Liquidation rewards and penalties: Who pays whom in DeFi?

Liquidation incentives and liquidation penalties: Two paths for DeFi's risky loan processing

Liquidation incentives and liquidation penalties represent two distinct compensation mechanisms adopted by the DeFi protocol for processing risky loans. In the incentive model, the liquidator receives additional collateral in return for repaying the debt; while in the penalty model, the borrower is charged an additional fee when the position is liquidated, which is retained by the agreement.

Who pays to whom depends on the underlying mechanism. In money market pools such as Compound and Aave, the liquidator receives rewards or discounts set in the agreement. This part of the gain comes from the value of the collateral obtained by the liquidator that exceeds the debt it pays off. In effect, the borrower paid the liquidator by losing additional collateral. In MakerDAO's treasury system, liquidation penalties are added to the treasury's debt, and the auction target is to recover the debt and penalties. Borrowers are required to pay penalties to the agreement, while auction participants still want to buy collateral at a price below the market price.

This difference has important implications for incentive mechanisms, liquidator participation, and agreement solvency. Higher rewards can speed up the liquidation process but also cause more loss of value to borrowers; higher fines can strengthen the funding buffer of the agreement, but must ensure that auction buyers are attracted. Governance sets these parameters, so each model reflects a different risk-reward balance.

How on-chain clearing rewards and penalties work

In the money-market model, liquidation takes effect as soon as a borrower's position hits a risk threshold. The Compound agreement sets a liquidation incentive factor, such as an 8% reward of 1.08, allowing a liquidator to obtain collateral worth more than he or she can repay his debts. The smart contract uses the current oracle price and exchange rate to calculate the number of tokens seized, and multiplies the amount repaid by the liquidation incentive coefficient to determine the liquidator's income. This logic can be seen in the Comptroller and CToken codes, and the formula for calculating the number of seized tokens is: actual repayment amount × liquidation incentive coefficient × borrowed token price ÷ (collateral price × exchange rate).

Aave adopts the same philosophy but uses different terms. It defines the health factors used to trigger liquidation, as well as a liquidation reward parameter that specifies the discounts or rewards that the liquidator receives when purchasing collateral. Aave's governance material also discusses rules similar to liquidation caps, and in the V4 design, proposes a floating reward plan that increases as health factors deteriorate to strengthen incentives.

MakerDAO's clearing mechanism is different. When a treasury is liquidated, liquidation penalties are added to its debt. The system then conducts an auction to sell the collateral, covering the debt and penalties. The fine is revenue from the agreement, rather than a reward paid directly to the liquidator, as in the money-market model. Maker's Clearing 2.0 system introduces Dutch-style auctions and their operating details, which the audit report notes are complex and sensitive to parameters.

Money market model: Borrowers provide incentive funds to liquidators

In Compound and Aave-style pools, third parties repay part or all of an insolvent loan and obtain collateral at a discount set by the agreement. This discount is reflected as a multiplier in Compound's clearing incentive coefficient and Aave's clearing incentive parameter. Because the value of the collateral acquired by the liquidator exceeds the amount of debt paid, the borrower's position will lose additional collateral in excess of the 1:1 swap ratio. This difference is a profit opportunity for the liquidator, but it is affected by the oracle price and execution costs. In this model, the agreement itself does not charge the borrower a separate liquidation fee, but implicitly achieves the transfer of value by transferring additional collateral to the liquidator.

These systems rely on liquidators and MEV searchers to compete with each other to perform liquidations as soon as they are profitable. The size of rewards affects the efficiency and reliability of clearing. Higher rewards can expand tolerance for slip points and Gas costs, making it more likely that off-chain robots will intervene in time. But the cost is that the borrower loses more collateral value, and if the reward is set too high, it may trigger unnecessary liquidation near the threshold window, which is an issue that governance needs to weigh when adjusting parameters.

Treasury/auction model: borrower pays penalty to agreement

When the vault's mortgage rate falls below the requirement, MakerDAO adds a liquidation penalty to its debt. It then negotiated an auction and sold enough collateral to pay off the debt and penalties. Fines are collected by the system, not by the liquidator. Liquidators, also known as "keepers," still try to make a profit by buying collateral at below-market prices in auctions, but their gains are not agreed payment rewards in the sense of the money-market model. Fines enhance the agreement's loss-absorbing ability and can offset auction costs, but they must be set to ensure that the auction can be completed efficiently.

At the operational level, this approach transfers value to the agreement rather than the first liquidator. It also introduces auction design options such as Dutch-style auction parameters and keeper incentives, which determine how close the actual selling price is to fair value. If auction parameters are improperly set or the market is not sufficiently liquid, it may extend the sales time and increase the risk that the proceeds from the sale of collateral will be less than the debt plus penalties, creating bad debts that the system must absorb.

Governance parameters and incentive adjustments

Liquidation parameters are settings controlled by the governance layer on the chain. In Compound, administrators can change the reward factor used to determine the amount of collateral seized through the_setLiquidationIncentive function. In Aave, reserve allocation includes clearing reward parameters and thresholds that affect health factors, and its governance material explores the impact of floating rewards linked to health factors in the V4 mechanism.

Changing these numbers changes behavioral patterns. Higher rewards usually increase keeper engagement and reduce outstanding non-performing loans, but they also increase the loss of value for borrowers. Higher fines can increase the amount of money the agreement retains during liquidation, but if fines or auction parameters are incorrectly set, the auction may underperform or inhibit bidders from participating. The DAO typically discusses, simulates, and implements these changes in stages because they affect MEV competition, bad debt exposure to the agreement, and user experience.

Example demonstration: Two clearing paths for the same position

The following example sequence shows the difference in "who pays whom" when the same insolvent position is processed using the reward model and the penalty model respectively. The amounts are only examples and are used to aid understanding and are not market data.

Starting point: The borrower's position exceeds the risk limit and meets the liquidation conditions.

Money market path (Compound/Aave style): The liquidator repays the borrower's 100 units of debt. With a liquidation incentive factor of 1.08, the system allows the liquidator to obtain collateral worth 108 units (calculated at oracle price). The additional 8 units of collateral are the total incentive for the liquidator. The borrower actually paid for the incentive through lost additional collateral. The agreement recovered 100 units of debt repayment and did not charge a separate clearing fee.

Treasury-Auction Path (Maker style): The agreement adds liquidation penalties to the borrower's debt, and then recovers the debt and penalties by auctioning collateral. Suppose the auction sells just enough collateral to cover 100 units of debt and penalties. The penalty belongs to the agreement. Bidders aim to buy collateral at a discount to the market price during the auction, which is their source of profit and is not related to fines.

Policy comparison:

Money market incentives: paid by the borrower through additional seized collateral, the recipient is the liquidator, and the parameter names include Compound's liquidation incentive coefficient and Aave's liquidation incentive parameters.

Treasury penalty: Payment by the borrower through the increased debt, the recipient is the agreement, and the parameter name is Maker's liquidation penalty.

Limitations, risks, special circumstances and common misunderstandings

Main limitations and risks include:

Parameter configuration is incorrect. An audit of the Maker Clearing 2.0 system warned that fines, auction times and keeper incentives must be accurately calibrated. Wrong settings can lead to loss of value or excessive auction times, creating the risk that debt cannot be covered.

Execution and liquidity risk. In the reward model, liquidators still rely on DEX and oracle liquidity to liquidate seized collateral. In stressful market environments, slippage can offset rewards and slow liquidation.

Price gaps and bad debts. Rapid market movements can push positions into severe insolvency before the robot can respond. Even with high rewards or penalties, the agreement could still face a funding shortage if collateral cannot be sold quickly at a price close to the oracle.

Keeper competes with MEV. If the reward is too small, only a small number of privileged searchers will profit after deducting Gas and slip points. If the reward is too large, the borrower's loss of value will increase, and opportunistic liquidations may be concentrated near the threshold.

Common misconceptions:

"Rewards are free money. "That's not the case. In the reward model, the borrower pays the reward through additional collateral seized, which is set by agreement parameters.

"Maker's fines were awarded to the liquidator. "The penalty belongs to the agreement. In Maker-type systems, the liquidator's profits come from buying collateral at a below-market price in an auction, rather than rewards paid by agreement.

"Changing parameters has little effect. "Adjustments to governance parameters can significantly change behavior and risks and usually require simulation discussions before implementation because of their systemic impact.

When will these mechanisms be exposed to

Borrowers face these rules when making mortgages. In Compound and Aave-style pools, published liquidation rewards tell you how much additional collateral a liquidator can seize if your health factor or mortgage rate falls below a threshold. In Maker-style vaults, liquidation penalties tell you how much additional debt will be added if your vault is liquidated. Both settings are visible in protocol documents and governance proposals, and may be changed through DAO decisions.

Liquidators and searchers adjust their robots based on reward mechanisms or auction design, including expected slip points and Gas costs. DAO contributors weigh parameter changes to strike a balance between timely liquidation, borrower costs, and protocol security. Understanding "who pays whom" helps clarify this balance: the reward model directs value from the borrower to the liquidator, while the penalty model directs value from the borrower to the agreement, but the auction still needs to attract competitive bids.

FAQs

Does an 8% clearing reward mean that the agreement pays 8% to the liquidator?

No. In reward models such as Compound and Aave, the liquidator obtains collateral in value that exceeds the debt it pays off. The borrower actually paid out this 8% through lost additional collateral, defined by the liquidation incentive factor or liquidation incentive parameter.

Who collects liquidation fines in MakerDAO?

Agreed collection. Maker adds fines to the treasury's debt and recovers the debt and fines by auctioning collateral. Fines are not paid to the liquidator.

If Maker does not pay rewards, how does the liquidator make a profit?

They aim to buy collateral at below-market auction prices and then sell it at or near market prices. The auction mechanism itself, rather than the rewards set by the agreement, creates profit opportunities.

Who changes the clearing reward or penalty, and how?

These parameters are set on the chain by the DAO. Compound provides an administrator function to change the clearing incentive coefficient, and Aave configures reserve parameters (such as clearing incentives). Because incentives and risks change systematically, governance proposals and simulations are usually made before changes.

What happens if liquidation fails during periods of extreme volatility?

If the collateral cannot be sold quickly enough or at fair value, the agreement may generate bad debts. Reward size, penalties and auction design are designed to alleviate this situation, but sudden price gaps and lack of liquidity remain risk factors.

Disclaimer: This article is for information reference only. Does not constitute and should not be used as legal, tax, investment, financial or other advice.