Gauntlet's USDC.e Initial Migration Report

[gauntlet]

Gauntlet Methodology & Recommendations Report

---

Abstract (TL;DR)

This report outlines a three-fold focus:

1. A framework for lending protocols that aligns with the changes in liquidity pool (LP) positions as a result of the initial recommendations

   a. For Radiant & Aave: Reduce LTV & LT for USDC.e by 3% weekly after initial liquidity analysis is conducted

2. Initial recommendations for liquidity requirements, including suggestions about liquidity pools and GMX recommendations.

   a. For GMX: Decrease Target Allocation for USDC.e (numbers below)
   b. For GMX: Increase Target Allocation for USDC (number below)
   c. (Not in Scope) For Arbitrum: Incentivize liquidity for a USDC/USDC.e pool

3. An alerting plan around USDC.e and USDC given the prescribed deprecation plan for USDC.e in lending protocols, aiming to ensure a smooth transition for users and providers.

Protocol Recommendations

GMX Tranching Recommendations

USDC.e Target Allocation 30.0% → 27.5%

USDC Target Allocation 3.0% → 5.5%

As the largest liquidity source for USDC.e in the Arbitrum ecosystem, this initial recommendation will serve two purposes:

1. Initialize the transition away from USDC.e in the GMX V1 pool
2. Provide data on how liquidity providers and trading volume respond to protocol changes.

We will monitor the changes and fold the analysis into our future recommendations based on the methodology below. We are also planning to review the buffer amounts for each of these assets on GMX V1 in the near future.

Liquidity Pool Considerations and Recommendations

USDC/USDC.e swaps on ETH.ARB should be stable, cost-effective and have low volatility.

(Not in scope, NIS) Gauntlet recommends to incentivize a liquidity pool [ which we will refer to as POOL ] with a robust set of lever(s) to adjust liquidity provider fee(s), in order to facilitate the USDC/USDC.e swap for global users. We recommend the Arbitrum Foundation and broader community push towards a concentrated, one-stop, solution to establish the market depth and mechanisms needed to successfully facilitate the migration from USDC.e on ARB. The example linked below is a current USDC.e/USDC pool with approximately 4.5M of TVL.

https://info.uniswap.org/#/arbitrum/pools/0x8e295789c9465487074a65b1ae9ce0351172393f

Aave and Radiant recommendations

We propose a weekly 3% reduction in the Loan-to-Value (LTV) and Liquidation Threshold (LT) for USDC.e on both the Aave and Radiant platforms, aiming to encourage users to switch to USDC. Alongside these adjustments, we plan to communicate with the respective communities in advance of any expected liquidations due to these changes. These recommendations will be made after collecting and analyzing preliminary data on changes in liquidity. Please take a look at the methodology section below for more details.

Introduction & Summary of Previous Report

In our Initial Migration Report, we outlined the following as the most relevant dimensions of market risks in context of the active ARB.USDC.e migration:

1. Price: Risk of asset prices moving unfavorably.
2. Liquidity: Risk of being unable to buy/sell assets in expected fashion.
3. Volatility: Risk of unexpected price changes.
4. Flow: Risk related to asset flow and redemption requests.
5. Credit: Risk of counterparty failure.

For this methodology and recommendations report, we want to highlight the trade-offs between a fast vs. slow migration, provide explicit bounds for liquidity, and identify user migrations to consider.

Towards a Proposal: Additional Detail

Some Prescription(s)

While these recommendations are ultimately up to the community to agree on, we want to highlight that the intention of these mechanisms is to ensure the net edge of asset swaps remains around 1 basis points (bps) (100 bps = 1%). While incentive optimization is out of scope of Gauntlet’s current engagement with the Arbitrum Foundation, we note to the community that edge levels in crypto (especially for stablecoin transfers) are often thought of in % when they should be thought of in bps. The issue is often the lack of competition that allows liquidity providers to charge more than they should. From an ecosystem perspective, a necessary condition of a globally optimal asset migration is minimized transaction cost. Put simply, ARB users shouldn’t have to pay much net-net to transfer USDC.e risk into USDC. As a consequence, liquidity provision of the USDC/USDC.e swap shouldn’t have that much “juice.”

What do we mean by “Juice”?

Juice is edge capture. Markets are profit-driven and the entities that provide liquidity to them do so for profit-driven intents (or they blow-up). As a result, an activity like asset swap liquidity provisioning needs to be profitable for the entities engaging in it [usually market makers, but we can’t deny that retail flow loves to punt here]. Game theoretically, these entities want their businesses to be as profitable as possible. As a result, they will “charge a lot” (e.g. wide spreads, front-running, other forms of indirect market impact, etc.) until someone else comes in specifically to compete against them. This emergence of competition and eventual “race to the bottom” (Howard Marks Memo) shows up in all liquid asset classes (Stocks, Bonds, Commodities, FX, Derivatives &c). There is no reason to expect anything different in cryptocurrency markets [SEE: US ETF Market Structure].

If the Arbitrum community wants to be forward-looking, it should condition expectations about market maker(s) and transaction cost(s) accordingly.

Localizing to the USDC/USDC.e Asset Swap

Barring weird market events, stablecoin prices are generally within [0,1]. As a downstream consequence of real numbers (or division algebras, if you will), for any two stablecoins X, Y their exchange rate E = X / Y can be any non-negative real number E. In the context of the Arbitrum USDC.e migration, what we are particularly concerned with is the USDC/USDC.e exchange rate which will be referred to as ER. In an ideal case, ER = 1 at all times. Not only would this require an ungodly amount of liquidity (more than needed for migration success), but in the same way as expecting 100% USDC.e deprecation on a medium term timeline is unreasonable, so is expecting ER = 1 at all times. A more suitable target is where. In effect, we expect the exchange rate between USDC and USDC.e to be normally distributed around 1 with variance to the tune of a 5 vol asset.

State of the Migration

In this Methodology & Recommendations Report, we will consider these methodologies in context of the USDC.e Migration at both the protocol and ecosystem level. We begin with fresh mark(s).

2023-08-16 06:00 PM	USDC.e	USDC
Total Supply	772,142,584	192,426,060
Numbers of Addresses	609,825	52,092

2023-09-15 10:00 AM	USDC.e	USDC
Total Supply	658,522,640 (-14.71%)	184,852,466 (-3.94%)
Numbers of Addresses	620,574 (+1.76%)	68,697 (+31.88%)

Protocol Level Recommendations

Objective

The objective for this section of the report is concentrated around mitigating risks & ensuring the stability of the ecosystem throughout the migration from USDC.e to native USDC on the Arbitrum Layer 2.

Risk Mitigation in Lending Positions

Objective

To ensure that lending positions involving both USDC.e and native USDC do not expose the ecosystem to outsized risks.

Lending Protocol Methodology

Our recommendation framework is summarized below.

We will suggest decreasing LT and LTV on USDC.e by 3% per week on Aave and Radiant to promote the migration to USDC. To accompany these changes, we will message communities when liquidations are expected as a result of the proposed.

An example alert can be seen below:

Attention: As of date, the data shows that the following accounts would be liquidated if these param changes were adopted. The list below specifies the USD value of collateral that borrowers would need to supply to reach a health factor of 1.1, on the assumption that their account’s composition of collateral types remains fixed.

Supply and Borrow Cap Changes

For lending protocols with supply and borrow caps:

Depending on the community preference, we propose a conservative supply and borrow cap of 80% of the circulating token supply on-chain or an aggressive borrow cap will be 120% of the circulating token supply on-chain.

• It should be noted that while the total_supply metric refers to the maximum number of tokens that can be generated, the circulating_supply metric provides a more accurate measure of the liquidity that is currently available in the market. This is due to the fact that the circulating_supply metric takes into account the tokens that are in circulation and readily accessible for trading, while the total_supply metric encompasses the entire quantity of tokens that can exist, regardless of whether they have been released to the market or not."

Liquidity constraints for liquidations

Of the 400+ liquidations that have happened on the Arbitrum blockchain since March 1, 2023 involving USDC.e, no significant liquidations have atomically swapped out of their positions. Below is an example of a liquidation that happened in June of this year. Below is the transaction flow of one of the largest liquidations of USDC.e in this time period. As a result, our borrow and supply cap changes are far more relaxed than Gauntlet’s general borrow and supply cap methodology. https://eigenphi.io/mev/eigentx/0xbfd3f31b9cb2e044e6be5303f0f683589fb47d7163f87f864eee4aa3e87fa4a6

Historic USDC price dislocation

See below for the price dislocation event and the corresponding impacts on liquidity for USDC.e on the Avalanche blockchain.

Liquidity Provisions and Support for Liquidity Providers

Objective

The objective for this section of the report is concentrated around identifying potential risks to liquidity providers and defining alerts for each potential risk when possible

Real-Time Monitoring

A real-time dashboard will be developed to monitor trade volumes, liquidity pools, and price discrepancies between USDC and USDC.e.

Notifications and Alerts

Alert mechanisms will be set up to notify liquidity providers of substantial shifts in trading volume or liquidity conditions, thereby enabling them to adjust their positions in a timely manner.

Prioritized Risks

Macro

1. Complicated UX could dis-incentivize USDC.e laggards from adopting USDC
2. USDC.e asset(s) being unbridged from Arbitrum might not return as USDC
3. Bridging problems
4. Disjointed migration process
5. Any knock-on effect(s) of asset flow in the context of unbalanced ecosystem incentive design
6. A faulty mechanism or system design causing either Circle, ARB users or protocols on ARB to unnecessarily lose collateral

Micro

1. Liquidity could dry up or pools could become unbalanced
2. Stable Asset(s) / Exchange Rate(s) could become Unstable
3. Likelihood of both liquidations and insolvencies could increase if protocol parameters are not suitably conditioned on volatility expectations in the context of public and private liquidity
4. Potential cross-protocol pricing discrepancies caused by nonstandard oracle usage
5. Token Transfer Issue(s) / Cross-Chain Asset-Liability Mismatch(es)

Preliminary Definition(s)

NIS: “NOT-IN-SCOPE”

SWAP: USDC/USDC.e ; ER: the “effective price” of SWAP

TOKEN: the “effective price” of asset T ∈ {USDC, USDC.e}

For a given asset T:

DPG_T is the de-peg threshold.

LIQ_T is the liquidity threshold.

BUF_T is the liquidity buffer.

POOL_T is the current amount of tokens in POOL

Methodology

Below we summarize strategies for responding to potential risks as they pertain to the dimensions of market risk identified above.

Price

Event(s)

• USDC.e goes below 1
• USDC goes below 1
• ER trades above 1 towards $\infty$

Response(s)

• If TOKEN < (1- DPG_T) for asset T, alert that T is de-pegging
• If ER >> 1 and ER’ > 0, identify if there is a pricing issue that is increasing demand (at the microstructure level) for the USDC leg of SWAP then alert accordingly
• If TOKEN >> 1 and TOKEN’ > 0, identify if there is a pricing issue that is increasing demand (at the microstructure level) for TOKEN then alert accordingly
• [NIS] If ER << 1 and ER’ < 0 and there is no heightened credit risk to Circle, conduct automated creation of USDC collateral and supply liquidity to deepen SWAP depth in POOL
• [NIS] Design embedded reactive mechanism like Aave Killswitch

Liquidity

Event(s)

• USDC liquidity in POOL dries up
• USDC.e liquidity in POOL dries up
• Relative liquidity between USDC and USDC.e get “out of whack”
• Liquidity Providers pull out of POOL

Response(s)

• If POOL_T << LIQ_T, alert that T’s liquidity in POOL has dried up
• If POOL_T ≥ LIQ_T AND POOL_T < (1+BUF_T)LIQ_T, alert that T’s liquidity in POOL is in the process of drying up
• [NIS] Liquidity Provider / Liquidity Taker Wallet-Level Tracking & Analytics
• If POOL_T > (1+2*BUF_T)LIQ_T, alert that POOL has excess T collateral
• [NIS] Dynamic fee adjustment in context of maker/taker behavior
• [NIS] Queuing Structure of Liquidity Provision (viz. price-time v. price-pro-rata)

Volatility

Event(s)

• USDC volatility spikes
• USDC.e volatility spikes
• Liquidity providers pull out from POOL due to heightened forward variance expectations regarding USDC, USDC.e and/or SWAP

Response(s)

• Market alerts in Telegram

Flow [NIS]

Event(s)

• Spike in volume flowing from USDC —> USDC.e
• Spike in volume flowing from USDC.e —> USDC
• Liquidity Providers pull out from POOL due to aggressive liquidity taking
• Liquidity Takers don’t utilize POOL due to UX complexity
• Liquidity Takers don’t utilize POOL due to maker impact /transaction cost expectations

Response(s)

• [NIS] Market-Condition-Dependent “White Glove” Risk Management

Credit [NIS]

Event(s)

• Circle defaults on its USDC and/or USDC.e liabilities
• Market Risk(s) damaging the value and/or access to Circle’s collateral (e.g. TradFi Banking insolvencies, funky duration + curvature moves in the risk-free world, repo market stress, government seizure, etc.)
• Liquidity Providers are short USDC and cannot deliver collateral
• Liquidity Providers are short USDC.e and cannot deliver collateral
• Liquidity providers pull out from POOL due to collateral delivery issues

Response(s)

• Market-Condition-Dependent “White Glove” Risk Management

Next Steps

• Share GMX, Aave, and Radiant recommendations to their respective communities
• Deploy alerting & real-time dashboards for scenarios covered in the Ecosystem Risks section