DeFi Lending and Borrowing: How Interest Rates Are Set

DeFi lending rates adjust automatically with pool utilization, not a bank decision, and overcollateralization protects lenders if collateral value falls.

There is no loan officer: it’s a pool and a formula

A DeFi lending protocol pools deposits from many suppliers into a single smart contract, and borrowers draw against that pool by posting collateral. There is no application, no credit check and no fixed term set by a person. Instead, the interest rate borrowers pay and the rate suppliers earn are both computed by the contract itself, recalculated with every transaction that touches the pool.

This is a fundamentally different mechanism from how a bank sets a savings or loan rate. A bank can hold a rate steady for months regardless of how much of its deposits are lent out. A DeFi pool cannot do that, because its only source of interest is the borrowers actively using it, and its only defense against running out of liquidity for withdrawals is to make borrowing more expensive as the pool empties out. The same pool-based mechanics that price a swap in an automated market maker, covered in our guide to liquidity pool math, show up here in a different form: a formula reacting to the pool’s own state rather than a person’s decision.

Utilization rate: the single number that drives the rate

The core input to almost every lending protocol’s rate formula is the utilization rate, defined simply as total borrowed divided by total supplied. A pool with $10 million deposited and $6.5 million currently borrowed has a 65% utilization rate. As Aave’s and Compound’s own documentation describe it, utilization is the one variable both protocols use to derive both the borrow rate and the supply rate for a given asset.

The shape of that relationship is what matters. Both major lending protocols use a two-slope, or “kink,” model:

• Below a target utilization (an “optimal” level the protocol sets per asset, often around 80%), the borrow rate rises slowly and roughly linearly as utilization climbs.
• Above that target, the rate curve gets dramatically steeper, so a small further increase in utilization produces a large jump in the borrow rate.

The logic is mechanical, not discretionary. A pool with plenty of spare liquidity can afford a low rate, since suppliers can still withdraw freely. A pool with almost nothing left to lend needs a rate high enough to do two things at once: discourage new borrowing, and pay suppliers enough that new deposits arrive before the pool runs dry for anyone trying to withdraw.

A worked example: the same pool at three utilization levels

Suppose a hypothetical stablecoin pool uses a two-slope model with a 0% base rate, an 80% optimal utilization (kink) point, a rate that climbs 4 percentage points across the first 80% of utilization, and then climbs a further 75 percentage points across the last 20%. Suppliers receive the borrow rate scaled by utilization, minus a 10% reserve factor the protocol keeps for its own safety module.

The jump between 80% and 92% utilization is the entire point of the model. A borrower who took out $6,000 when utilization sat at 65% was paying about $16.03 in interest over 30 days. If utilization climbs to 92% because other borrowers pile in, or because suppliers withdraw, that same $6,000 loan now accrues about $241.64 over 30 days, roughly 15 times more, with no change in the borrower’s own behavior. Run any of these annualized figures through our APR to APY calculator to see what daily or continuous compounding would do to the number over a full year.

Why lenders need overcollateralization

A lending protocol has no way to chase down a borrower who disappears, so it protects suppliers a different way: by requiring every loan to be backed by collateral worth more than the amount borrowed. This is overcollateralization, and it is the mechanism that makes an anonymous, permissionless loan possible at all.

Suppose you deposit 5 ETH as collateral at a hypothetical price of $2,000, worth $10,000, and borrow $6,000 in a stablecoin against it. If the protocol sets an 80% liquidation threshold for ETH, meaning it treats up to 80% of that collateral’s value as available to back a loan, your position’s health factor is:

Health factor = (collateral value × liquidation threshold) ÷ borrowed amount = ($10,000 × 0.80) ÷ $6,000 = 1.333

A health factor above 1.0 means the position is still safely collateralized. It falls whenever the collateral’s price drops or interest keeps accruing on the debt, and once it reaches 1.0, the position becomes eligible for liquidation.

Example: If ETH’s price falls 20%, from $2,000 to $1,600, your collateral is worth $8,000, and the health factor becomes ($8,000 × 0.80) ÷ $6,000 = 1.067, still above 1.0 but noticeably thinner. A further drop to $1,500, a 25% decline from your entry price, brings the health factor to exactly 1.0, the point at which liquidation becomes possible.

What actually happens during a liquidation

Liquidation in DeFi is permissionless: once a position’s health factor is at or below 1.0, any outside participant, usually an automated bot watching the chain, can step in as the liquidator. The liquidator repays part or all of the outstanding debt on the borrower’s behalf and, in exchange, receives an equivalent value of the borrower’s collateral plus a bonus, funded from that same collateral. This bonus is what makes liquidation profitable enough for bots to compete over it, which keeps the process fast once a position crosses the line.

For the borrower, this means the loss is not just the collateral used to repay the debt; it also includes the liquidation bonus paid to whoever closes the position. That is the real cost of letting a health factor drift toward 1.0, on top of whatever interest has been accruing at the pool’s current utilization-driven rate. Since collateral is usually a volatile asset and the loan itself may be a stablecoin, the risk here compounds the same kind of price exposure discussed in our guide to yield farming, just applied to a borrowed position instead of a liquidity pool share.

Reading a lending market before you deposit or borrow

A few checks apply whether you are supplying or borrowing:

• Check current utilization, not just the posted rate. A rate that looks attractive today can move sharply if utilization is already near a protocol’s kink point.
• Know your liquidation threshold and health factor, not just your collateral ratio at entry. The threshold, not your starting ratio, is what determines how much room you have before liquidation risk begins.
• Separate the base borrow rate from any incentive rewards layered on top. Some protocols pay extra token rewards to borrowers or suppliers; the framework in our guide to real yield versus inflationary yield applies here too, since a rewards-boosted rate can fall once emissions slow.
• Remember the smart contract itself is a risk, separate from the rate math. A lending pool holds every depositor’s funds in one contract; our smart contract risk checklist covers what to verify before committing capital.

Common mistakes to avoid

1. Assuming a posted APR is fixed. Both borrow and supply rates recalculate continuously as utilization moves, sometimes within the same block.
2. Treating the starting health factor as a permanent buffer. It erodes as interest accrues on the debt even if the collateral price never moves.
3. Ignoring correlation between collateral and debt assets. Borrowing a stablecoin against a volatile asset carries different risk than borrowing one stablecoin against another.
4. Confusing liquidation threshold with maximum loan-to-value. Protocols typically set a lower ratio for opening a new loan than the threshold used to trigger liquidation, leaving intentional headroom.
5. Comparing APR across protocols without checking the reserve factor. Two pools at the same utilization can pay suppliers different amounts depending on how large a cut the protocol keeps.

The bottom line

DeFi lending rates are not set by a person; they are computed from pool utilization through a formula that stays gentle at moderate usage and turns sharply punitive once a pool runs low on spare liquidity. On the borrowing side, overcollateralization and a monitored health factor are what let a protocol extend credit with no credit check at all, and letting that health factor drift toward 1.0 is what turns a normal loan into a liquidation.

This content is for education only and is not financial, investment, tax or legal advice. Crypto assets are volatile and you can lose money. Examples use hypothetical numbers. See our disclaimer and editorial policy.