Market Risk Managers Recommendation
Below is a joint proposal from Gauntlet and @ChaosLabs.
Rocketpool rETH first began trading in November 2021, with supply at 182000, FDV at $320mm, and 30-day average daily volume at $6mm. rETH is a growing asset - whether in terms of holders, or total supply, particularly as more users stake their ETH with Rocketpool.
rETH is the third largest ETH liquid staking derivative, behind stETH/wstETH and cbETH, holding 3.5% of the market share of liquid staking derivatives. The quality of rETH as collateral can be directly compared to that of stETH/wstETH. As such, it is important to evaluate what components of rETH make its risk profile different from that of stETH/wstETH.
- rETH currently has a weaker liquidity profile than wstETH - As of 1/23/22, rETH → WETH swap of $27mm incurs 1% slippage, whereas a wstETH → WETH swap of $180mm incurs 1% slippage.
- As a liquid staking derivative, rETH is more flexible than wstETH. rETH can be redeemed immediately to ETH at the exchange rate via smart contract, conditioned on the Rocketpool ETH reserves, which are roughly ~5120 ETH. On the other hand, wstETH cannot be redeemed to ETH until post Shanghai upgrade, the only way to convert wstETH to ETH is via the market.
- rETH and wstETH follow similar price trajectories. Both assets represent some form of staked ETH (rETH represents staked ETH by validators with Rocketpool, stETH represents staked ETH by validators with Lido) that has the potential to be withdrawn at the upcoming Shanghai upgrade. As such, both assets naturally accrue interest over time (rETH and wstETH) and their prices have positive drift. However, there are +/-5% fluctuations in price between the two. Moreover, rETH was less affected by LSD discounts during June 2022 than stETH. This may be due to rETH not being enabled for collateral across the market, so cascading liquidations of rETH did not have tangible impact.
rETH/wstETH time series
We can also imagine the type of debt that rETH, because it is a liquid staking derivative, will support to be similar to that of stETH on Aave v2. The biggest use case for stETH on Aave v2 is to borrow WETH, roughly 57% of the debt supported by stETH is WETH, compared to 37% for stablecoins.
On the other hand, this is the distribution of debt supported by WETH collateral - 75% are stablecoins, whereas only 18% is WETH.
As such, we recommend the following parameters:
Isolation Mode - No
While isolation mode provides newer, untested assets a safe backdrop to evolve, by only allowing stablecoin borrowing, in the context of rETH however, it excludes potentially the largest use case of rETH collateral, which is to borrow correlated assets in WETH or wstETH. Enabling in isolation mode will likely not capture how rETH collateral will likely be used, and as a result, only delays the path towards understanding rETH dynamics on the Aave markets. Unless the definition of Isolation Mode includes correlated assets to rETH, we recommend against initializing in isolation mode.
Borrowable - Yes
We are aware of the risk that comes with enabling small LSD borrowing. However, disabling borrowing for rETH (given the profile shown above) may be excessive and misses out on likely use cases for rETH. As demand for liquidity for liquid staking derivatives increases, disabling borrowing cuts off a potential revenue opportunity for Aave v3 - specifically WETH/wstETH supply + rETH borrowing. We believe a better way to balance this intrinsic risk and opportunity would be to allow borrowing rETH, thus giving the protocol an opportunity to explore this potential form of usage but mitigate tail risks via stringent borrow caps. This will allow rETH more time to evolve in the context of Aave’s markets.
LT - 74%; LTV 67%
We recommend lower parameters compared to wstETH of 79.5% to capture this difference in collateral quality that we’ve identified above.
LB - 7.5%
We recommend initializing with an LB of 7.5% to give liquidators slightly more incentive to liquidate rETH collateral, as compared to wstETH.
RF - 15%
LPF - 0.1
Debt Ceiling - NA
Supply Cap - 10K
Borrow Cap - 1200
Our analysis shows that initializing rETH with borrows enabled but with a stringent borrow cap - would be most beneficial to see how rETH evolves on the v3 markets.
| Symbol | Isolation Mode | Borrowable | Collateral Enabled | LTV | LT | LB | RF | LPF | Debt Ceiling | Supply Cap | Borrow Cap |
|---|---|---|---|---|---|---|---|---|---|---|---|
| rETH | NO | YES | YES | 67% | 74% | 7.5% | 15% | 0.10 | N/A | 10K | 1200 |
Interest Rate Curve and Reserve Factor
rETH is similar to stETH, which has been listed on Ethereum v2 since last Spring, but since borrowing has always been disabled for stETH, we have no data about how users respond to changes in interest rates. As such, we recommend starting with conservative parameters that can be optimized later.
| Parameter | Recommendation |
|---|---|
| Base | 0 |
| Slope 1 | 0.07 |
| Uoptimal | 0.45 |
| Slope 2 | 3.0 |
| Reserve Factor | 0.15 |
Under these parameters, the borrower interest rate increases linearly from 0% at 0% utilization to 7% at 45% utilization, and then linearly to 307% at 100% utilization. This interest rate curve matches that of the more volatile assets on Aave (1INCH, CRV, ENS, LINK, MKR, UNI). From a risk perspective, interest rate curves need to be designed to reduce the chances of utilization reaching 100%, which would prevent stakers from withdrawing rETH and prevent liquidators from seizing rETH collateral when performing liquidations. The proposed interest rate curve is thus desirable because it uses a low optimal utilization and has a high maximum interest rate.
Question to the Community on E-Mode
One question which we would appreciate community feedback on is whether we should initiate rETH, cbETH, and other LSD markets with e-mode, or wait to see how usage evolves before initiating e-mode? Gauntlet and Chaos are currently working on aligning methodologies. If the community has a preference for enabling e-mode upon market launch, Gauntlet can provide recommendations in the meantime as we continue working with Chaos on aligning methodologies. We would value community feedback.
Below are tradeoffs:
- A benefit of not initiating with e-mode is to see how market usage evolves, and using that data to inform e-mode recommendations.
- A downside of not initiating with e-mode is that a primary use case for these LSDs is recursive borrowing, so not enabling eMode at launch means Aave is less attractive to user needs.
- It also may be important to have full functionality of the asset before making further parameter changes in order to understand how new Aave v3 features perform (regular mode relative to e-mode for that asset). Not enabling e-mode can delay the path towards understanding rETH dynamics on the Aave markets.