An interest rate model is the formula an onchain lending market uses to set its rates. It takes one main input, the share of the deposited pool that is currently borrowed, and returns a borrow rate. The rate paid to lenders is then derived from that. No party quotes a price; the pool calculates one from how much of it is in use.
In a bank, a credit committee sets lending rates, and in a pooled onchain market there is no committee in the loop, so the pricing is written into a contract in advance and governance changes the parameters rather than the rate. It is one of the main mechanisms behind where onchain yield comes from.
What does an interest rate model do?
The model converts one measurement into two prices, continuously and without anyone approving it. Depositors supply an asset. Borrowers take some of it against collateral. The contract measures the proportion borrowed and computes the borrow rate from a fixed curve. The deposit rate follows from the borrow rate.
Aave's rate strategy contract does this in a few lines. The supply rate is the borrow rate multiplied by utilization and then reduced by the reserve factor, the share the protocol keeps, and the contract expresses it as supplyRate = borrowRate * U * (1 - reserveFactor).
Two things follow from that single line. Lenders earn less than borrowers pay, because idle capital in the pool earns nothing and gets averaged in, and if nobody is borrowing at all the deposit rate is zero, a value the contract returns explicitly when there is no debt outstanding.
What is utilization?
The share of the pool lent out. Written from the contract: U = totalDebt / (availableLiquidity + totalDebt).
At 40% utilization, four in every ten deposited units are out on loan and six are sitting in the pool. Withdrawals come out of the six. Utilization is therefore two facts at once: it is the input that sets the price, and it is the inverse of how much of the pool can leave today.
Why does the curve kink?
Because a market needs a calm rate in normal conditions and a violent one near the point where withdrawals stop working, and one curve has to do both.
Aave's documented model uses two slopes, one below an optimal utilization point and a second from that point up to 100%, and below the optimum the borrow rate rises gently from a base rate along the first slope: borrowRate = base + slope1 * (U / U_optimal). Above it, the first slope is fully applied and the second one takes over across the remaining range: borrowRate = base + slope1 + slope2 * ((U - U_optimal) / (1 - U_optimal)). The contract enforces that the second slope is greater than or equal to the first, and constrains the optimal point to somewhere between 1% and 99%.
The kink is the market's only defense against running dry. There is no reserve requirement and no lender of last resort, so the steep section has to do the work: it makes borrowing expensive enough that borrowers repay, and it makes depositing attractive enough that new supply arrives. Other markets use differently shaped curves, and some adjust toward a target over time rather than mapping utilization to a rate instantly, but the purpose of the steep region is the same.
The lender's side when borrowers crowd in
The rate rises twice over. The number on a dashboard can therefore move faster than the underlying market. When utilization increases, the borrow rate goes up along the curve, and the multiplication by utilization in the supply rate goes up as well, so past the kink both effects compound on the steep part of the curve.
For a depositor this looks like the position suddenly working much harder. The same movement, read from the other side, says most of the pool has been lent out and the buffer that funds withdrawals has thinned, so a rate that has jumped reports on liquidity as much as on demand.
Why does a high rate mean a hard exit?
Because the amount available to withdraw is the part of the pool that is not lent out, and the rate is high precisely because that part is small. At 90% utilization, one tenth of deposits can be withdrawn; the rest is out on loan and comes back only when a borrower repays or a new depositor arrives. A position small relative to the free liquidity comes out fine. A large one waits.
The model is self-correcting, and the correction does work. High rates pull in supply and force repayment, so utilization normally falls back. What the correction does not give the manager is a time, and between the request and the liquidity the position is illiquid at exactly the moment the rate is telling everyone how attractive it is.
For an allocator this makes utilization a position limit, not just a yield input. Sizing against available free liquidity, establishing who sets the curve parameters, and watching what changes between governance votes are the standing tasks that follow from it. How DeFi lending works sets out the basic mechanics, and what a liquidation is explains what happens on the collateral side when rates rise.
Frequently asked questions
Who sets the parameters of an interest rate model?
Governance does, not the market. The base rate, the two slopes and the optimal utilization point are configured per asset and changed by proposal, and the contract enforces bounds on what can be set, including that the second slope must be greater than or equal to the first. Parameter changes are therefore a monitoring item for anyone holding a sizeable position.
Why do lenders earn less than borrowers pay?
Two reasons, both in the formula. The supply rate is scaled by utilization, so the idle portion of the pool dilutes the return across all depositors, and then the reserve factor takes a share for the protocol: supplyRate = borrowRate * U * (1 - reserveFactor).
What happens at 100% utilization?
Borrowing has consumed all available liquidity, so no depositor can withdraw until a borrower repays or new deposits arrive, and the rate is at its maximum precisely to make both of those happen quickly. The market is behaving as designed, but no withdrawal can be paid until then, and that distinction is what matters to a redemption schedule.
Is a high supply rate a good signal or a warning?
It is a measurement of utilization, so it is both. The rate says borrowers want the asset, and the same figure says the withdrawable portion of the pool has shrunk, so treating it only as a return figure, without sizing the position against remaining free liquidity, is how an allocator ends up unable to exit at the rate they were attracted by.