DeFi Liquidations & Market Impact
Empirical findings on collateral loss, fixed-discount vs auction mechanisms, and secondary market slippage
This section reviews the empirical literature examining how lending protocols execute liquidations and how protocol design choices affect borrower collateral loss, liquidator profits, and AMM price stability.
1. An Empirical Study of DeFi Liquidations
Citation: Kaihua Qin, Liyi Zhou, Pablo Gamito, Peter Jovanovic, Arthur Gervais. ACM Internet Measurement Conference (IMC), 2021.
Read Paper (PDF)
Core Takeaway (TL;DR)
Existing fixed-discount liquidation mechanisms (like Aave and Compound) successfully incentivize liquidators to clear debt, but force borrowers to surrender significantly more collateral than economically necessary.
Key Findings & Analysis
- Multi-Protocol Evaluation: Empirically analyzed on-chain data across Aave, Compound, MakerDAO, and dYdX over multiple market cycles.
- Excessive Collateral Loss: Fixed discounts (e.g. 5–10%) transfer oversized economic surplus from distressed borrowers to liquidators. Even when market liquidity is deep and execution costs are low, the borrower pays the full fixed penalty.
- Optimal Liquidator Strategies: Formulated an optimal liquidation game showing that liquidators maximize returns by strategically selecting positions, structuring partial repayments, and timing transactions.
- Economic Game Formulation: Demonstrated that a liquidation mechanism cannot be judged solely by whether bad debt is repaid; it must be evaluated on how much unnecessary value is extracted from borrowers.
Relevance to Reforge
Directly motivates Reforge's core mission: replace fixed collateral discounts with a competitive batch auction that minimizes borrower collateral loss while guaranteeing debt repayment.
2. Liquidation Mechanisms and Price Impacts in DeFi
Citation: Phoebe Tian, Yu Zhu. Bank of Canada Staff Working Paper, 2025.
Read Paper (PDF)
Core Takeaway (TL;DR)
Comparing fixed-spread (Aave) vs. auction-based (MakerDAO) liquidations reveals that competitive auctions reduce liquidation discounts and market price impact by over 70%.
Empirical Comparison: Aave vs. MakerDAO
| Metric | Aave (Fixed Spread) | MakerDAO (Dutch Auction) | Reduction with Auctions |
|---|---|---|---|
| Average Liquidation Discount | 5.0% | 1.8% | 🔻 64% lower collateral penalty |
| Secondary Market Price Impact | 0.054% | 0.010% | 🔻 81% lower market distortion |
| Liquidator Participation | Lower diversity | Higher competition | 🟢 More robust liquidity |
Two Competing Economic Effects
- Participation Effect: Auctions attract more competing liquidators because profit opportunities are flexible, ensuring unhealthy loans are cleared even during market distress.
- Competition Effect: Bidding competition pushes the clearance price closer to fair market value, meaning less borrower collateral must be sold to cover the same debt.
Relevance to Reforge
Serves as direct empirical validation for Reforge's thesis: auctions preserve borrower capital and reduce secondary AMM price slippage compared to static discount models.
3. Recent Advances: OEV and Gas-Sensitive Liquidations (2025–2026)
Recent studies in 2025 and 2026 explore the intersection of Oracle Extractable Value (OEV), transaction fees, and liquidation profitability:
- Participation Costs: Research indicates that when liquidator participation costs (gas, infrastructure) are high, auction efficiency can degrade unless the protocol explicitly optimizes for execution cost.
- Fee-Sensitive Manipulation: Models show that gas fees materially determine whether manipulation-based liquidations (oracle manipulation + frontrunning) are economically viable for attackers.
Design Takeaway for Reforge
Reforge's scoring formula explicitly incorporates Gas Cost and DEX Slippage Cost into Net Economic Value, ensuring the auction remains resilient even under volatile gas regimes.