Mitigating liquidity risk through the borrow interest rate model

Aave’s interest rate strategy is calibrated to manage liquidity risk and optimise utilisation. The borrow interest rates come from the Utilisation Rate $U$.

$U$is an indicator of the availability of capital in the pool. The interest rate model is used to manage liquidity risk through user incentivises to support liquidity:

When capital is available: low interest rates to encourage loans.

When capital is scarce: high interest rates to encourage repayments of loans and additional deposits.

To retrieve the interest rate strategy contract on-chain, see this section of the developer docs.

Liquidity risk materialises when utilisation is high, its becomes more problematic as $U$ gets closer to 100%. To tailor the model to this constraint, the interest rate curve is split in two parts around an optimal utilisation rate $U_{optimal}$. Before $U_{optimal}$the slope is small, after it starts rising sharply.

The interest rate$R_t$follows the model:

$if \hspace{1mm} U < U_{optimal}: \hspace{1cm} R_t = R_0 + \frac{U_t}{U_{optimal}} R_{slope1}$

$if \hspace{1mm} U \geq U_{optimal}: \hspace{1cm} R_t = R_0 + R_{slope1} + \frac{U_t-U_{optimal}}{1-U_{optimal}}R_{slope2}$

Both the variable and stable interest models, are derived from the formula above from the Whitepaper with different parameters for each asset.

When $U < U_{optimal}$ the borrow interest rates increase slowly with utilisation

When $U \geq U_{optimal}$ the borrow interest rates increase sharply with utilisation to above 50% APY if the liquidity is fully utilised.

Variable loans see their rate constantly evolving with utilisation. This means they are not ideal for financial planning.

Hence stable loans, that maintain their interest rate at issuance until the specific rebalancing conditions are met. For rebalancing the stable rate down, the loans stable rate$S$needs to be greater than the current stable rate$S_t$ plus a delta equal to 20%: $S \geq S_t + 20\%$.

For rebalancing the stable rate up, these two conditions need to be met:

Utilisation Rate: $U_t > 95\%$

Overall Borrow Rate, the weighted average of all the borrow rates: $R_O < 25\%$

The interest rate parameters have been calibrated per cluster of currencies which share similar risk profiles.

When market conditions change, the interest rate parameters can be adapted. These changes must adapt to utilisation on Aave’s market as well as to incentives across DeFi.

The interest rate curve of SNX is much higher than that of other assets due to the staking incentives of the Synthetix platform. This makes SNX the most utilised pool generating the highest yields.

With the rise of liquidity mining, Aave also adapted its cost of borrowing by lowering $U_{optimal}$ of the assets affected. This increased the borrow costs that are now partially offset by the liquidity reward.

Following the favorable historical review of liquidity risk, the interest rate models have been optimised to be more competitive while keeping theirs risk mitigation properties.

Asset | $U_{optimal}$ | Base | Slope 1 | Slope 2 |

BUSD | 80% | 0% | 4% | 100% |

DAI | 80% | 0% | 4% | 75% |

GUSD | 80% | 0% | 4% | 100% |

sUSD | 80% | 0% | 4% | 100% |

TUSD | 80% | 0% | 4% | 75% |

USDC | 90% | 0% | 4% | 60% |

USDT | 90% | 0% | 4% | 60% |

AAVE | | | | |

BAL | 45% | 0% | 7% | 300% |

BAT | 45% | 0% | 7% | 300% |

CRV | 45% | 0% | 7% | 300% |

ENJ | 45% | 0% | 7% | 300% |

ETH | 65% | 0% | 8% | 100% |

KNC | 65% | 0% | 8% | 300% |

LINK | 45% | 0% | 7% | 300% |

MANA | 45% | 0% | 7% | 300% |

MKR | 45% | 0% | 7% | 300% |

REN | 45% | 0% | 7% | 300% |

REP | 45% | 0% | 7% | 150% |

SNX | 80% | 3% | 12% | 100% |

UNI | 45% | 0% | 7% | 300% |

WBTC | 45% | 0% | 7% | 100% |

XSUSHI | 45% | 0% | 7% | 300% |

YFI | 45% | 0% | 7% | 300% |

ZRX | 45% | 0% | 7% | 300% |

The stable rate provides predictability for the borrower which comes at a cost, as the interest rates are higher than the variable rate. However the rate of a stable loan is fixed until the rebalancing conditions are met:

Utilisation Rate: $U_t > 95\%$

Overall Borrow Rate, the weighed average of all the borrow rates: $R_O < 25\%$

The currencies the most exposed to liquidity risk, TUSD, sUSD and BUSD, do not offer stable rate borrowing.

The base rate of the stable rate model correspond to the average market rate of the asset.

Asset | $U_{optimal}$ | Base | Slope 1 | Slope 2 |

BUSD | | | | |

DAI | 80% | 4% | 2% | 75% |

GUSD | | | | |

sUSD | | | | |

TUSD | 80% | 4% | 2% | 75% |

USDC | 90% | 4% | 2% | 60% |

USDT | 90% | 3.5% | 2% | 60% |

AAVE | | | | |

BAL | | | | |

BAT | 45% | 3% | 10% | 300% |

CRV | | | | |

ENJ | 45% | 3% | 10% | 300% |

ETH | 65% | 3% | 10% | 100% |

KNC | 65% | 3% | 10% | 300% |

LEND | 80% | 3% | 10% | 300% |

LINK | 45% | 3% | 10% | 300% |

MANA | 80% | 3% | 10% | 300% |

MKR | 45% | 3% | 10% | 300% |

REN | | | | |

SNX | | | | |

UNI | | | | |

WBTC | 80% | 3% | 10% | 60% |

XSUSHI | | | | |

YFI | | | | |

ZRX | 45% | 3% | 10% | 300% |

When market conditions change, risks change. The utilisation of reserves is continuously monitored to check liquidity is available. In case of prolonged full utilisation, the interest rate parameters are adapted to mitigate any risks emerging from market conditions

Date | Asset | Uoptimal | Variable Rate | Stable Rate |

| | | | |

This section shows Aave's interest rate curves per asset.

No stable borrows for BAL, CRV, REN, XSUSHI and YFI

| Token | aToken | Stable Debt Token | Variable Debt Token | Interest Rate Strategy |

BUSD | | ||||

DAI | |||||

GUSD |