eIP 15: Reduce Reserve Factors and Amend Interest Rate Models

  • Title: Reduce Reserve Factors and Amend Interest Rate Models
  • Author(s): Seraphim (delegate)
  • Submission Date: 12.07.2022

eIP 15: Reduce Reserve Factors and Amend Interest Rate Models

Simple Summary

This proposal offers to reduce reserve factors on all cross and collateral assets, as well as to amend interest models.

Motivation

The motivation behind the proposal is to reduce the spread between the borrow and lending APYs. As of now, the spreads are way too wide even by DeFi standards:

The reason is two-fold:

  1. Reserve factor is too high
  2. Utilisation has gone down during market volatility

On (1), reserve factor of 0.23 means 23% of all the interest borrowers pay goes into reserves, ie lenders only get 77% of all the interest. This is arguably too much and lenders should be receiving more interest.

On (2), utilisation has decreased across the board as fewer people borrow from pools to liquidity mine $EUL. To illustrate:

USDC:

WETH:

DAI:

WBTC:

Why does utilisation affect the spread? Because if the lending pool contains $1,000,000 and only $100 is borrowed, the interest the borrowers pay on $100 is distributed across $1,000,000 worth of lent assets. Alternatively, if the entire $1,000,000 is being borrowed, then lenders get a lot more interest on average.

How to address (2) then? Increase utilisation by moving the kink further out and/or lowering the interest rate around the kink. This way, your borrow vs lending APY spreads are reasonable even at high utilisation levels.

Implementation

Exact smart contract implementation will be posted shortly.

Current Model
Base IR Kink IR Max IR Kink% Reserve Factor
USDC 0% 4% 100% 80% 0.23
WSTETH 0% 8% 200% 80% 0.10
WETH 0% 4% 100% 80% 0.23
DAI 0% 4% 100% 80% 0.23
WBTC 0% 8% 200% 80% 0.23
USDT 0% 20% 300% 80% 0.23
AGEUR 0% 4% 100% 80% 0.23
UNI 0% 20% 300% 80% 0.23
LINK 0% 20% 300% 80% 0.23
ENS 0% 20% 300% 80% 0.23
MATIC 0% 20% 300% 80% 0.23
OSQTH 0% 20% 300% 80% 0.23
RBN 0% 20% 300% 80% 0.23
SHIB 0% 20% 300% 80% 0.23
MKR 0% 20% 300% 80% 0.23
CVX 0% 20% 300% 80% 0.23
PERP 0% 20% 300% 80% 0.23
AXS 0% 20% 300% 80% 0.23
New Model
Base IR Kink IR Max IR Kink% Reserve Factor
USDC 0% 1% 100% 80% 0.02
WSTETH 0% 1% 200% 80% 0.02
WETH 0% 1% 100% 80% 0.02
DAI 0% 1% 100% 80% 0.02
WBTC 0% 1% 200% 80% 0.02
USDT 0% 1% 300% 80% 0.02
AGEUR 0% 1% 100% 80% 0.02
UNI 0% 1% 300% 80% 0.02
LINK 0% 1% 300% 80% 0.02
ENS 0% 1% 300% 80% 0.02
MATIC 0% 1% 300% 80% 0.02
OSQTH 0% 1% 300% 80% 0.02
RBN 0% 1% 300% 80% 0.02
SHIB 0% 1% 300% 80% 0.02
MKR 0% 1% 300% 80% 0.02
CVX 0% 1% 300% 80% 0.02
PERP 0% 1% 300% 80% 0.02
AXS 0% 1% 300% 80% 0.02

Risk Assessment

Assuming we only change the reserve factors to 0.02 without amending the interest rate models, we get the following improvement in lending APYs:

Asset Current Utilisation Current Borrow APY Current Lending APY New Borrow APY New Lending APY RF Improvement
USDC 51.07% 2.55% 1.00% 2.55% 1.28% 0.27%
WSTETH 38.00% 3.80% 1.30% 3.80% 1.42% 0.12%
WETH 45.45% 2.27% 0.80% 2.27% 1.01% 0.22%
DAI 28.81% 1.44% 0.32% 1.44% 0.41% 0.09%
WBTC 29.24% 2.92% 0.66% 2.92% 0.84% 0.18%
USDT 20.95% 5.24% 0.84% 5.24% 1.08% 0.23%
AGEUR 58.39% 2.92% 1.31% 2.92% 1.67% 0.36%
UNI 32.26% 8.07% 2.00% 8.07% 2.55% 0.55%
LINK 37.23% 9.31% 2.67% 9.31% 3.40% 0.73%
ENS 44.58% 11.15% 3.83% 11.15% 4.87% 1.04%
MATIC 10.27% 2.57% 0.20% 2.57% 0.26% 0.06%
OSQTH 9.00% 2.25% 0.16% 2.25% 0.20% 0.04%
RBN 73.39% 18.35% 10.37% 18.35% 13.20% 2.83%
SHIB 7.87% 1.97% 0.12% 1.97% 0.15% 0.03%
MKR 0.00% 0.00% 0.00% 0.00% 0.00% 0.00%
CVX 43.47% 10.87% 3.64% 10.87% 4.63% 0.99%
PERP 74.78% 18.70% 10.76% 18.70% 13.70% 2.94%
AXS 0.00% 0.00% 0.00% 0.00% 0.00% 0.00%

These are rather decent in % terms, especially on the more liquid assets like USDC, WSTETH and WETH.

Next, if we amend the interest rate model WITHOUT changing the reserve factor. The rate model change is just changing the IR% at kink to 1%.

Asset Current Utilisation Current Borrow APY Current Lending APY Current Spread New Borrow APY New Lending APY New Spread
USDC 51.07% 2.55% 1.00% 1.55% 0.64% 0.25% 0.39%
WSTETH 38.00% 3.80% 1.30% 2.50% 0.48% 0.16% 0.31%
WETH 45.45% 2.27% 0.80% 1.48% 0.57% 0.20% 0.37%
DAI 28.81% 1.44% 0.32% 1.12% 0.36% 0.08% 0.28%
WBTC 29.24% 2.92% 0.66% 2.27% 0.37% 0.08% 0.28%
USDT 20.95% 5.24% 0.84% 4.39% 0.26% 0.04% 0.22%
AGEUR 58.39% 2.92% 1.31% 1.61% 0.73% 0.33% 0.40%
UNI 32.26% 8.07% 2.00% 6.06% 0.40% 0.10% 0.30%
LINK 37.23% 9.31% 2.67% 6.64% 0.47% 0.13% 0.33%
ENS 44.58% 11.15% 3.83% 7.32% 0.56% 0.19% 0.37%
MATIC 10.27% 2.57% 0.20% 2.36% 0.13% 0.01% 0.12%
OSQTH 9.00% 2.25% 0.16% 2.09% 0.11% 0.01% 0.10%
RBN 73.39% 18.35% 10.37% 7.98% 0.92% 0.52% 0.40%
SHIB 7.87% 1.97% 0.12% 1.85% 0.10% 0.01% 0.09%
MKR 0.00% 0.00% 0.00% 0.00% 0.00% 0.00% 0.00%
CVX 43.47% 10.87% 3.64% 7.23% 0.54% 0.18% 0.36%
PERP 74.78% 18.70% 10.76% 7.93% 0.93% 0.54% 0.40%
AXS 0.00% 0.00% 0.00% 0.00% 0.00% 0.00% 0.00%

The new spread is unsurprisingly tighter than the current spread.

Here’s what happens if we decrease the IR% at Kink to 1% and the reserve factor to 0.02:

Asset Current Utilisation Current Borrow APY Current Lending APY Current Spread New Borrow APY New Lending APY New Spread
USDC 51.07% 2.55% 1.00% 1.55% 0.64% 0.32% 0.32%
WSTETH 38.00% 3.80% 1.30% 2.50% 0.48% 0.18% 0.30%
WETH 45.45% 2.27% 0.80% 1.48% 0.57% 0.25% 0.32%
DAI 28.81% 1.44% 0.32% 1.12% 0.36% 0.10% 0.26%
WBTC 29.24% 2.92% 0.66% 2.27% 0.37% 0.10% 0.26%
USDT 20.95% 5.24% 0.84% 4.39% 0.26% 0.05% 0.21%
AGEUR 58.39% 2.92% 1.31% 1.61% 0.73% 0.42% 0.31%
UNI 32.26% 8.07% 2.00% 6.06% 0.40% 0.13% 0.28%
LINK 37.23% 9.31% 2.67% 6.64% 0.47% 0.17% 0.30%
ENS 44.58% 11.15% 3.83% 7.32% 0.56% 0.24% 0.31%
MATIC 10.27% 2.57% 0.20% 2.36% 0.13% 0.01% 0.12%
OSQTH 9.00% 2.25% 0.16% 2.09% 0.11% 0.01% 0.10%
RBN 73.39% 18.35% 10.37% 7.98% 0.92% 0.66% 0.26%
SHIB 7.87% 1.97% 0.12% 1.85% 0.10% 0.01% 0.09%
MKR 0.00% 0.00% 0.00% 0.00% 0.00% 0.00% 0.00%
CVX 43.47% 10.87% 3.64% 7.23% 0.54% 0.23% 0.31%
PERP 74.78% 18.70% 10.76% 7.93% 0.93% 0.69% 0.25%
AXS 0.00% 0.00% 0.00% 0.00% 0.00% 0.00% 0.00%

It is important to consider the fact that lower rates across the board may cause higher utilisation. This increases liquidity risk. Let’s see at what utilisation will borrow APYs match the current borrow APYs:

Asset Current Utilisation Current Borrow APY Current Lending APY Current Spread New Borrow APY New Lending APY New Spread Utilisation where Current Borrow APY = New Borrow APY
USDC 51.07% 2.55% 1.00% 1.55% 0.64% 0.32% 0.32% 80.314%
WSTETH 38.00% 3.80% 1.30% 2.50% 0.48% 0.18% 0.30% 80.281%
WETH 45.45% 2.27% 0.80% 1.48% 0.57% 0.25% 0.32% 80.257%
DAI 28.81% 1.44% 0.32% 1.12% 0.36% 0.10% 0.26% 80.089%
WBTC 29.24% 2.92% 0.66% 2.27% 0.37% 0.10% 0.26% 80.193%
USDT 20.95% 5.24% 0.84% 4.39% 0.26% 0.05% 0.21% 80.283%
AGEUR 58.39% 2.92% 1.31% 1.61% 0.73% 0.42% 0.31% 80.388%
UNI 32.26% 8.07% 2.00% 6.06% 0.40% 0.13% 0.28% 80.473%
LINK 37.23% 9.31% 2.67% 6.64% 0.47% 0.17% 0.30% 80.556%
ENS 44.58% 11.15% 3.83% 7.32% 0.56% 0.24% 0.31% 80.679%
MATIC 10.27% 2.57% 0.20% 2.36% 0.13% 0.01% 0.12% 80.105%
OSQTH 9.00% 2.25% 0.16% 2.09% 0.11% 0.01% 0.10% 80.084%
RBN 73.39% 18.35% 10.37% 7.98% 0.92% 0.66% 0.26% 81.160%
SHIB 7.87% 1.97% 0.12% 1.85% 0.10% 0.01% 0.09% 80.065%
MKR 0.00% 0.00% 0.00% 0.00% 0.00% 0.00% 0.00% 79.933%
CVX 43.47% 10.87% 3.64% 7.23% 0.54% 0.23% 0.31% 80.660%
PERP 74.78% 18.70% 10.76% 7.93% 0.93% 0.69% 0.25% 81.184%
AXS 0.00% 0.00% 0.00% 0.00% 0.00% 0.00% 0.00% 79.933%

As can be seen, the trade-off is that if borrow APYs are matched, the utilisation rises to about 80%. However, given the current market conditions and lower demand for leveraging up and $EUL liquidity mining (as can also be seen on Aave and Compound), it’s unlikely utilisation will persistently spike to unsustainable levels.

Even if the market appetite for leverage massively increased over a short period of time, we can quickly propose more conservative kink models. At this point in the market, however, I think it’s important Euler shows competitive interest rates with low spreads.

Voting

Vote here on snapshot:
https://snapshot.org/#/eulerdao.eth/proposal/0x526a71ca8f52c2df72f53aac6e908afcada883e33daca00d49211a7118d99c00

Conclusion

Lower RF and lower Kink IR will cause lower interest rates and tighter spreads between Borrow and Lending APYs, hence making Euler much more attractive for borrowing. The possible negative consequence is increase in average utilisation, which increases liquidity risk.

Given the current market conditions, it’s unlikely we’ll see spikes in utilisation. Hence we think it’s appropriate to prioritise the tightening of spreads.

5 Likes

Yes, I support this. The spread has stood out for some time on Euler. I think reducing the spread is the right thing to do to make the platform even more attractive than it already is.

1 Like

I absolutely agree with this proposal.

At the moment, the spreads in pairs are really wide and the market phase is good for making them smaller and trying to increase their utilization.

Generally speaking, agreed…

But i want to say maybe we should reduce later when mkt is stable or bullish.

when market will be bullish, you dont need to care about spread, demand will be high and wide spread mostly is not a problem if you really want to borrow.

Great eIP, globally agree with the proposition and the motivation.

Reducing reserve factor to 2% will improve efficiency and will make Euler closer to a Peer 2 Peer lending.
As for the kink I get the reason being this large decrease → to be more closer to other lending protocol APYs. But I think the best action would be to first decrease the kink such that the supply APY will be close to the one currently (i.e 1%). This is Pareto optimal, because suppliers do not loose anything and borrower are still getting a better APY than before. This can lead to an increase in borrow → increase in supply APYs → you can drop back the kink → Loop.

I think a more adaptive kink will be better overall for Euler :slight_smile:

1 Like

It seems like this proposal has a goal of increasing borrow utilization by making the rates that borrowers pay more attractive. This is challenging because by adjusting the IR model you’re simultaneously lowering the rate that suppliers earn, so it might have the unintended consequence of reducing supplied assets and reducing overall TVL.

I think it’s also very important to mention the critical function that reserves play in money markets:

  • Reserves decrease the risk of the protocol, by acting as insurance in case of an emergency, and act as a liquidity cushion for illiquid markets.
  • Increase supply interest rates - please correct me if I’m wrong here, but in Compound interest on reserves goes directly to the suppliers in each market, which increases the appeal of supplying liquidity
  • Excess reserves can either be returned to the community or used to pay for future improvements to Euler

I also want to mention that in the current market environment, where liquidity is very thin and the risk of further asset price decreases is significant, this is not the time to be increasing the risk in the protocol, it is the time to be building reserves and decreasing risk, so that suppliers have confidence that they will be able to withdraw their liquidity in the future, and won’t have to take a haircut due to insolvent debt.

It’s worth taking a look at a similar (but far less aggressive) proposal proposed by Gauntlet Network on June 22nd for Compound: Gauntlet Reserve Factor Methodology for the Compound Protocol - Proposals - Compound Community Forum

They proposed reducing ETH reserve factor from 20% to 18% and keeping USDC at 7.5%, based on their mathematical analysis. This proposal was still voted down, and while Robert and others that voted against it did not leave a forum post explaining their reasons why, we can surmise they are concerned about reducing reserve factor in extremely bearish and volatile market conditions.

I plan on voting against this proposal, but since there is still 4-5 days left in the voting period, would love to hear if people disagree with this.

1 Like

Agree here. I think it’s kinda pointless to talk bout risk where there’s no activity. Even aave only has less than 10 borrowers daily.