Bad debt scenarios
When a position becomes impossible to fully liquidate — because collateral has fallen below debt — the protocol absorbs the loss as bad debt. This is the worst-case for suppliers.
How it can happen
The basic flow:
- A borrower has supply of $V and debt of $D, with V > D × (1 + buffer).
- A crash drops V below D + bonus_compensation.
- By the time the oracle reflects the crash, the position is unsalvageable: even taking 100% of the supply, the protocol can't fully recover the debt.
How sized is the buffer?
The default buffer (50% over-collateralisation, derived from LTV 66.67%) is sized so that:
- A 33% instantaneous crash (~half the buffer) leaves comfortable margin for liquidation.
- A 50% instantaneous crash (eats most of the buffer) is the borderline.
- A 60%+ instantaneous crash exceeds the buffer; bad debt can occur.
The Merton jump-diffusion Monte Carlo in the whitepaper computes:
| Configuration | Bad debt at 99% CVaR |
|---|---|
| Default (LTV 66.67%, BUFFER 50%) | 0.02% of pool |
| Conservative floor (LTV 33%, BUFFER 200%) | 0.00% even at 50% crash |
So at the default, very rare tail events can produce ~0.02% bad debt. At the conservative floor, even severe crashes produce no bad debt.
What happens to bad debt
Bad debt reduces the pool's totalAssets(). This means:
- Suppliers' redemption value decreases pro-rata.
- The pool may take time to recover via positive interest accrual.
- In severe cases, governance may inject capital (from a treasury, if one exists).
There's no "socialisation switch" — the loss is automatic and proportional.
Mitigations
- Lower-LTV pools. A pool running at the 33% conservative floor has effectively zero bad-debt risk for crashes up to 50%.
- Diversify across pools. A bad-debt event in one pool doesn't spread to others.
- Watch the risk dashboard. Aggregate H distribution and oracle staleness are leading indicators.
What this means for suppliers
If you're supplying XPOW into a pool with diverse collateral types, your worst-case loss in a tail-event crash is bounded around 0.02% at default parameters. That's small but non-zero.
For users who can't tolerate any bad-debt risk, the conservative floor pools (when available) are the appropriate choice. They earn lower yields in exchange for the enhanced safety.
Where to go next
- Liquidation risk — the borrower's view
- Oracle staleness — why crashes can outpace the oracle
- Pool parameters — LTV and BUFFER configurations