The Aave V2 protocol implements a more aggressive IR for stable and variable rates compared to V1. Specifically, the risk parameters have been set as follows:
USDC — USDT
DAI — TUSD
Borrowers are taking this opportunity to borrow at a lower risk leading to a high share of stable borrows in V2. This is problematic as the stable rate only rebalances under stressed conditions, losing some of the risk management benefits of the variable rate.
The share of stable rates in V2 has significantly increased for all stablecoins:
- DAI from 18.57% to 65.3%
- USDC from 29.13% to 69.65%
- USDT 45.73% to 77.2%
The reason for that is that compared to the market conditions when V2 launched, many new stablecoin farms appeared, which dramatically increased the demand for stablecoins. The current implementation of the Lending rate oracle cannot keep up with the ever changing market conditions.
This shows to be unhealthy in the long term: too aggressive stable rates fail to attract new liquidity, which dampens the protocol growth and imposes more UX overhead (more difficulty to withdraw liquidity as there is high utilization).
The genesis team proposes two solutions:
the short term fix is to manually override the lending rate oracle (which is still under the control of the Aave genesis team) and impose a base stable rate of 7% (against the current 4%). A base that is almost twice as high will at the same time disincentivize stable borrowing for the short medium/term, or provide higher APY for LPs if the borrowers still choose to borrow at stable.
The long term fix is to get rid of the Lending rate oracle completely by algorithmically set the spread between the stable rate and variable rate. This can be achieved by calculating the ratio
Where Ds is the total debt stable and Dv is the total debt variable. When R > 1, it means more of the reserve is being borrowed at stable, which increases the spread between stable and variable for new loans. When R < 1, the spread can be lower, to incentivize stable rate borrowing. The equation for the spread can be adjusted to target whatever value of R.
This removes the need for the lending rate oracle altogether, which also reduces the centralization risk and decreases gas costs.