Introduction
Currently, Alchemix accepts three collateral types for alUSD borrowing, DAI, USDC, and USDT. These assets are all paired against alUSD in its most liquid market, alUSD3CRV on Curve Finance. These are the big 3 stablecoins in the crypto space, and while each has their own tradeoffs, they have strong track records of being very tightly pegged to USD. 3CRV (DAI, USDC, USDT) is the single largest unit of liquidity on Curve, being indirectly incentivized by other various centralized and decentralized stablecoins.
However, over the past few months, the Frax protocol has started an initiative to supplant 3CRV as the dominant source of liquidity on Curve with their FRAX Base Pool (FBP), which consists of FRAX and USDC. Currently, Frax is aggressively collaborating with willing protocols across DeFi to pair their stablecoins and governance tokens with FBP in a way that is mutually beneficial for both parties. When protocols pair with 3CRV, they must provide all of the incentives for the pool themselves, which in turn gives 3CRV an enormous amount of liquidity incentives for free, as they are paid for by other protocols. So while the teams behind the 3CRV assets are proud of how they didn’t require liquidity mining, they surely benefitted from other protocols incentivizing their liquidity.
FBP is different in that Frax assists with collaborating protocols by incentivizing FBP pairs proportional to how much FBP is deposited in their pools.
Analysis of alUSD Liquidity Pools
Currently Alchemix is collaborating with FRAX with two FBP pairs on Curve, one for alUSD and one for ALCX, and one FBP alUSD pair on Saddle. It is worthwhile doing an economic analysis and comparison between the alUSD3CRV and alUSDFBP pools on Curve.
Currently, there is $106m locked into alUSD3CRV, with ~$57m alUSD and ~$48m in 3CRV assets. We incentivize this pool with 4000 ALCX via the Votium platform every other week, alongside the voting power of our generous sdCRV and vlCVX holdings, with the pool currently yielding 4% APR. Currently, for every $1 in direct incentives/2 week epoch, we can generate ~$1300 in liquidity for the pool.
The alUSDFBP pool has 10.7m deposited, with alUSD consisting 6.1m of the pool’s liquidity. We incentivize this pool with 500 ALCX every other week, but do not use our sdCRV and vlCVX voting on it. This pool is currently earning 8.5% APR, which is also bolstered by FXS emissions in addition to the CRV and CVX. For every $1 in direct ALCX incentives, we can generate ~$1100 in liquidity for the pool. This pool does not receive votes directly from us, but is aided by Votium incentives from Frax.
If we adjust the price of liquidity to factor in the pool APR, we see that via FBP, we secure almost twice the yield per $1 of incentives as compared to alUSD3CRV, which is a major difference.
Alchemix AMOs
The alUSD Elixir Algorithmic Market Operator (AMO) holds approximately 46m of the 106m of total liquidity, and in tandem with our voting and incentivization, this AMO usually returns 70-120% of value for what we put into it from our incentives (these numbers are highly variable based on market conditions). For example, the alUSD AMO generated 50k CRV and 2500 CVX in the previous epoch for a total of ~$60k in revenue with $80k in incentives. This is a huge boost in economic efficiency from a 100% loss on incentives if there were no AMO. However, let’s try to simulate what an alUSDFBP AMO would look like using the numbers outlined above:
If the entire $46m of the Elixir AMO moved to FBP, and emissions were adjusted to maintain the current levels of yield, an alUSDFBP AMO would have generated approximately $130k, as opposed to the $58k the alUSD3CRV AMO generated in the previous epoch. Instead of getting 75% of the value we put in back, we would get 162.5%. This is a game changing difference that would fundamentally improve Alchemix’s tokenomics for the better.
Frax and Alchemix
So, if Alchemix were to go forward with transitioning to an alUSDFBP AMO, it begs the question: If alUSD is so heavily paired against FRAX, wouldn’t it make sense to also add FRAX as a collateral asset for borrowing alUSD?
First, let's dive into what FRAX is and how the Frax protocol works. When FRAX first launched, the only way to generate FRAX was via a dual deposit of both their governance token, FXS, and USDC, with the ratio generally being 80% USDC and 20% FXS. In this setup, FRAX is a semi-algorithmic stablecoin with a portion of its backing being in the form of its native governance token. Frax had mild success with this formula, generating over 100m FRAX via this method, however their protocol found a safe avenue for hyper-scaling via the introduction of the AMO (from which the Alchemix AMO was modeled).
While Frax has several AMOs deployed, their main one historically has been the FRAX3CRV AMO. This mechanism works by setting a minimum accepted price for the FRAX stablecoin. Currently, they target a peg of 1 FRAX per USDC. If FRAX is at or above this peg, then the protocol can directly mint FRAX into the liquidity pool, and conversely, if it is under the peg, Frax can withdraw and burn FRAX from the AMO to restore its peg. Then this Protocol Controlled Value (PCV) is put to work earning yield in Convex, with the farmed assets going to bolster the FXS price and increasing their voting influence on Curve gauges. This strategy then created a flywheel effect, where their liquidity incentives became ever more profitable for the protocol while also increasing their market share, making FRAX the second largest decentralized stablecoin.
While some people may understandably have some hesitation around a semi-algorithmic stablecoin, especially after what had happened to Luna and other algo-stables, since the Frax protocol owns the vast majority of the FRAX in circulation, they are able to gracefully expand and contract with the market. While every other decentralized stablecoin saw a somewhat elongated period of depegging following the psychological damage done to LPs after the Luna meltdown, FRAX’s peg was stable because their AMOs properly contracted by removing and burning the right amount of FRAX.
This is even more impressive when you consider that they were affected by the Rari Fuse exploit and the death of the 4Pool with UST.
FRAX vs DAI
Once people understand the mechanism design of Frax, the next criticism that comes to mind is “Doesn’t this just make FRAX wrapped USDC?” In some senses this is true, however, it is becoming less true over time. Frax has recently launched FraxLend, which introduces a CDP system for borrowing FRAX using volatile collaterals. As this system sees more adoption, the percentage of FRAX backed by USDC will decrease over time.
While DAI has widespread recognition for being the first successful decentralized stablecoin, it is increasingly becoming more and more FRAX-like. During the black swan market event in May 2020, chain congestion allowed for the MakerDAO keeper system to be attacked in a way that left the protocol below its desired collateralization ratio. MakerDAO had to mint new MKR tokens to auction off to fill this gap, and they also responded by adding USDC as collateral for borrowing DAI. This evolved into MakerDAO’s DAI Peg Stability Module (PSM), which paired DAI with USDC collateral to bring DAI’s peg back down to $1. MakerDAO even takes USDC/DAI Uniswap LP tokens as collateral, using DAI as backing for…DAI. According to https://makerburn.com/, currently 81% of DAI is backed by fiatcoins (mostly USDC), and now MakerDAO is proposing using their PSMs to earn yield on Coinbase and other centralized custodial parties.
Clearly, the two protocols are beginning to converge on a single design, albeit with some minor differences. MakerDAO governance is completely on-chain whereas Frax is conducted via snapshot and a multisig. Thus MakerDAO is arguably more decentralized, but by taking user funds in the PSM to deposit in providers such as Coinbase and Gemini, they would require centralized managers and introduce custodial risk to facilitate such operations. Frax has publicly outlined their transition to full on-chain governance, and is expected to finish this transition in the next few months. Information about their transition plan can be found in this tweet thread :https://twitter.com/blockworksres/status/1574789308082372608?s=20&t=IgGdDmRdXRAYMkMQ3dcD6g. Furthermore, Frax is decidedly DeFi focused and keeps their assets on-chain, with a robust UI showing exactly where all funds are currently deployed in real-time. So while Frax is progressively decentralizing, MakerDAO is progressively centralizing by partnering with CeFi custodians to chase yield.
There is also some uncertainty regarding how MakerDAO will treat DAI. Maker’s founder has pushed for the “MakerDAO Endgame” that would see the PSMs converted to ETH and allow DAI to go off peg, which does pose a major risk to Alchemix since our system assumes DAI is $1 in value. If that were to significantly change, DAI would no longer be an acceptable collateral for alUSD. If we had to remove DAI as collateral, it would be extremely preferable to have an alternative decentralized stablecoin be a part of our system.
So if both stablecoins have centralization vectors, are majority-backed by USDC, have had their pegs battle tested, and are extremely liquid, why would Alchemix only accept one as collateral but not the other?
Request for Comment
This document lays out the economic case for a transition to an alUSDFBP AMO and the addition of FRAX as alUSD collateral.
This is not a proposal yet, but rather a request for the Alchemix community to share their thoughts and to gauge opinion to see if Alchemix core should pursue this direction. If the community is supportive of this initiative, we shall draft a formal proposal with more specifics.
So, please share your thoughts with us, as your feedback is very important.