinding of the position, in the event the collateral falls below margin requirement.

Let’s give a quick example of margin requirement. Currently minting DAI in the ETH-A pool requires a 145% minimum collateral ratio. Suppose you deposit 1ETH, 1ETH = $4,000, and you take out a 1,000 DAI loan. Your collateral ratio is 4,000/1,000 = 400%. If ETH starts crashing and falls below $1,450/ETH, then your vault will be subject to liquidation. During liquidation, the protocol sells off your collateral at a discount to market rate, giving an incentive for an arbitrage to come in and retire the debt.

So how do we retire debt? Well, we can’t really repossess the DAI that was issued to the borrower (DAI is an ERC20 token protected by user private key, plus the borrower may have already transferred it to someone else). Instead, we sell the ETH collateral in the open market in exchange for DAI. Then, we send back the 1,000 DAI to the protocol to be burned, closing the outstanding debt balance. Again, the incentive for the arbitrageur to perform this function is that they receive a small haircut of the collateral during the liquidation process.

What’s really cool here is that the protocol is permissionless; no credit history check, no KYC process, no credentialing whatsoever; so long as you meet the margin collateral demands the system doesn’t care who you are, nor can anyone hinder you.

Another cool feature is that due to the permissionless nature, the borrower is sort of serving as both the creditor and borrower simultaneously. There isn’t really a direct counterparty on the other end of the agreement (contrast this to traditional finance where the bank serves as the direct counterparty in the event of a loan default).

Instead, default risk is aggregated across a tranche of system participants, as illustrated below:

If the protocol is not able to auction off the borrower’s margin collateral in time before the vaults debt > the vaults collateral, then the system will mint MKR out of thin air and sell it into the market until the outstanding vault liabilities are recapitalized. If for whatever reason MKR liquidity evaporates and there are no marginal buyers left willing to purchase MKR and liabilities in the vaults still remain, then DAI holders ultimately bear the cost as their DAI is no longer fully backed, and the markets reprice it accordingly.

Abracadabra (MIM Stablecoin)

Fundamentally, the Abracadabra protocol looks incredibly similar to the system mechanics of the MakerDAO protocol. Debt (MIM) is permissionlessly minted by overcollateralizing a vault. If the collateral: debt ratio falls below a certain threshold then that debt is retired by liquidating the borrower’s collateral in the open market, and using it to buy back MIM to close the vault.The primary difference between MakerDAO and Abracadabra is that Abracadabra vaults are funded by depositing interest bearing tokens, such as yvYFI, yvUSDT, yvUSDC, xSUSHI etc. Interest bearing tokens are simply their vanilla counterpart (YFI, USDT, USDC etc) deposited in a money market protocol, thereby generating continual interest. The interest token is just an accounting token demonstrating how much claim on collateral the user has in the money market protocol.

So, instead of simply having these interest tokens sitting passively in Yearn or Compound, we can use them to mint MIM debt against. An advantage Abracadabra offers is that because collateral in the vault is generating yield over time, the loan: value ratio tends to decrease. From a user perspective, this is nice, because it means their debt obligations are reduced over time. The user can then use this reduction to mint more MIM without needing to post additional collateral (and still maintain the same loan:value ratio that they originally started with). Or, they can keep their MIM debt the same and just have a lower real debt burden to pay back when they go and close the vault.

And unlike MakerDAO, which charges a variable interest rate on open vaults, Abracadabra offers fixed rate interest. Abracadabra also exists on multiple chains, allowing native MIM exposure cross chain for users. A drawback however is that MIM introduces additional smart contract risk, as the collateral is ultimately held in the yield generating protocol outside of the Abracadabra. Thus, a contagion event in an external protocol will have a direct effect on users.

Alchemix (alUSD)

Similar to Abracadabra, Alchemix is a debt protocol that allows users to post collateral to a vault and mint stablecoin debt against it. Both protocols involve depositing yield generating assets (yearn assets) as collateral. The key selling point Alchemix offers is non-liquidatable loans. Let’s explore how the protocol is able to offer this:

Alchemix initializes with the user overcollateralizing y-asset into the Alchemix protocol, minting alUSD debt against it (with a 200% minimum collateral-to-debt ratio). The only collateral accepted to mint alUSD is yDAI. This creates a 1:1 correlation between the volatility of the collateral, and the volatility of the debt.

The whole reason why liquidation exists in protocols like MakerDAO or Abracadabra is to prevent the system from accruing negative equity. If I deposit $200 of ETH in a Maker vault to mint $100 of DAI, and the value of that ETH subsequently falls to $95, the system no longer has enough credit to guarantee that Dai’s $1 peg. Now, each Dai only has a collateral backing of $0.95. This is the inherent dilemma any debt protocol encounters when using volatile collateral to backstop non-volatile debt.

However, if we revoke volatility of the collateral, we no longer face this challenge and can remove liquidation requirements, bringing peace of mind to users about no longer needing to monitor margin requirements of the vault.

In addition, due to the fact that our collateral is a productive asset generating yield for us behind the scenes, we can use this accrued yield to actually pay off the underlying principle of our loan over time! This is how Alchemix is able to facilitate the concept of “self-replaying loans”.

If there’s no threat of liquidation, why require overcollateralization?

Without it, we run into a recursive loop where a depositor can mint unlimited debt. For example:

1. User deposits 100 yDai into Alchemix, mints 100 alUSD debt
2. Sells 100 alUSD on Curve.fi for 100 additional Dai
3. Deposits new Dai proceeds into Yearn, receives 100 yDai
4. Takes new yDAI back to Alchemix, mints 100 more alUSD
5. Repeats cycle indefinitely

Overcollateralization of 200%  caps the total recursive leverage at 1x the deposit. Meaning that if a user deposited $100 of yDAI into the protocol, they would never be able to mint more than $100 of alUSD debt.

UXD Protocol (UXD)

UXD is a Solana based protocol that offers a novel composition structure from any previous decentralized stablecoin alternative. The key innovation here is that UXD uses on-chain perpetual futures contracts (also known as perpetual swaps) to create an on-chain market neutral derivatives position that isolates directional risk and produces a stable $1 value. Let's unpack what all that means.

Derivatives markets such as bitmex, binance, FTX  allow users to trade synthetic contracts, called perpetual swaps, that mimic the spot market. For example, Bitmex’s XBTUSD allows traders to get exposure to BTC/USD price without having to physically own that Bitcoin. Perpetual contracts have generated tremendous adoption in the crypto space primarily due to their simplicity, convenience (no expiration, so no need to roll the contract), and most importantly, leverage offerings. While many have scaled back in recent months, 100x leverage was commonplace at one point for almost every exchange in the space.

Every synthetic contract faces an inherent dilemma- how can our instrument maintain price parity with the spot market we are trying to tether our derivative to?

While the objective of XBTUSD contract is to mimic the behavior of BTCUSD as tightly as possible, ultimately there will never be perfect union of the two prices, unless there is perfect supply/demand equilibrium between the spot and derivative orderbooks: an impossible assumption. What happens if the buy pressure on Bitmex explodes, but remains relatively constant on the spot? In such a scenario, the price of the XBTUSD contract would increase, while the BTCUSD index would remain flat. The derivative has deviated from the underlying oracle.

To overcome this situation, Bitmex introduces an incentive mechanism called “funding rate” which naturally aligns the peg. Funding is a reoccurring 8-hour interest payment, paid by the contract side where the premium exists.

If funding is positive (i.e. derivative price > spot price) then longs will pay shorts the premium. If funding is negative (i.e. derivative price < spot price) then shorts will pay the longs the premium.

Example : Suppose the price of spot index is $50,000/BTC, while the derivative trades at $55,000/BTC. In this scenario, funding is positive (derivative price > spot price). Thus, XBTUSD longs would pay an interest rate to maintain their position, with the proceeds accruing to their counterparties that are short XBTUSD. Longs are paying for counterparty liquidity. This relationship would exhibit in the opposite direction if there were more short demand than long demand; shorts would pay longs.

The greater the differential between the derivative and spot prices, the higher the interest payment. As that interest rate floats higher, it incentivizes more counterparty liquidity to step in and take the opposite side of the trade to collect that fee.

But what happens if the price of bitcoin is rapidly appreciating? Sure, you are set to make a marginal interest payment every 8h, but what if the loss of principle on the short is drastically outpacing your accrued funding payment?

Enter Arbitrage

An interesting arbitrage opportunity exists where traders can hedge away directional risk, isolating the interest payment from funding. Here’s how it works:

Suppose the trader bought $100,000 of BTC and then opened up a 1x short position with it. To calculate how much interest he would earn/pay, we would simply multiply the interest rate * the notional size of the trader’s position. Assuming the “default” funding rate of 0.01%, over the 8h window the trader would receive:

$100,000 * 0.01% = $10

This might not seem like a lot at first, but what if we annualize this value?

$10 * 3 = $30/day

$30 * 365 = $10,950/ year

$10,950/ $100,000 = 10.95% for a directional risk-free trade

0.01% is the baseline default rate, however many times funding payments can be much higher. Historically, this trade has annualized out in the range of 20-30% APR. During a bull run, it can be much higher, as demand for leverage skyrockets, and participants are willing to pay a large premium for that counter-party liquidity.

UXD Mechanics

Described simply, the UXD protocol receives capital, and automates this arbitrage strategy on behalf of its users. If a user sends in 1 BTC to the protocol, UXD will transfer that 1 BTC to a decentralized derivatives trading platform (currently starting with Mango Markets, with plans to expand into other markets in the future) and subsequently execute a 1 BTC short using that collateral as margin for the trade. Assuming the market price of BTC trades at $50,000, this effectively locks in a $50,000 USD stablecoin position. If the spot price of BTC goes up, our short position incurs a negative PnL, but it is offset by the linear rise in the BTC collateral value. Likewise, if the spot price of BTC goes down, our short position incurs a positive PnL, but it is offset by the loss in BTC value underpinning it. The total value of our derivative position stays fixed at $50,000. UXD then issues 50,000 tokens to the user, representing their claim on collateral from this position. To redeem their UXD tokens for collateral, they would simply send back the tokens to the mint contract, the protocol would unwind the derivatives trade on Mango markets, deliver the BTC collateral back to the user, then subsequently burn the UXD tokens.

Now, the market neutral derivative position is typically accruing an interest rate this whole time. What happens to those funds?

This component is still being worked on by the community governance process, but it's expected that a portion of those proceeds will accrue to the UXD insurance fund, to offer system protection in the event of a liquidation crisis (more on that later). The other portion will accrue to UXP holders, which is the UXD protocol’s native governance token. So, UXP offers a quasi-equity claim on protocol cash flow accrued through delta neutral interest. Additionally, there is talk about offering a portion of this accrued interest to  UXD holders directly, offering them an ability to collect automatic yield on their stablecoin holdings, akin to if they deposited it into Compound, AAVE or Yearn.

FRAX

Frax in its current state can be best described as a hybrid centralized/ decentralized stablecoin. Frax’s core innovation is that it uses a credit-based model, determined by various external market forces, to collateralize the stablecoin.

In brief summary, the protocol uses an algorithm to perform 3 various functions:

1. Increase the collateral ratio
2. Decrease the collateral ratio
3. Maintain current equilibrium

Each FRAX is backed by a certain % of USDC collateral. At system onset, in order to mint 1 FRAX, a user had to deposit 1 USDC into the protocol. Over time however, the price of FRAX began to fluctuate above or below $1. This isn’t an uncommon feature, in fact, both USDT and USDC experience similar variability in their exchange rate. It is a process of arbitrage that ultimately reconverges the stablecoin back to parity.

If the price of USDC = $1.05, an arbitrage agent will mint USDC by depositing $1 of USD into Circle’s bank account. Circle takes that USDC and sell it on the market, collecting a $0.05 risk free profit. Each time the arbitrager sends more USDC into the market, they are increasing the supply, and pushing price down. They will keep iterating through this cycle until 1 USDC trades back at its intrinsic peg.

The reverse arbitrage exists if 1 USDC = $0.95. They will buy USDC on the open market, send it to the USDC contract to be burned, and redeem the USD in Circle’s bank account, pocketing a $0.05 spread. As they remove USDC from the market, supply is reduced, pushing price up. They will iterate through this cycle until 1 USDC trades at its intrinsic peg. **
**

FRAX works differently. When the price of FRAX deviates from its $1 peg and stays there for a certain period of time,  an algorithm will kick in which either increases or reduces the collateralization requirements for the creation of a FRAX.

For example, when the price of FRAX goes to $1.05, this signals to the algorithm that demand is outweighing supply, as market participants are essentially purchasing an asset at a 5% premium to par value. The USDC centralized model lets arbitrage agents resolve the situation organically by increasing the total supply of USDC, and collect a risk free profit by doing so. Frax operates differently. Instead of increasing the supply of FRAX, the protocol responds by not only increasing the supply of FRAX via arbitrage, but also reducing the collateral backing outstanding FRAX float.

As a simplified analogy, imagine a restaurant is faced with a 5% increase in food costs. They are faced with a few choices to pass along those expenses to the end consumer:  raise meal prices by 5%, reduce meal portion sizes by 5%, or a combo of both. We can think about FRAX responding to supply/demand characteristics in a similar way. When price is above peg, supply is increased by arbitrageurs by some proportion (increasing the cost of the meal), while the remaining proportion is captured by the protocol via a collateral reduction (making the meal size smaller).

An inherent feature to the Frax protocol is FXS, the protocol governance token that plays a crucial role in facilitating system credit. FXS accrues value during periods of credit expansion , and bleeds value during periods of credit compression.

The best way we can think about the role of the FXS token in the ecosystem is that FXS essentially serves as a credit backstop mechanism that ensures full collateralization of the FRAX stablecoin. Even though the amount of USDC backing each FRAX may only be 83%, that does not mean that FRAX is undercollateralized. Rather, that remaining 13% void is replaced by an implicit credit guarantee of FXS token backstop: if the market starts demanding a higher USDC collateralization ratio (expressed by FRAX trading below $1 for some extended period) then the protocol will respond by minting more FXS out of thin air and using the proceeds to acquire USDC in the open market, which it can then use to beef up the USDC collateralization ratio of FRAX token.

Throughout the lifecycle of protocol, we’ve seen FRAX trade > $1 more often than not, which has allowed the protocol to algorithmically wean off USDC over time. Currently each FRAX in existence is only 83% backed by USDC. As noted above, this value could change in a moment's notice (though there is an inherent time delay built into the algorithm…i.e. it couldn't flip to 100% USDC collateralization in minutes).

The primary risk FRAX faces is a FXS liquidity crisis. If buy side demand for FXS were to disappear for whatever reason, the protocol could continue minting more and more FXS to try and increase the collateral ratio (and thereby restore the peg) but without buyers it would have diminishing marginal returns. At such a point, the protocol would be trapped without a way to acquire additional USDC to add to the reserve. Market participants might recognize this and begin selling their FRAX into alternative stables, pushing the price below the peg and amplifying the problem even more.

The following graphic illustrates an expansionary period, ie when the market price of FRAX > $1:

• Algorithm recognizes demand < supply, increases USDC collateralization by 0.25%
• FRAX price trading below $1, arbitrageurs
• To mint new FRAX, minter must deposit $83 of USDC and $17 of FXS
• $17 of FXS token is burned after the procedure, price of FXS rises as a result

Fei

Fei is another algorithmic stablecoin that shares some similar characteristics with FRAX. However, unlike Frax, FEI aspires to be a fully trustless stablecoin without any reliance on centralized intermediaries (USDC, USDT etc). Like the Frax protocol, Fei protocol introduces 2 separate tokens in its design architecture: FEI (stablecoin) and TRIBE (governance token).

Original mechanics operated by utilizing protocol controled value (PCV) to maintain the peg. PCV is just a fancy term for a protocol selling its native token ( FEI) in exchange for some exogenous capital (ETH). The protocol now owns that ETH collateral and can deploy it as it sees fit. Contrast this to the liquidity mining model standardized by most Defi projects where the protocol is essentially renting liquidity at an expensive rate, and forced to pay out high APYs to attract mercenary capital. Once those subsidies die down, liquidity evaporates and the protocol has nothing tangible to show for it.

In the original design, a user would deposit say $100 of ETH and in return the protocol would mint back 100 FEI via a bonding curve. However this mint process of FEI via the bonding curve was a one way channel; if you wanted to sell back your FEI for ETH you'd have to do so through a secondary market, such as Uniswap. The system used something called direct incentives to maintain the peg.

At equilibrium 1 FEI = $1 of ETH collateral, but what happens if this inherently volatile ETH collateral fell below $1? The Uniswap pool has a custom parameterization where selling FEI while FEI traded its $1 peg, incurred a penalty.  Penalty was calculated by doubling the delta b/t $1 and the current FEI price. So, if FEI was at $0.98, you'd be charged ([1.0-0.98]*2) = $0.04 penalty.

Likewise if you purchased FEI while below the peg, you got a premium, this premium coming from the surcharge tax on the seller mentioned. So in this instance, if you minted 1 FEI at $0.98, you ended up getting $1.02 worth of value. These features comprised the direct incentives architecture.

These mechanics proved to be inefficient at maintaining price parity with the preg breaking on numerous occasions and having trouble converging to equilibrium.  Since then, a v2 has emerged that utilizes a fully different approach to keep FEI pegged at $1.

V2

The system scraps direct incentives as the primary module for maintaining the peg, replacing it with PCV yield allocation and TRIBE token backstopping. The protocol has reduced exposure from just 2 narrowly concentrated assets (ETH and FEI) to now 7 (FEI, ETH, DAI, RAI, INDEX, DPI, LUSD). The goal here is to spread risk across a wider asset base, to reduce downside volatility and the threat of de-collateralizing the existing FEI peg if ETH price were to crash. In addition, the PCV is being deployed across a variety of money market protocols (AAVE, Compound, Lido, Tokemak ) to earn continual interest on the collateral, thereby boosting the collateral ratio. In a way, the FEI protocol almost looks like an investment DAO, with an investment mandate to keep the system PCV > user circulating FEI.

If the protocol fails in that department, then TRIBE holders serve as the backstop mechanism to keep price aligned with the $1 peg. If under collateralization occurs (ie $ value of PCV < $ value of user circulating FEI) then TRIBE will be minted and sold on the market to raise capital to finance the gap.

Likewise, if the PCV: Circulating FEI ratio continues to expand, TRIBE tokens will benefit from direct value accrual, as a portion of the PCV will be used to purchase TRIBE on the open market and subsequently burned, driving price up.

Terra (UST)

UST is a decentralized stablecoin whose price is backed by $LUNA, the native token of the Terra blockchain. At the heart of the Terra ecosystem lies the algorithmic market module that balances system incentives to stabilize the peg.

2 components exist:

• Terra UST: Dollar pegged stablecoin
• LUNA: underlying gas token of the Luna smart contract blockchain (ie ETH in Ethereum)

The LUNA token essentially absorbs volatility of the UST stablecoin. When the price of UST trades outside of its peg, . This is where incentives kick in; the algorithmic module will incentivize arbitrageurs to mint more UST supply, thereby pushing the price back down to parity. New supply can be minted by converting through the module, which will always mint/redeem  1 UST for $1 of LUNA, regardless of the broader market price for UST. This gap between the prevailing market rate and the rate that the module will mint at offers a risk free profit for the arbitrager, while simultaneously assisting in maintaining $1 parity of the peg.

Let’s work through an example. Suppose the price of UST on Binance trades at $1.05

• Arbitrager would acquire $1 of LUNA token and send it to the protocol module
• Module would burn that $1 of LUNA token, and simultaneously mint 1 UST
• Arbitrager would then send that 1 UST to binance and sell it for $1.05, pocketing a 5% spread

End result: UST float expands while LUNA float contracts. UST expansion coincides with a LUNA contraction, leading to value accrual for LUNA token holders as price appreciates. LUNA holders are cheerleaders for UST growth and integration across DeFi, as they are the direct value accrual recipients of its adoption.

Now, suppose the price of UST on Binance trades at $0.95

• Arbitrager would purchase 1 UST on Binance for $0.95 send it to the protocol module
• The module will always honor a redemption rate of $1 for UST, regardless of broader market price. Arbitrager swaps 1 UST for $1 of LUNA.
• Arbitrager then sells $1 of LUNA for USD, pocketing $0.05 spread

End result: UST supply contracts while LUNA float expands. UST supply contraction pushes price back down to its $1 peg. LUNA token holders suffer a value dilution in the process, as their current holdings are diluted by the new LUNA inflation.

The UST/LUNA interchange is subject to highly reflexive feedback loops in both directions. UST has demonstrated a relatively  strong track record will be interesting to see how it fares in a bear market when increased pressure threatens the peg. The biggest concern is a death spiral situation where LUNA liquidity dries up and goes no bid. With no route to unwind LUNA proceeds after burning UST at the module, arbitrageurs evaporate and the peg breaks

Analysis

Stability

Stability can be described as the effectiveness of a protocol to maintain its $1 peg. Maintaining parity when price is > $1 is easy, getting it up to $1 when trading below is the tough part.

Having an efficient arbitrage process is the most crucial component of stability. The faster this arbitrage process can be completed, the more robust the protocol, as it reduces the scope of time price outside the peg. In the ideal setting, when even the slightest deviations occur, we want professional arbitrage to be able to come in and extract that profit, pushing price back in line immediately. Long buildup of deviation outside the range increases protocol risk, causes users to lose confidence in the market mechanisms, and compounds downward reflexivity.

Rather than trying to ascribe an ranking index for stability (too subjective), we will talk about the strengths/weaknesses of each project with respect to stability:

Dai : Keeper bots are incentivized to call liquidation functions against undercollateralized vaults. Overcollateralization keeps DAI price from trending below $1 but historically has had a tough time preventing price from trending above, as no direct arb exists when DAI > $1

MIM : same premise as Dai

alUSD : Again, no direct arbitrage exists to re-equilibrate alUSD when out of band. alUSD in particular is faced with a challenge where in natural setting, more sell pressure exists than buy pressure, pushing price below $1 regardless of whether the collateral exists to support its intrinsic value of $1. This is because people minting alUSD are primarily doing so to acquire leverage, where after minting users will immediately sell their alUSD for X asset of choice. Meanwhile, the demand for alUSD on the buy side is relatively low, with very little DeFi application support for alUSD that would spawn some demand to hold it. Thus, price continues to bleed down slowly. In order to overcome this, liquidity rewards at the protocol layer are directed towards 3pool LP vaults that provide liquidity between alUSD and DAI, USDT, USDC.

UXD : UXD probably offers the strongest guarantee of stability out of any non-overcollateralized stablecoin, with a unique structure that is even competitive with overcollateralized models.  However like any other decentralized protocol, UXD is subjected to peg instability during periods of severe volatility, where the derivative protocol may not be able to properly offload liquidated positions and the protocol forced to incur socialized losses. A good portion of this risk however is quasi-insured by tranches of various system stakeholders (both in the Mango markets derivative protocol, and the UXD protocol) before it gets passed on to the UXD token holders themselves. First line of defense is offered by the Mango Markets insurance fund. If that were to be depleted, a second line exists in the UXD protocol insurance fund, which would fill any excess gap. Lastly, if both those resources were to be exhausted with outstanding debt still remaining, a 3rd line exists in the UXP governance token holders who backstop protocol risk (the protocol will mint UXP out of thin air and sell it in the open market for additional collateral to fill the void). If all UXP token liquidity evaporates completely, then UXD itself would be undercollateralized and repriced accordingly in the market. We think the probability of all 3 defense lines failing are incredibly slim, and would likely only occur from an industry ending event (e.g. an undisclosed quantum computer is developed by a secret agency that steals everyone’s assets across every crypto protocol)

FRAX : Has ‘strongish’ stability in the exogenous USDC backstopping the FRAX token. There are direct incentives that help revitalize the peg if price trades below $1, but one of the inherent issues is that these incentives, by protocol design, take a while to kick in.

Additionally the ‘undercollateralization’ gap  is guaranteed by FXS holders in the event market forces signal demand for higher USDC collateralization. If no buyers are willing to come in and sell USDC for FXS, the system becomes inherently undercollateralized by whatever portion of FXS accounts for 1 FRAX.

FEI : Probably the weakest in terms of stability due to a) volatility of the assets backing FEI, uses native crypto collateral instead of USDC/USDT b) lack of overcollateralization buffer c) lack of direct arbitrage mechanics

UST : Unlike some of the other decentralized stables UST has a very sound and deliberate mechanism that incentivizes direct arbitrage capture when price breaks parity, and just as importantly this arb can be performed extremely quick. However, the system offloads UST volatility onto LUNA to support the peg. There are various reflexive downward loops that can emerge if LUNA liquidity dries up. However liquidity crisis events are inherent risks that affect all stablecoin protocols across the board, so it’s not a unique risk isolated solely to UST.

Censorship Resistance

People often intertwine the properties of censorship resistance and decentralization when speaking about blockchain systems. But decentralization for its own sake is meaningless; rather, we use decentralization as a tool to achieve a feature: that feature being censorship resistance. Centralized systems are non-objectively better performance engines. They are cheaper, faster and more scalable than their decentralized counterparts. But they create a central point of failure due to their autocratic nature, which opens the door for certain users to be profiled against by those that control said power.  Once a system has decentralized its design architecture enough so that censorship resistance is achieved, additional decentralization produces diminishing marginal returns.

Capital Efficiency

Protocols that require overcollateralization are inherently less capital efficient, which makes the prospect of minting new stablecoin less attractive for the average user. In fact, the only real function of minting these forms of stablecoin is credit facilitation, where someone is holding these existing spot assets and wants to take out debt against them without forfeiting the underlying.

Scalability

The outstanding supply of stablecoin needs to be able to expand in order to respond to an increase in demand. If growth and adoption of the stablecoin in question continues to rise due to integration with different DeFi services, but the total supply remains fixed, basic economics tells us that price will have to go up. However, that fundamentally violates the coveted properties of our stablecoin, which is achieving parity with the USD. Thus, we need a mechanism in which arbitrageurs and various other market participants can respond to increased market demand (reflected by price > $1) by depositing new collateral into the protocol and minting new float, offsetting that demand.

In 2021 we saw an emergence of a number of new decentralized stablecoin models and witnessed significant acceleration of adoption across the board. Each market makes certain tradeoffs with regard to the scalability trilemma and no solution has emerged as the “perfect” decentralized stablecoin. While few have demonstrated that they have achieved product / market fit, the market capitalization of decentralized stablecoins sits at just a fraction of the market caps of fiat backed stablecoins like USDC and USDT and, given that the explicit goal it targeting a much larger TAM of the $90T+ supply of fiat currencies, we believe the design space for decentralized stablecoin space will only expand with the winning model(s) for internet-native reaching outstanding supplies in the trillions.

If you want to support or contribute to our mission to reduce noise and create thoughtful DeFi research, please follow us on twitter @analystdao and join us in Discord @ https://discord.gg/eAkrN4XN.