The Steem approach:
* witnesses publish price feed of STEEM in $ (like BTS/BitUSD);
* under normal conditions, Steem Dollar holders can convert their Steem Dollars to STEEM at median feed price with a delay (like force-settlements in BitShares);
* when median feed ($ per STEEM) is so low that total amount of existing Steem Dollar (current SBD supply) is more than 10% of the whole STEEM market cap (which is median price feed * current STEEM supply), the 10% threshold is used as conversion rate. In other words, the system breaks the peg intentionally at the time.
For bitAssets in BitShares E.G. bitUSD, it's possible to adopt similar approach:
* after a new price feed is published,
* if the median feed price ($ per BTS) is so low that would trigger a black swan event (which means median_feed / MSSR < black swan price, aka debt/collateral of the debt position with least collateral ratio), don't execute a global settlement, instead, update the margin call execution price to the black swan price;
* if the median feed price is higher, use the median.
In short,
1. use the black swan price as a floor when matching margin calls,
2. don't trigger global settlement.
Outcomes of this approach:
* when price of collateral asset (BTS) is too low, the peg may be effectively broken, which is same as globally settled;
* the debt positions are still in the borrowers' accounts, which IMHO is the most significant benefit to the ecosystem since decentralization (decentralized borrowing / supply creation) is kept as is, in comparison to the globally-settle approach which will close all positions to form a centralized pool;
* the borrowers who are in margin call territory can still increase collateral or reduce debt;
* new bitAsset supply can be created (borrowed) with sufficient collateral to eat the margin calls;
* bitAsset holders can buy into the margin calls at black swan price, which is same as globally settled;
* if there are orders selling collateral asset (BTS) with lower price, they'll get matched before the margin calls;
* after the position with least collateral ratio got bought out (completely eaten), or the debt position owner increased collateral ratio, the bitAsset may or may not be effectively revived immediately, depends on whether the next debt position has high enough collateral ratio;
* even if not revived immediately, new black swan price and margin call execution price might move down a bit, so the peg would be more or less tighter;
* if later a new price feed caused the median feed to be higher than current black swan price, the bitAsset is effectively revived immediately.
* the down side for debt position owners: it's possible that the positions with higher collateral ratio would sell collateral at even lower price, in comparison to the globally-settle approach that all positions will get "bailed out" at black swan price.
Other things that need to be taken into consideration:
* if force-settlement is enabled on the bitAsset, the settlement price should also has the same floor;
* ideally,
effective MCR should be kept at same level when margin call matching price or settlement price is higher than the real price, so people can't borrow with
effectively low collateral ratio. That said, the nominal MCR used in calculation may need to be increased.
Thoughts?
IMHO this approach is better than the bad-debt-holder approach proposed in
https://bitsharestalk.org/index.php?topic=27273.0By the way, although this requires a hard fork to implement, actually similar result can be achieved by witnesses before the hard fork: witnesses are able to feed adjusted price to prevent global settlement from being triggered.