StableSims: Optimizing MakerDAO Liquidations 2.0 Incentives via Agent-Based Modeling

TL;DR

This paper uses agent-based modeling to optimize MakerDAO's Liquidations 2.0 incentive parameters, finding that increasing the constant fee reduces liquidation time more cost-effectively than increasing proportional fees.

Contribution

It introduces an agent-based simulation to analyze and optimize the incentive structure of MakerDAO's Liquidations 2.0 mechanism, a novel approach for this protocol.

Findings

Increasing the constant fee (tip) decreases liquidation time more effectively.

Optimizing incentives reduces costs while maintaining efficient liquidation.

Agent-based modeling provides insights into keeper behavior and protocol performance.

Abstract

The StableSims project set out to determine optimal parameters for the new auction mechanism, Liquidations 2.0, used by MakerDAO, a protocol built on Ethereum offering a decentralized, collateralized stablecoin called Dai. We developed an agent-based simulation that emulates both the Maker protocol smart contract logic, and how profit-motivated agents ("keepers") will act in the real world when faced with decisions such as liquidating "vaults" (collateralized debt positions) and bidding on collateral auctions. This research focuses on the incentive structure introduced in Liquidations 2.0, which implements both a constant fee (tip) and a fee proportional to vault size (chip) paid to keepers that liquidate vaults or restart stale collateral auctions. We sought to minimize the amount paid in incentives while maximizing the speed with which undercollateralized vaults were liquidated. Our findings indicate that it is more cost-effective to increase the constant fee, as opposed to the proportional fee, in order to decrease the time it takes for keepers to liquidate vaults.