Bad Debt

What Is Bad Debt?

Bad debt in decentralized finance refers to a lending position where the value of the collateral is less than the value of the outstanding loan. When this happens, even a successful liquidation cannot recover enough value to make lenders whole. The difference between what is owed and what can be recovered is the bad debt: a realized loss to the protocol or its depositors.

In traditional finance, bad debt describes loans that a creditor writes off as uncollectible. DeFi lending protocols face the same fundamental problem, but the mechanics differ. Because DeFi loans are overcollateralized, bad debt should theoretically never occur: collateral should always exceed the loan value. In practice, market conditions can move faster than liquidation systems, creating windows where collateral value drops below the debt before anyone can act.

How It Works

To understand how bad debt forms, consider the lifecycle of a typical DeFi loan:

1. A borrower deposits collateral (e.g., ETH) and borrows a stablecoin (e.g., DAI or USDC) at a collateralization ratio above the minimum threshold
2. As the collateral asset's price fluctuates, the position's health factor changes: if it drops below 1.0, the position becomes eligible for liquidation
3. Liquidators repay part or all of the debt and receive the collateral at a discount (the liquidation bonus), which incentivizes them to act quickly
4. If everything works correctly, the debt is repaid and lenders suffer no loss

Bad debt emerges when step 3 fails or is incomplete. Several conditions can cause this:

Rapid Price Crashes

If the collateral asset's price drops sharply in a short period, the position can pass through the liquidation threshold and become undercollateralized before any liquidator acts. In extreme cases, the collateral may be worth less than the gas fees needed to execute the liquidation.

Oracle Delays

DeFi protocols rely on price oracles to determine collateral values. If oracle updates lag behind real market prices, the protocol may not recognize that a position is underwater until it is too late. The gap between the true market price and the oracle-reported price creates a window where bad debt can accumulate undetected.

Network Congestion

During market crashes, blockchain networks often experience severe congestion. Gas prices spike, transactions queue up, and liquidation bots may fail to get their transactions included in blocks quickly enough. This was a key factor in MakerDAO's Black Thursday event in March 2020, when Ethereum network congestion prevented timely liquidations.

Illiquid Collateral

Even when liquidators act promptly, they need sufficient market depth to sell the seized collateral. If the collateral token has thin liquidity, large liquidations can cause slippage that pushes the effective recovery below the debt amount. Liquidators may also simply refuse to participate if the available discount does not compensate for the price risk of holding an illiquid asset.

Simplified Example

Borrower deposits 10 ETH at $2,000 each = $20,000 collateral
Borrower takes a $15,000 stablecoin loan
Collateralization ratio: 133%
Liquidation threshold: 120% ($18,000 collateral value)

--- ETH flash-crashes to $1,200 ---

Collateral value: 10 ETH × $1,200 = $12,000
Outstanding debt: $15,000
Bad debt: $15,000 - $12,000 = $3,000

No rational liquidator will repay $15,000 of debt
to receive $12,000 worth of ETH.
The $3,000 gap is bad debt.

Notable Bad Debt Events

Several high-profile incidents illustrate how bad debt materializes in practice:

MakerDAO Black Thursday (March 2020)

On March 12, 2020, ETH's price dropped roughly 43% in a single day. Ethereum network congestion caused gas prices to spike, and MakerDAO's collateral auction system failed to attract sufficient bidders. Some liquidators won auctions with bids of 0 DAI, acquiring collateral for nothing while leaving roughly $6 million in unbacked DAI as bad debt. Maker governance responded by minting and auctioning new MKR tokens to recapitalize the system: the protocol's built-in debt auction mechanism.

Aave CRV Incident (November 2022)

A trader (later identified as Avraham Eisenberg) borrowed large amounts of CRV tokens on Aave V2 and attempted to manipulate the CRV market. When the position was eventually liquidated, the illiquidity of CRV on-chain meant that the liquidation could not recover the full debt. Aave was left with approximately 2.66 million CRV (roughly $1.6 million) in bad debt. Governance later passed AIP-144 to purchase CRV tokens using protocol treasury funds to clear the shortfall.

THORChain Insolvency (January 2025)

THORChain's lending and savers programs accumulated approximately $200 million in liabilities that exceeded the protocol's available liquidity. The lending design burned RUNE tokens on loan issuance and minted them on repayment, creating a price-dependent debt spiral. As RUNE's price declined, the system became insolvent. Validators paused THORFi operations and initiated a 90-day restructuring plan that proposed converting defaulted debt into equity.

Iron Bank vs. Alpha Homora (2021-2023)

In February 2021, an exploit on Alpha Homora resulted in approximately $30 million in bad debt owed to Iron Bank (a Cream Finance fork). The two protocols spent years disputing responsibility for the loss. In March 2023, Iron Bank unilaterally froze Alpha Homora's lending accounts, locking roughly $50 million in depositor assets. The incident demonstrated that bad debt disputes between composable protocols can persist for years and affect users who had no involvement in the original exploit.

Mitigation Strategies

Protocols use several mechanisms to prevent or absorb bad debt:

Safety Modules

Aave's Safety Module is a staking pool where users deposit AAVE tokens in exchange for rewards. If a shortfall event occurs, up to a defined percentage of staked tokens can be slashed and sold to cover bad debt. This effectively creates a protocol-owned insurance layer funded by token holders who accept the risk for yield.

Insurance Funds and Surplus Buffers

Many protocols maintain reserve pools funded by a portion of protocol revenue. MakerDAO's surplus buffer accumulates DAI from stability fees, serving as the first line of defense before MKR debt auctions are triggered. Perpetual DEXs like dYdX and DeepBook use insurance funds to absorb losses from underwater positions before socializing them across traders.

Debt Auctions

When reserves are insufficient, some protocols can mint governance tokens and auction them for stablecoins to recapitalize. MakerDAO pioneered this approach: if the surplus buffer cannot cover the bad debt, the system mints new MKR tokens and auctions them for DAI. This dilutes existing MKR holders but restores the system to solvency. It aligns incentives because MKR holders bear the cost of governance failures.

Loss Socialization

Some protocols distribute losses across all lenders proportionally. If bad debt occurs in a lending pool, the loss is spread across all depositors rather than concentrated on specific users. While this approach is simpler to implement, it means passive depositors bear losses from risks they did not choose. Protocols like Compound have historically used this model where bad debt reduces the exchange rate of cTokens, effectively socializing the loss.

Soft Liquidation

Curve's LLAMMA (Lending-Liquidating AMM Algorithm) introduces a gradual liquidation model where collateral is progressively converted into the borrowed asset as prices decline, rather than executing a single abrupt liquidation. If the collateral price recovers, the process can reverse (de-liquidate), potentially avoiding bad debt entirely. This market-based approach reduces the reliance on external liquidator bots but does not eliminate bad debt risk: sustained price declines can still produce losses, as seen in Curve's CRV-long lending market where roughly $750,000 in bad debt accumulated during 2025.

Risk Factors for Lenders

Depositors in DeFi lending protocols should understand several risk factors that increase bad debt exposure:

• Collateral quality: protocols that accept volatile, illiquid, or correlated collateral assets face higher bad debt risk than those restricted to high-liquidity assets
• Liquidation parameters: lower liquidation thresholds and higher liquidation bonuses reduce bad debt risk but also reduce capital efficiency for borrowers
• Oracle design: protocols using multiple oracle sources and circuit breakers are more resilient to price manipulation and stale data
• Historical track record: checking whether a protocol has experienced bad debt events and how they were resolved reveals governance quality and risk management maturity
• Reserve adequacy: the size of insurance funds or surplus buffers relative to total deposits indicates how much loss the protocol can absorb before socializing it

Why It Matters

Bad debt is the primary solvency risk in DeFi lending. When a protocol accumulates bad debt, the accounting no longer balances: depositors are owed more than the protocol holds. If depositors lose confidence and withdraw simultaneously, the result resembles a bank run: late withdrawers receive less than their deposits.

The interconnected nature of DeFi amplifies this risk. A bad debt event in one protocol can trigger liquidation cascades across others, as depositors withdraw to cover losses, reducing liquidity elsewhere. THORChain's 2025 crisis demonstrated how bad debt in a lending module can threaten an entire cross-chain protocol's viability.

For Bitcoin-native ecosystems, bad debt risk is a key consideration in protocol design. Systems like Spark that focus on payment infrastructure rather than leveraged lending avoid the overcollateralization and liquidation dynamics that create bad debt exposure, offering a different risk profile for users who want to hold and transact with Bitcoin and stablecoins without lending-layer solvency concerns.

Bad Debt vs. Related Concepts

This glossary entry is for informational purposes only and does not constitute financial or investment advice. Always do your own research before using any protocol or technology.