[ARFC] Increase Optimal Borrow Rates for Ethereum Stablecoin Markets

[monet-supply]

title: [ARFC] Increase Optimal Borrow Rates for Ethereum Stablecoin Markets
author: @monet-supply - Block Analitica
created: 2023-10-11

Summary

This ARFC proposes to make adjustments to stablecoin interest rate models across Aave v2 and v3 Ethereum markets.

Motivation

Over the past several months, equilibrium stablecoin borrowing costs across the broader defi space have trended upwards, and may now be higher than the governance defined “optimal borrow rate” set within Aave’s stablecoin interest rate models. For example, Maker borrow and supply (DSR) rates have been mostly over 5% since mid August. Rates in tradfi have also risen significantly, with short term treasury bills yielding around 5.5% and even long term treasury rates rising to nearly 5% across various maturities.

This mismatch between Aave’s rate models and the broader market leads to a range of potential inefficiencies, including deadweight loss from the cost of liquidity being artificially constrained away from equilibrium value, and negative UX impact due to greater rate volatility when utilization is above the optimal ratio.

First, we present data backing up the assertion that Aave’s current stablecoin rate models are not finding efficient market equilibriums. Then, we discuss the negative impacts of this including deadweight loss and UX impacts. Finally, we present a plan of adjustment that we estimate will address the market inefficiencies and lead to improved market utility across borrowers, suppliers, and the community of AAVE token holders.

UX Impacts

We gathered block by block interest rate data for the largest stablecoin markets across v2 and v3 (USDC, USDT, and DAI) from 1 August to 7 October, block numbers 17816255 to 18300344. Raw data can be downloaded here.

In the middle of this period (28 August), Aave governance executed proposal 306, which added sDAI as collateral on v3 while increasing the slope1 parameter for DAI to 5% on both v2 and v3. This offers useful data to A/B test the premise of whether increasing the borrow rate at optimal utilization can help improve market functioning.

Reviewing data for the Aave v3 DAI market, we can see that the increase in slope1 was strikingly effective at reducing borrow rate volatility, with the percent of time utilization was above optimal point falling from 62% to below 1%. There were some confounding factors, such as market turbulence over CRV leveraged positions at the beginning of August, and Maker’s elevated 8% DSR rate effective from 4 August to 18 August. But we can see heightened rate volatility from utilization exceeding the optimal ratio persists even excluding these periods.

Source: Stablecoins ARFC Market Data

Source: Block Analitica Aave Dashboard (DAI v3)

Comparing this with historical rates data for Ethereum v3 USDC and USDT markets (shown below), we can see a significant reduction in borrow rate volatility from the increase in the slope1 parameter. Note that the USDC market seems to exhibit higher volatility because it has a slope1 parameter of 3.5%, vs 4% for the USDT market.

Source: Block Analitica Aave Dashboard (USDC v3)

Source: Block Analitica Aave Dashboard (USDT v3)

We can expect reduced rate volatility to drive an increase in borrower participation. Paradoxically, raising the slope1 parameter for the DAI markets also seems to have been associated with a reduction in average (geometric mean) borrowing rates; for example, the average v3 DAI borrow cost from 1 August to 28 August (when proposal 306 was executed) was 6.49%, while the average rate from 28 August to 7 October was 4.56%.

Looking at the USDC v3 market over the same 28 August to 7 September period, we see the average (geometric mean) borrowing rates were 4.52%, so similar to DAI after the adjustment to interest rate models but with much higher volatility. This supports the notion that raising slope1 can bring greater rate stability without negatively impacting borrower welfare.

Deadweight Loss

Deadweight loss is an economic concept that describes the loss of total utility created by a market when prices are artificially constrained either above or below the optimal equilibrium where supply and demand are in balance.

It is commonly demonstrated with graphs of the supply and demand curve; any quantities exchanged below borrowers’ (consumers’) marginal willingness to pay represents consumer surplus, while quantities exchanged over suppliers’ (producers’) marginal willingness to produce represents producer surplus, and together these represent the markets overall utility.

When prices are artificially constrained above or below the market equilibrium, this leads to a lower quantity of goods exchanged, which reduces the total area of consumer and producer surplus and creates an area on the graph showing reduced utility known as deadweight loss.

With borrowing rates at optimal utilization falling below broader market levels, this can have the unintentional effect of constraining price below equilibrium value, which reduces overall utility of the Aave stablecoin markets. This has a likelihood to constrain market size growth over the medium term as suppliers are insufficiently incentivized to deposit more capital, while borrowers have a very low amount of liquidity available at reasonable prices before pushing the utilization above the optimal point and potentially pushing rates far above the market equilibrium.

Proposed Adjustments

We propose an adjustment to the variable slope1 parameter for stablecoins across Aave v2 and v3 Ethereum markets to better align optimal utilization rates with the broader market. Specifically, we proposed to increase variable slope1 parameters to 5% for each of the following assets: USDC, USDT, FRAX, sUSD, LUSD, GUSD, and USDP. No changes are proposed for DAI as it already uses 5% variable slope1 across both v2 and v3 Ethereum markets.

This will result in borrowing rates at optimal utilization still being equal or slightly below most other defi options (eg. Maker and Spark rates), while significantly increasing supply rates. Increasing the expected supply rates for stablecoins on Aave v3 Ethereum market may lead to a bit more downward price pressure on GHO, but this is not expected to be material considering that users can already borrow GHO with sDAI collateral (at 5% collateral yield, likely higher than the average for other stablecoin markets after this change) and Aave governance is also considering additional measures to support the GHO peg.

Specification

Ethereum v2 Market Changes

• Increase USDC slope1 parameter to 5%
• Increase USDT slope1 parameter to 5%
• Increase FRAX slope1 parameter to 5%
• Increase LUSD slope1 parameter to 5%
• Increase sUSD slope1 parameter to 5%
• Increase USDP slope1 parameter to 5%
• Increase GUSD slope1 parameter to 5%

Ethereum v3 Market Changes

• Increase USDC slope1 parameter to 5%
• Increase USDT slope1 parameter to 5%
• Increase FRAX slope1 parameter to 5%
• Increase LUSD slope1 parameter to 5%

Disclaimer

Block Analitica is not presenting this ARFC on behalf of any third party and is not compensated for creating this ARFC.

This proposal is presented on an “as is” basis and without warranty of any kind. This is not intended or offered as financial, investment, regulatory, or tax advice. Any assets or protocols referenced are mentioned purely for illustrative purposes, and this is not a recommendation or solicitation to engage with any asset or protocol.

Next Steps

1. Gather community feedback on this ARFC.
2. If community consensus is reached, escalate this proposal to the Snapshot ARFC stage.
3. If the Snapshot outcome is YAE, escalate this proposal to AIP stage.

Copyright

Copyright and related rights waived via cc0.

[Gauntlet]

Gauntlet supports the recommendations by @monet-supply. Stablecoin utilization on Ethereum v3 has been consistently at Uopt, and US1Y and US2Y yields have been north of 5%, implying that slope 1 could be raised without compromising on utilization.

We would also like to propose the following. We will share some more color below.

1. raise slope 1 to 5% for USDC, USDT, DAI, FRAX, LUSD across all v3 deployments
2. raise Uopt to 90% for USDC, USDT, DAI, FRAX, where applicable, across all v3 deployments
3. raise RF from 20% to 25% for USDC, USDT, LUSD on v2 Ethereum

On raising slope 1 to 5% for USDC, USDT, DAI, FRAX, LUSD for all v3 deployments

Utilization across other chains is also consistently at Uopt. Slope 1 across v3 deployments is 4% for USDT, FRAX, DAI and 3.5% for USDC. Supporting IR parity across deployments also serves to increase borrow revenue without compromising on utilization.

On raising Uopt to 90% for USDC, USDT, DAI, FRAX, where applicable, across all v3 deployments

AIP-306, in addition to raising slope 1 to 5% for DAI, also raised Uopt from 80% to 90%. The reduction in DAI borrow rate volatility can also be attributed to this Uopt increase, which allowed more borrowers to come in at slope 1 borrow rates. Gauntlet believes raising Uopt for USDT, DAI, FRAX can increase borrow revenue without adding excess risk.

• Currently, USDC is the only stablecoin with Uopt at 90%. The main considerations for setting Uopt to 90% is promoting enough liquidatable supply when liquidations of stablecoin collateral occur.
  • Assuming large market move, only 10% of supply is available for liquidation, instead of 20% should Uopt move to 90% and equilibrium utilization to rest at Uopt.
• Given the current loanbook for stables, Gauntlet believes increasing Uopt to 90% does not add excess risk. The below graph shows cumulatively how much of each stablecoin is supplied at each LTV for Ethereum v3.
  • as an example, only 10% of USDC is supplied at 0.6 LTV and higher
  • 10% of DAI is supplied at 0.55 LTV and higher

• Other v3 deployments show similar structure. The market would need to move considerably abnormally for liquidation capacity to come into question for stablecoin collateralized liquidations for 90% Uopt.

On raising RF from 20% to 25% for USDC, USDT, LUSD on v2 Ethereum

AIP-306 also raised DAI RF on v2 Ethereum to 25%. This would bring RF parity for USDC, USDT, and LUSD to DAI, and help encourage supply migration from v2 to v3.

[ChaosLabs]

Borrowing Interest Rate Slope 1 Increase

We concur with the analysis provided by @monet-supply regarding the necessity of increasing interest rates for stablecoins and the potential benefits it can bring. In terms of the optimal value, we propose setting UOptimal above 5%:

1. Our assessment of the average interest rates paid by borrowers indicates that the equilibrium borrowing rate presently hovers just above 5% for DAI and USDC, and approximately 6.5% for USDT. This is corroborated by data from other protocols.
2. The broader macroeconomic context and the consistent upward trend in interest rates over the last six months suggest an ongoing rise in stablecoin interest rates. To accommodate some flexibility, given that the exact equilibrium borrowing rate cannot be pinpointed down to a few basis points, we advocate for setting the UOptimal interest rate slightly above the equilibrium interest rate.
3. Interest rates are subject to fluctuations driven by diverse market dynamics external to traditional interest rates. In fact, in 2022, there was minimal to no correlation between stablecoin borrowing rates and interest rates on treasury bills. Part of the rationale behind slope 1 is precisely to offer a mechanism for balancing supply and demand across a spectrum. To ensure a seamless transition in the critical segment of the slope, i.e., around the equilibrium borrowing rate, it’s advisable to set the UOptimal borrowing rate slightly above this rate. Otherwise, it may experience frequent fluctuations on Slope 2.

In light of these considerations, our recommendation is to set the borrowing interest rates for Slope 1 for all USD stablecoins across Aave deployments at 5.5% as an initial step. Following this increase, we will continue monitoring the usage and equilibrium rate and make additional recommendations as necessary.

Other Changes

One of the most effective ways to gauge the responsiveness of participants in financial markets is to measure the impact of alterations. When changes are isolated, specifically by modifying only one variable, the assessment of sensitivity to this change holds greater significance.

Furthermore, adjustments to Slope 1 may lead to repercussions that could influence decisions regarding the increase of UOptimal or RF. We propose a cautious approach - initiate the elevation of Slope 1 first. After observing and assessing the impact, weigh the risks against the potential benefits of modifying other parameters.

[Gauntlet]

@monet-supply and Gauntlet have published the following two snapshots:
Increase slope 1 for stablecoins on V3

Increase Uopt for stablecoins on V3 and RF on V2 Ethereum

[monet-supply]

Both of the above Snapshot polls have passed. Hoping to move the proposed changes to an onchain proposal shortly!

[Gauntlet]

As an update, AIP-375 was executed. We thank the community for their participation.