Some Ideas to avoid Liquidity Squeezes

Problem: Collateral factor for stables is too high. At some point its better for you to get liquidated then to pay back. We want that it hurts more to get liquidated.

Proposal: Make it more expensive to get liquidated

  • Change Collateral Factor of USDC/USDT to 85% and Collateral Factor of USDN to 75% (give people time to pay back. Timescale ~4-8 Weeks)
  • Change Liquidation Penalty from 5% to 7.5% for all stables.

Problem: Borrower are trapped with high APY because of significant liquidity outflow in short time frames and not because money was borrowed.

Proposal: Dynamic withdraw limits

  • Introduce Dynamic Withdraw Limits:
    If Utilization >95% set daily withdraw limit based on the total inflow in the last 24h of the underlying asset.
    Make it possible to reserve your withdraw so you actually dont have to withdraw but you can accumulate to withdraw if needed. Of course this reservation must decay if utilization normalizes.

Problem: People are scared to deposit during times withdraw is not possible.

Proposal: Make it more attractive to deposit during times withdraw is capped/not possible

  • Boosting Power for new deposits during high Utilization (>95%):
    Introduce a Treasury for times of high Utilization (1% of protocol fee) and use this treasury to boost the interest of new deposits during high utilization for a short time frame (2 to 4 weeks).
  • Boosting Power for old deposits:
    Use parts of the Treasury to boost long time participants based on the average deposit of the last 365 days.

Not good to make liquidation penalty too high. Liquidators typically sell the collateral, so after a certain point too high of a liquidation penalty ends up being counterproductive to successfully liquidating assets in a market squeeze in a broad market sense. There simply needs to be a fair pricing oracle and a small sufficient incentive for liquidations to happen.

Arbs will happen whether the gain is 7.5% or 2%. Leeway can be given for short term oracle fluctuations, but we are dealing with stablecoins so this is likely not necessary. Lowering the collateral factor and using external price feeds should be sufficient so that especially unhealthy positions can begin to be liquidated.

No comment on the first liquidation part because I’m not knowledgeable enough to say good/bad… But the second and third options there seem interesting at least. Again, I’m probably the wrong person to suggest how this improves (or negatively impacts) the protocol, but they’re certainly interesting.

I’ve seen some withdraw limit ideas floating around, and the biggest negative thing I see people saying is “it’ll take me too long to get my money out!”… But $1/day is literally more than $0/day, which we currently have. Balancing that with some sort of additional incentive to actually just stick around and wait it out (because that’s also the name of the game as a lender) is interesting too. Layer on that there’s dynamic changes to be introduced depending on the utilization and I just feel like there’s some way to optimize this for goodness.

Anyway, I hope smarter people than me read this and can absorb goodness from it. Thanks for sharing.