Background

FraxBasePool(FraxBP) is a base pool token on Curve that is an alternative to 3CRV. FraxBP consists of only FRAX and USDC, while 3CRV consists of DAI, USDC and USDT. FraxBP is deployed on mainnet in June 2022, as the second base pool token in addition to 3CRV on Curve Finance.

Motivation:

1. Bootstrap liquidity by enabling partnership with smaller protocols

Before having FraxBP, the only available Curve stableswap pair with FRAX was FRAX / 3CRV. With the new FraxBP as the base pool asset, there are more variations of DIRECT stable and risky-asset pairs that are made possible. Further, FRAX team was committed to recycling their share of Curve bribe back to whatever new FraxBP metapools. In the original forum post on Curve, Sam Kazemian, the founder of FRAX protocol suggested that the Frax team would allocate their Curve bribe generated from the demand for FraxBP back to the corresponding metapool:

FRAX will make a commitment to return revenue generated from FRAXBP demand back to metapools paired with FRAXBP thus allowing us to contribute more cash flows back to the ecosystem. Essentially, FRAX will proportionally distribute revenue to any project that pairs with FBP equal to the demand for FBP in their metapool.

This is made possible since the CurveAMO from Frax would mint FRAX into the base pool to create the additional supply needed for a particular new metapool. The additional supply would never be redeemed to USDC and only sits on the “balance sheet”; and would be burnt when later on the curve pool receives the excess FraxBP back. Meanwhile, the AMO would receives a considerable part of the Curve bribe (CRV + CVX) which the FRAX team is committed to re-invest back to that particular pool.

2. Circumvent Exposure to DAI and USDT.

In the underlying mechanism, FRAX is minted by providing a fractional amount of USDC, coupled with FXS on a dynamic ratio, to mint the equivalent amount of FRAX. The only stablecoin exposure is USDC under the hook. When Curve was launched in 2020, the only available base pool token was 3CRV. This creates an unavoidable pairing to 3CRV when FRAX needs external liquidity to facilitate inter-protocol exchange of value at that time. However the choice of 3CRV / FRAX also leads to additional exposure on USDT and DAI, in addition to USDC. Over the year, DAI has increased exposure to various underlying assets in its Peg Stability Module(PSM). You can refer to the initial governance vote (MIP29) on makerDAO forum to understand the detail and motivation of PSM. What it means is that DAI has taken small, but many assets’ risk.

At the same time, USDT has undergone multiple times of major de-peg. There are also concerns that Tether have not fully disclosed their treasury exposure, nor verifying that they have sufficient liquid assets to back up the circulating USDT.

Therefore creating FraxBP that only pairs with USDC, could provide a more direct pathway for liquidity providers that are concerned about the risk of DAI and USDT.

Rapid Development since Launch

As of now (Jan 2023), FraxBP has been paired with 16 stablecoins as well as 11 risky assets, totalling 27 pairs on Curve since its deployment on June 2022. It has diversified the liquidity that once clustered in FRAX/3CRV to both FRAX/3CRV and FRAX/USDC respectively, and hence using the derived FraxBP to further provide liquidity leverage to the many more curve pools.

Comparison with 3CRV as a BasePool

Currently, FraxBP has 20% less liquidity than 3CRV, but 50% higher bribe APY, but virtually no swapping reward from the base fee.

What this implies, while the FRAX team is proactively bribing for the base pool (with the aim to make it as attractive as joining 3CRV), the current total APR is not significantly more attractive than 3CRV.

Comparison with FRAX/3CRV

FRAX/3CRV has 30% higher APR since it takes additional exposure to DAI & USDT. The size of liquidity is comparable.

Balance-wise, FraxBP has a slightly higher ratio of FRAX (65%) versus Frax/3CV (61%), while FraxBP and Frax/3CRV both have a relatively high amplification factor (1500) to tolerate a small swapping slippage in an imbalanced pool.

Risk Taking Implication

• DAI & USDC & USDT => 3CRV

• USDC & *FXS => FRAX, FraxBP

• DAI & USDC & USDT & FXS => Frax/3CRV

• USDC & FXS & X(custom) => FraxBP/X *FRAX / FraxBP is not directly exposed to the price of FXS; but if FXS gets free-minted to infinity, then essentially the peg of FRAX and FraxBP could NOT be maintained. (due to the design mechanism of FRAX)

Liquidity Leverage and Money Supply

Imagine u have 1 USDC: You let the USDC sit in your wallet. The blockchain world has 1 dollar. What a stablecoin protocol likes Frax does: It creates a liquidity leverage:

=> You put the USDC into Frax (neglect the technicality during mint for now), you get 1 FRAX. And Frax protocol has your USDC. Frax would deploy your USDC to somewhere for yield generation, while you can deploy your FRAX to a FRAX/X metapool. The blockchain world has 2 dollars now.

With the addition of FraxBP (FraxUSDC pool) => You put the USDC into Frax to get 1 FRAX. And Frax protocol has your USDC. Frax would deploy your USDC to somewhere for yield generation.

=> You put your FRAX into FraxUSDC to get 1 FraxBP, and then you put your FraxBP into a FraxBP/X metapool for yield generation. Frax Treasury has your USDC. FraxUSDC has your FRAX; and the FraxBP/X pool has your FraxBP.

The blockchain has 3 dollars now.

This is why FraxBP has the strategic value to leverage 50% more money supply from the previous scenario.

Curve Bribe Retention

As a protocol, let’s say you would like to create a Curve metapool to supply liquidity for your native token (be it stablecoin or risky governance token) by pairing it with 3CRV. To incentivise liquidity providers to join your pool, you can choose to provide “bribe” for veCRV voters to vote for your pool such that the liquidity providers of your pool can receive more CRV emission, effectively having a higher APR. However there are two caveats:

1. CRV would inflate. It means that the impact of bribing for 1 veCRV vote would decrease gradually over time. You either have to increase the number of veCRV to bribe, or tolerate a reduced CRV emission.

2. Usually liquidity providers are not loyal. As soon as the APR in your pool drops out of the relative yield of other pools, they would hop to other liquidity pools, hurting the price stability of your native tokens.

In light of this, what Sam said in the original FraxBP proposal specifies how FraxBP would help to address these two problems:

1. When you create a FraxBP metapool with your native token, Frax protocol would help to provide the necessary FraxBP to pair with it (on the stablecoin scenario, probably be case-by-case for a risky asset), by minting FRAX from the CurveAMO into FraxUSDC and sending the FraxBP to the designated metapool.

2. With 1.), Frax Protocol is now one of the liquidity providers for your metapool, and receiving CRV and CVX bribe. Frax protocol would then utilize these CRV and CVX farmed to continue “bribing” or “voting” for your pool next time.

What this implies:

• Less resources to maintain a stable peg, since Frax Protocol would provide some liquidity for pairing.

• Your Curve bribe creates a recurring impact, effectively mitigating the effect of CRV inflation.

• higher price stability, as the Frax AMO would not withdraw liquidity as a result of reduced APR but would only do it for a price rebalance.

Summary

FraxBP is a strategic play by Frax protocol to retain liquidity and establish more partnership. It smartly plays the Curve game and provides a win-win situation for other protocols to integrate. With 6 months in mainnet, It has rapidly led to the creation of 27 pools.

Appendix

Initial Post of FraxBP on Curve Forum: https://gov.curve.fi/t/deploy-a-fraxbp-pool-whitelist-frax-for-vecrv-staking/3958 Incentive Strategy to bootstrap liquidity: https://snapshot.org/#/frax.eth/proposal/0xd4cda7880b314cadc8cb3960aeefa25770c736dcb23af591d329a2ddbcfed345 FRAX protocol DAO actions Overview: https://messari.io/dao/frax-finance-governance/proposals

• FPI is an instrument that is inflation-pegged, 100% collateralized by FRAX, with yield backed by AMO strategy earning and FPIS auction in case of deficit.

What does it offer?

• FPI offers US inflation pegged yield, which realised 7% APY in Q4 2021.

• It is implemented by a peg price that is updated MONTHLY when inflation rate is announced by US government: https://www.bls.gov/cpi/

• FPI peg price appreciation is smoothed out and increased per second in contract implementation. (check currPegPrice in CPITrackerOracle: https://etherscan.io/address/0x66B7DFF2Ac66dc4d6FBB3Db1CB627BBb01fF3146#code)

• The TrackerOracle is owned by Frax Multisig:

• Oracle job is maintained by Chainlink Operator with jobId: 0x3163333039643432633730383462333462316163663161383965376235316663

How to mint and redeem FPI?

• Anyone can mint and redeem FPI with FRAX based on the latest peg price with fixed 30bps fee, FRAX is hard-pegged to 1 USD (some checks to make sure FPI is closely pegged in market AMM during mint & redeem).

• FPI has a cap of 150m total circulation. This parameter is configurable by FPI Multisig.

Treasury Backing

• Currently there is no AMO borrowed asset through the smart contract AMO functions, FPI is backed by:

  • 12.5m FRAX in the FPI ControllerPool, for immediate FPI redemption.

  • 67m worth of LP position in the FPI Multisig. 66.5m in Curve & Convex and 0.5m in Uniswap.

  • 2.5m Frax in the FPI multisig.

• totalling 82m which matches the 77m circulating FPI valuation with latest peg price at 1.065 (as of Jan 2023). Meaning FPI is 100% backed by FRAX, even though the treasury has some exposure to FPI itself that contributes to an inflated total supply. (meaning that 82m Treasury actually “OWNS” some FPI itself)

• One thing that is concerning to the decentralization is that there is a backdoor recoverERC20 function in the FPI ControllerPool. It allows the Multisig to pull any token out (INCLUDING FRAX). The actual AMO operation is mainly conducted through this method, instead of the pre-defined AMO operation in the smart contract.

  • Example Transactions:

    • https://etherscan.io/tx/0x1d07afa0b2c4d9721cc5d760057f461122f5022d150fabf80e2c8f72b4034c7e

    • https://etherscan.io/tx/0x326ccb9834efb953355270d9f3b193186e5827fd6c5002e5223b35a9b334f6b3

How does FPI Treasury upkeep the yield?

• The realised yield in Q4 2021 was around 7% APY, which is considerably higher than the risk-free rate of other stablecoins. The FPI Treasury achieves this by:

  • farming CRV and CVX on the Convex LPs (yes which comes from FXS emission/bribe)

  • collecting 0.3% upon minting/burning FPI

  • By AMO operation if someone purchases/sells FPI with premium on the market. (no AMO deployed atm)

• If the above does not generate enough yield, FPIS would be sold for FRAX to cover the gap, FPIS is the governance token of FPI. There are 1.3m FRAX in the FPIS/FRAX AMM atm.
 https://docs.frax.finance/frax-price-index/overview-cpi-peg-and-mechanics

Lack of Clarity in ControllerPool Migration

Accounting during migration to current ControllerPool:

• FRAX Multisig transferred (transfer, not buying FPI or anything) 7.5m FRAX to the current FPI ControllerPool in the beginning; However, on that time there were 18m FPI in circulation. https://etherscan.io/tokencheck-tool

• There were still 2.7m FPI in fpis.eth which was the initial holder upon initial mint (https://etherscan.io/address/0xf2c4592813b5b3f79ac522e4efb2c19a666e937c), at the time of this transfer. Even discounting 2.7m there were 18 - 2.7 = 15.3m FPI outstanding. This leads to a gap in the reconciliation during the migration. Would need more background from the FRAX team on when they later sent back enough FRAX to maintain the 100% collaterialization.

Is FPI riskier than FRAX?

FPI risk in additional to FRAX:

• Robustness of yield source In 2021, the APY of FRAX/FPI was quite high as the curve bribe is in unit of FXS and FXS had huge momentum and liquidity. No AMO or FPIS auction was needed to happen. But now the yield of FRAX/FPI is on sub-2% APY level as the price of FXS enters into bearish territory together with the wider market, it causes concern over the need to leverage AMO and even selling FPIS to cover the deficit in order to maintain the higher inflation-pegged yield.

• Decentralisation: As mentioned above, even though the circulating FPI is 100% backed by treasury, the fund is split into a multisig and a smart contract respectively. Only the fund in smart contract is immediately and permissionlessly redeemable. In order to enable a whale to withdraw, manual intervention from the FPI multisig is needed to send back the fund. Further, there is an recoverERC20 function that enables the FPI multisig to pull FRAX out.

Summary:

FPI is an instrument that is designed to be pegged to the US inflation rate. It has the below pros and cons:

Pros:

• Offers an attractive yield source that is pegged to the US inflation rate

• Can be easily minted by and redeemed for FRAX by paying fixed 30bps fee

• Yield is back-stopped by FPIS liquidity.

Cons:

• Redemption NOT fully decentralised;

  • only some funds stays in the ControllerPool smart contract,

  • a major part remains in the Multisig and invested in the Curve & Convex Pool. This part is subject to centralisation risk.

• There is a lack of clarity surrounding the accounting during the migration to the current ControllerPool.

• Concern over the robustness of yield sources

Appendix:

• FPI stats: https://stats.frax.academy/

• FRAX Risk Assessment:
 https://cryptorisks.substack.com/p/risk-assessment-frax-governance

• FPI Multisig:
 https://debank.com/profile/0x6a7efa964cf6d9ab3bc3c47ebddb853a8853c502

• There were legacy FPIControllerPools which could mint FPI before the current FPI ControllerPool went into production. so the current contract is started with non-zero FPI circulation.
 https://etherscan.io/address/0x94cff60496c71a0302ababa0da1a1f21626f9613#code
 https://etherscan.io/address/0x309ac8840f9b4c7eeb5bab1e89669d8dbb86c060#code

• There is a TWAP AMM (FRAX/FPI) price in UniswapV3 is monitored in the ControllerPool SC against the peg_price. This is used as a check against the peg_price from CPITracker.

Uniswap-V3 concentrated liquidity & collateralised lending.

Abstract:

crvUSD provides an innovative way of stablecoin service that drastically reduces the loss during liquidation through the use of Lending-Liquidating AMM Algorithm (LLAMMA). LLAMMA spreads liquidity into bands and introduces “devaluation” of partial collateral when collateral price drops, and recovers the collateral when price bounces back; In a historical back-test, if the price of the collateral drops and then rebounds, the liquidation loss under LLAMMA was as little as 1% upon liquidation, if the collateral loan is spread across a range of 20%. It also provides a permissionless arbitrage point of service where arbitrageur can compete without need of capital (through PegKeeper). With PegKeeper, it closes the arbitrage loop by bridging curvePool AMM with the new lendingAMM. Once crvUSD is released into production, it is expected veCRV would increase in utility as veCRV holder can decide the addition of collateral type and upper limit of crvUSD that collateral market can accommodate.

Context:

crvUSD is a Curve’s upcoming stablecoin. The code is published on GitHub and is still under active development.

Problems to Solve:

From the whitepaper, crvUSD is developed as a stablecoin with a more advanced and gentle liquidation mechanics. In classic collateral-enabled stablecoin protocols, a position would be liquidated when the value of that collateral position drops below the loan value, plus certain buffer. This has the benefit of simplicity but create massive losses for the loan taker if liquidation happens. For example, the liquidation penalty on Liquity is 9.09%, quite significant even though Liquity is already aggressive in lowing this buffer relative to other stablecoin protocols. This buffer is needed mainly to offset the volatility in the price of the underlying risky-asset collateral, as well as the delay in oracle price update(s) in the blockchain space.

How crvUSD solves this?

crvUSD reduces the liquidation penalty significantly, by providing an innovative liquidation mechanism, called LLAMMA. (yes a very lengthy name). Outlined in a historical back-test in the whitepaper, under LLAMMA, if the price hit liquidation threshold and then rebounded later, the liquidation loss would be as little as 1% upon liquidation, if the collateral loan is spread across a range of 20%.

LLAMMA works by creating “bands”, where collateral is deposited into a range of them. Similar to ticks in Uniswap-V3, each band have a non-linear price relationship to each other. Health factor of a position is then computed by calculating the effective collateral value based on user’s share on each of the band. As the price of collateral drops, the deposited collateral in the top of the band ranges would be swapped into crvUSD gradually (devaluation), while the health factor could remain positive (healthy). Liquidation would only happen if the health factor drops to negative with regards to the liquidation discount, then the entire position would be repaid with penalty; Since part of the collateral would be already devalued (without penalty) when liquidation happens, loss over the entire position would be reduced considerably relative to the traditional one-size liquidation.

ps: Self-liquidation without the penalty is possible: If the position drops into the liquidation threshold and no liquidator goes ahead to liquidate the position. The loan taker can call self-liquidation to close out the position without taking any penalty (alternatively, he can take the penalty to himself, essentially rewarding himself with his own money zeroing out any gain).

Architecture:

There are multiple components in the design based on the whitepaper and published code:

ControllerFactory:

• Factory is eligible to deploy AMM markets and control versioning of all the contracts.

• Factory can update the upper limit of crvUSD each collateral market can mint.

• Factory would initialize the liquidation threshold, borrowing_discount (aka LTV), address of controller, MonetaryPolicy and admin fee for a particular market upon addMarket(). adminFee and LTV can be changed later in controller.

• Factory is governed by an admin which likely to be holder of veCRV.

Controller:

• Controller is the main entry point for users to interact with the AMM market. All user interaction for example create_loan, repay, liquidate etc, are called through the controller.

• The controller is the admin of AMM.

• Each controller would map to only 1 AMM.

• Controller log the value of loan and collateral of each depositor.

• Controller can update adminFee and LTV.

AMM

• AMM is the core logic at which any swapping is calculated.

• Except Exchange(); all other functions can only be called by Controller.

• exchange() is the entry point to swap collateral to crvUSD or vice versa given the deposited price band is above the active band.

• Once deployed, AMM for a market is immutable. However the deployed version for the next market in factory can be upgraded.

MonetaryPolicy (MP)

• MP is a separate module defining the rate at which the loan would grow.

• From the repo, there is a constantRate MP as well as a dynamicRate MP

• MP is initialised upon creation of a market and can be updated by controller.

PegKeeper

• PegKeeper is an actor that interacts with the AMM in crvUSD and the collateral curvePool (another AMM pool) when any misprice happens.

• PegKeeper would offset the misprice by add_liquidity/remove_liquidity into curvePool.

• Supposedly, pegKeeper would be allowed to mint uncollateralized stablecoin and deposit it to the stableswap pool single-sided in such a way that the final price after this is still no less than the pegged price minus some threshold.

• The PegKeeper is allowed to withdraw (asymmetrically) and burn the stablecoin.

• With above anyone can called the update() function to check for mis_price, without he/she having to provide the capital for arbitrage; if mis_price exists above the threshold then the caller would share a fraction of the arbitrage profit. (caller_share)

• PegKeeper is added into MonetaryPolicy.

Example In Action

https://crvusd.0xreviews.xyz/ Oracle Price = 1000 Top 5 bands: 990/980.1/970.3/960.6/950.99

In this simulation, AMM price is assumed to be 1020.3, where there is no arbitrage and all collateral is still healthy in the top bands. active band is only 1, where only band 0 can be “devalued” (which is emptied by default). If the price of AMM increases, top band could be devalued.

As you can see, if AMM price increases to 1040.80, then the collateral in tick1 would be swapped into crvUSD mostly. If AMM price increases slightly to 1041.02, then active band can be updated to 2, and band 2 would be eligible for swapping now.

As you can see on tick1, collateral (y) is already swapped fully into crvUSD (x), whereas tick2/3/4/5 still have some collateral. The execution price would be calculated based on the invariant I = (f + x)(y + g), whereas x and y are balance of crvUSD and collateral in that band respectively. The process continues when AMM price increases.

All top 5 ticks would be “devalued” when AMM price turns in 1128.18.

This is a position-agnostic, price-only simulation. If at any point the position drops below the health factor then the entire position would be liquidated (taken) by the liquidator, essentially withdrawing all user share from bands. This is where liquidation discount (penalty) would be introduced. This can be triggered by loan sizes due to interest accural, not necessarily by price action.

Impact:

crvUSD is a new and innovative way to create stablecoin with a much more gentle liquidation mechanism, by blending uniswap-V3-like concentrated liquidity with collateralized lending. It leverages on the success of Curve-V2 AMM to create a seamless arbitrage model that makes the lending market much more efficient. Once crvUSD gets released and proves itself to what it promises:

1. veCRV increases in governance power Besides boosting yield and voting for gauges, veCRV can increase in utility by governing the type of collateral that can be added to mint crvUSD, as well as the upper limit.

2. Bridging liquidity from lending market to AMM With crvUSD, liquidity in lending market can then be provided to AMM as well, increasing the liquidity provision in the bigger market. Vice versa, if price rebounds, then the price action would be bridged back to the lending market through pegkeeper, cushioning any subsequent liquidation. This would help reduce market volatility that is incentivised to trigger liquidation.

3. Create a new wave of lending market research and protocols engineering LLAMMA is an elegant way of combining concentrated liquidity and stablecoin lending. This new codebase would be explored further once it is proven in the production environment. New protocols and aggregators that build on top of lending market as well as Curve would erupt and push the Defi boundary further.

Appendix:

• https://crvusd.0xreviews.xyz/

• https://github.com/curvefi/curve-stablecoin

Tl:dr Depositor Flow:

• Permissioned factory model controls the type of collateral and upper limit crvUSD of each collateral can mint

• Once a lending market is created, User create loans by depositing collateral into bands based on the latest oracle price.

• During creation of a loan, the depositor send in collateral as well as specify the range of band that he/she would like for the deposit to spread across. This value can be ranged from 5 to 50.

• There is an interest rate, increasing the size of loan per second. This can be static, or a dynamic variable as a function of total loaned amount.

• User owns share of liquidity in bands upon depositing; If price of collateral stays above the price where deposit happens; the collateral would remain there and nothing would happen.

• If price of collateral drops, active band hitting the band where the depositor put int money, the collateral in the top band would be converted to crvUSD by arbitrageur. This is called devaluation. Devaluation does not incur penalty to depositor.

• Once a position suffers devaluation, it means the collateral within the devalued band would be swapped to crvUSD. If price recovers later, the crvUSD would be swapped back to collateral.

• The the price of collateral continues to drop, the health factor of the position would hit negative to a point, where liquidation could happen and penalty is incurred.

• Once health factor drops to negative, liquidation of an entire position would happen as an atomic operation. Liquidator would pay the entire debt and takes out the user_shares of depositor across all bands (both crvUSD and collateral).

• Each band has an upper and lower price limit that has the relationship of (A-1) / A. During devaluation (collateral being swapped to crvUSD), the price within the band would moves within this narrow range as ratio of collateral decreases.

• Therefore, liquidity in each band would smooth out prices between them, making execution price continuous even in a band.

What is Yield Farming ?

Providing liquidity to AMM in return for trading fee, and governance tokens has been a standard way of generating yield for stablecoins since defi summer in 2020. This operation is nicknamed as “farming” as liquidity providers often would “harvest” the interest in order to compound it back to the principle to maximize gain.

Thanks for reading Gro Farmer! Subscribe for free to receive new posts and support my work.

Why we no longer see 1000% as much as earlier?

In the early days of 2020, stablecoin farming was very lucrative due to high valuation of the governance tokens of new protocols, as the market was still relatively small and capital inflow was sizeable. Fast forward to now, with only a few exceptions, many AMM protocols (a.k.a farms) cannot maintain their high APYs due to the fact that their governance tokens can no longer maintain the hype and thus high valuation.

Added into the reasonss of reduced APY is the clustering of capital over a few battle-tested protocols, as people getting to realize the many types of risk over farming on new and unproved protocols.

This article would explore some of the common risk of stablecoin farming with examples.

Price Risk: when stablecoin does not equal to $1.

ALL stablecoins are engineered with the purpose to peg to $1. However in reality, the price of stablecoins would vary within a tight range around $1 on a weekly, if not daily basis. This is due to arbitrage and trading activity as supply and demand fluctuates. In most cases, the price would repeg back to $1 very quickly. In rare occasions, or during volatility or some news-driven movements though, stablecoins would depeg from a few basis points to a few percentages.

When would price fluctuation be risky?

IN the SHORT run, price fluctuation is NOT a risk, but benefit. If the stablecoin works as expected, price fluctuation does not harm any stablecoin farming positions. In fact, price fluctuation would benefit the farmers by bringing additional trading fee as long as the stablecoin repegs to $1 later on. For example, USDT, DAI, USDC all faced major depeg at certain points, and they all repegged back later on. This created arbitrage opportunity, as well as sizeable trading fee for liquidity providers.

In the LONG run, though,  extended and major price fluctuation would cause confidence issue, and thus reducing the strength and price floor at which arbitrageur would step in to recover the market price of the stablecoin. This would further reduce the price resilience. For example, 90% of time MIM’s price is discounted and investor would expect certain discount in its price now, same for arbitrageur.

Liquidity Risk: When you cant sell whenever you want without hair cut.

Assuming you have $1000 DAI, and you would like to sell it for $1000 into your bank account for a lovely Chrsitmas. How? In the case of DAI, which follows the deposit → mint procedure, you cannot redeem DAI for USDT/USDC directly unless you have put down collateral in the first place; thus you are left with the only option: trade in existing AMMs.

In most time, you should be able to trade $1000 DAI to $1000 USDC. But like what is mentioned in the price risk session, if it happens to be the volatile time where people are selling $DAI as well, you may not find enough liquidity in the market to back $DAI to $1. It may be only $0.99., $0.98 or whatever. In that case, you would be forced to take a hair cut because you have a time constraint to sell $DAI to buy your wife’s Christmas present.

For whales (investors with sizeable positions), liquidity risk is more daunting relative to that to retail investors, since their positions are big enough to feel the slippage first when everyone is pulling money.

Oracle Risk: when oracle feeds a delayed or wrong price.

Oracle refers to the smart contract that feeds price data from the real world to the blockchain. While for centralised stablecoins like USDC or USDT, oracle risk is not applicable since they hold real-world USD asset as collaterals, for stablecoins that hold risky assets (anything that is not USD) as collateral, like DAI, LUSD etc. They would have to reflect the price of the underlying risky assets all the time for book keeping.

This price reflection can be maintained by the protocol itself, or sometimes the protocol would rely on a third-party oracle networks for this operation. Famous oracle protocols are Chainlink, API3, UMA etc.

If the oracle price is wrong, or delayed, the protocol would have a risk of turning a over-collateralized positions into an under-collateralized one, and suffering bad-debt, leading to Collateral Risk, which would be mentioned below.

For example, USDH on Solana suffered an oracle attack which leads to a loss of 1.2m.

Collateral Risk: When liquidation does not work (or not timely enough)

Collateral refers to the liquid assets that back the amount of minted stablecoin. Collateral risk refers to the general insufficiency of those assets to cover the totalSupply of the stablecoin. This can be caused by failed business endeavours, or some of the above risks like oracle failures/hacks, freezing of funds due to sanction etc. For protocols that accept risky assets (non USD) as collaterals, simply delayed liquidation leading to bad debt would also incur collateral risk, if their own treasury cannot cover the bad debt. A famous example was MakerDAO DAI’s 2020 black swan collapse, where ETH plunged 50% in within 24 hours. Since DAI was using ETH as collateral, a lot of positions went sour and had to be covered by auctioning $MKR. Subsequently a lot of positions were exploited due to zero-bid in the market for $MKR.

Regulation Risk: When government chips in.

Centralized stablecoin issuers are registered company that would have to be compliant with the jurisdiction at which they are registered.

For example, in the incident of the Tornado Cash sanction issued by OFAC (Tornado Cash is a protocol that facilitates anonymous transaction), while Tether(USDT issuer) ignored the sanction, Circle (USDC Issuer) complied quickly to that, since USDC is registered in the United States, while USDT is registered in British Virgin Island, and held by a Hong Kong based mother company.

Since many decentralized stablecoin protocols have already accepted USDC as a collaterals for minting their own stablecoin, holding decentralised stablecoin would also be impacted by regulation risk now. So if one day your favourite stablecoin protocol get sanctioned by US government and Circle decides to comply with that, then the USDC in the treasury might turn sour, leading to bad debts.

Smart Contract Risk:

Last but not least, smart contract inherently have risk of vulnerability due to logic missing, bridge hacks or malicious insiders. While some decentralized protocols dont permit unbacked stablecoins, some centralised stablecoin that takes real-world assets or bridge assets, do not hold this guarantee within one blockchain. If some stablecoins are minted out of a bug without proper assets backing, the newly minted stablecoin would also flood the market, as well as the the price of the stablecoin.

Example: Recent Hacks of aUSD on Acalas and Beanstalk on Ethereum.

Summary:

There are many risk associated with yield farming, this article tries to cover some of the common risk category that a stablecoin farmer would be exposed to. Make sure what you get pays off for the risk you take.

Disclaimer: Not financial advice. Do your own research.

$GHO is AAVE’s upcoming stablecoin. AAVE’s proposal to create it’s own stablecoin was passed in August and there is an audit already done by Openzepplin. Progress update is being done in the AAVE forum. The code continues being tested and audited and $GHO is still under active development.

What is $GHO

$GHO is a decentralised stablecoin that is governed by AAVE. Initially, AAVE lending pool would be the only “facilitator”(explained below) that can mint and burn $GHO by depositing USDC. AAVE governance aims to gradually bring in other facilitators who can have their own logics in accepting different collaterals for minting $GHO. AAVE governance would maintain the whitelist of facilitators and quota of $GHO minting of each of them.

How $GHO would work:

• Facilitator There would be a number of facilitators that can mint and burn $GHO permissionlessly. Each facilitator would have a bucket, which defines the upper limit of $GHO this facilitator can mint. facilitator is meant to be a smart contract that defines how it would allow $GHO minting.

• AAVE Governance AAVE Governance has the power to whitelist facilitators, make change(s) to its bucket, but not affecting any $GHO that is already minted. Governance would also be able to affect the interest rate of $GHO being minted out of each facilitator.

• AAVE lending protocol Based on the whitepaper, AAVE lending protocol is going to be the first $GHO facilitator in the beginning, meaning that $GHO can only be sourced through AAVE lending pool in the beginning.

• $GHO interest rate model AAVE governance will statically adjust interest rates depending on the need for the $GHO supply to contract/expand. AAVE governance has the ability to adjust this rate for EACH FACILITATOR.

• Discount model to safety module participants Holders of stkAAVE would enjoy 20% of interest rate discount based on the disclosure in the audit report. Amount of $GHO eligible for this 20% discount per stkAAVE is yet to be disclosed.

Comparison with existing stablecoins

• Comparison with $DAI Similarity: Both require collateral whitelisting (facilitator in $GHO); $DAI charges a stability fee and $GHO has a static interest. Difference: $GHO would put a bucket on each type of collateral while $DAI has no such constraint.

• Comparison with $FRAX Similarity: Nothing I can think of. Difference: $GHO is meant to be overcollateralized while $FRAX is under-collateralized. $FRAX has only stablecoins or stablecoin liquidity position in the treasury while $GHO can have risky assets. $GHO does not associate with $AAVE directly, while $FXS, the governance token of Frax protocol, is also involved during the minting & burning of $FRAX.

• Comparison with $MIM Similarity: Both $MIM and $GHO are designed to be over-collateralized. Both have a different mint quota for each type of collateral. Both have different interest rates parameters for each type of collateral. The mint quota is controlled by multisig for $MIM and controlled by governance for $GHO respectively. Difference: $MIM has no interest rate discount for SPELL holders or SPELL locking holders; while $GHO would provide interest rate discount to stkAAVE holders.

• Comparison with $LUSD Similarity: Both $LUSD and $GHO are designed to be over-collateralized. Difference: $LUSD is entirely permissionless and accept only $ETH as the collateral, while $GHO would accept multiple collaterals and it is permissioned to add further collaterals. $GHO would have a static interest, while $LUSD is interest-free. $LUSD incurs a 1-time 0.5% mint fee while $GHO does not specify any mint/borrow fee.

What would be the opportunity?

AAVE is the first lending protocol that taps into the stablecoin war. No doubt this is a new experiment that carries some risk. But based on the above understanding there are a few potential impacts that may change the stablecoin landscape:

AAVE governance carry a lot of power: AAVE governance have the ability to:

1. make change to interest rate

2. up or down the bucket of $GHO that a facilitator is entitled to

3. whitelist (so delist as well) a facilitator The power that AAVE governance can exert on $GHO is much bigger than the governance of other stablecoins protocols do to their counterparts. If $GHO indeed has a market, then the utility of $AAVE would increase for sure. Interested protocols may fight for AAVE’s voter for whitelisting for facilitator, increase in $GHO bucket etc.

Facilitator model adds in contract risks but can be a synergy play

With existing model, stablecoin protocols consider collateral as a pure ERC20 tokens, and the acceptance of the collateral is enacted by whitelisting the token address for collateral. This narrows the types of collateral accepted for one particular stablecoin. On the other hands, $GHO attempts to modularize the collaterals to a “Facilitator” entity, such that each facilitator can adjust the minting and burning of $GHO with regard to their own collaterals while AAVE governance focus on managing the aggregate risk by adjusting the bucket(upper limit) of $GHO each facilitator can mint. If this model works properly, we can imagine $GHO to be an inclusive stablecoin protocol that has both algorithmic, centralised and uncentralized stablecoins, RWA or off-chain assets, all to be accepted as collaterals. This would bring synergy and additional scale of economy to the stablecoin market.

Bootstrapping $GHO would be difficult, but if AAVE is determined there would be lots of incentives in the beginning. Based on the whitepaper, Aave lending protocol is going to be the first $GHO facilitator in the beginning, meaning that GHO can only be sourced through AAVE lending pool in the beginning. When there is no utility in the beginning, borrowing $GHO, and paying $GHO interest seems not profitable. I imagine AAVE protocol would introduce some incentive(s) to bootstrap the total supply of $GHO in the beginning. It can be a bribe to its curve pool , provide $AAVE incentives to the borrowers, or something brand new.

Over the long run, AAVE also have the ability to entitle part of its revenue to a staked pool of $GHO similar to the $LUSD model.

Disclaimer: Not financial advice. Do your own research.

Reference:

• Forum post: https://governance.aave.com/t/gho-development-update/10267

• Technical White Paper: https://governance.aave.com/uploads/short-url/6B2t8gC8Sf4WOafAcgOrdCg0Nka.pdf

• Audit: https://github.com/aave/gho-public/blob/main/audits/Aave GHO Audit Report Updated - August 2022.pdf