New mechanism to handle bad debt (black swan)

[armin]

Wait this doesn't make sense. If a debt position is < 100% CR, lets say 80% CR, then this new account will take the 80% CR BTS and buy the bitasset and burn it? But the problem is that an extra 20% of the bitassets was issued by the blockchain. Effectively, the blockchain is "printing bitassets" out of thin air?

No, the new account will try to buy the 100% debt for the available 80% CR. This is probably to far away from the market price, so the order will stay on the book. As soon as the price recovers, the order will be filled and the full debt will be repaid.

(In reality, with MSSR > 0, the account will try to buy 100% debt for 110% CR. For large positions this will result in the same premium that we have had with the old margin call rule, at least temporarily until the feed price adjusts.)

The problem here is that the price is not guaranteed to recover. With BSIP18 at least the bitasset is "guaranteed" to be de-pegged. And it doesn't seem like you can de-peg a small percentage of the outlying bitasset...

[pc]

Wait this doesn't make sense. If a debt position is < 100% CR, lets say 80% CR, then this new account will take the 80% CR BTS and buy the bitasset and burn it? But the problem is that an extra 20% of the bitassets was issued by the blockchain. Effectively, the blockchain is "printing bitassets" out of thin air?

No, the new account will try to buy the 100% debt for the available 80% CR. This is probably to far away from the market price, so the order will stay on the book. As soon as the price recovers, the order will be filled and the full debt will be repaid.

(In reality, with MSSR > 0, the account will try to buy 100% debt for 110% CR. For large positions this will result in the same premium that we have had with the old margin call rule, at least temporarily until the feed price adjusts.)

[armin]

Wait this doesn't make sense. If a debt position is < 100% CR, lets say 80% CR, then this new account will take the 80% CR BTS and buy the bitasset and burn it? But the problem is that an extra 20% of the bitassets was issued by the blockchain. Effectively, the blockchain is "printing bitassets" out of thin air?

[clockwork]

I like it too.

Of course specifics would have to be thought out very carefully but eliminating that mechanism which I think we all agree is very very flawed is a huge incentive to do this right.

On a side-note, can we call the account: collection-agency ?

[abit]

I like it!

Not sure you've fully covered the effects if item 3 is only partially filled tho.

Where do proceeds from sale of undercollateralized debt positions go? Are they returned to the debtor or go to reserve fund or elsewhere?

Collateral will go to the one who bought it. Debt will go to "bad-debt-holder" then immediately be destroyed as described in OP.

I think a penalty for such undercollateralized debt position holders will help keep them more attentive to their collateral and dis-incentivize traders from operating with too little backing.

The penalty is MSSR.

Bad debt appears when: collateral / (feed_price*MSSR) <= debt

Although I feel strongly that we need to maintain at least 100% backing of MPA assets at all times, I feel it would be quite safe if collateral requirements were reduced while market is bullish, and committee could set a global collateral ratio. If that occurred about as infrequently as witness pay is adjusted (ofc that depends on the market), coinciding with a sustained bull market trend, it might work. Question for the committee is how long does "sustained" need to be? That's why I compared with how committee decides to maintain or change witness pay. Point being the CR level would only change with longer spanning trends, and not with every rise and fall of chart candles.

> "collateral requirements were reduced while market is bullish"

IMHO, in % sense, collateral requirements should be increased while market is bullish; then when market turns to bearish, there will be more spaces to release risks.

[Thom]

I like it!

Not sure you've fully covered the effects if item 3 is only partially filled tho.

Where do proceeds from sale of undercollateralized debt positions go? Are they returned to the debtor or go to reserve fund or elsewhere?

I think a penalty for such undercollateralized debt position holders will help keep them more attentive to their collateral and dis-incentivize traders from operating with too little backing.

Although I feel strongly that we need to maintain at least 100% backing of MPA assets at all times, I feel it would be quite safe if collateral requirements were reduced while market is bullish, and committee could set a global collateral ratio. If that occurred about as infrequently as witness pay is adjusted (ofc that depends on the market), coinciding with a sustained bull market trend, it might work. Question for the committee is how long does "sustained" need to be? That's why I compared with how committee decides to maintain or change witness pay. Point being the CR level would only change with longer spanning trends, and not with every rise and fall of chart candles.

[abit]

A bad debt means value of a debt position's collateral is less than value of its debt.

Currently, when a bad debt appears in the system, no matter how small the debt is, global settlement will be triggered and ALL debt positions will be forced closed. IMHO this mechanism is flawed.

Here I propose a new mechanism to handle bad debt, the core ideas are:

1. set up a special account "bad-debt-holder" in the system, with an impossible authority like null-account, so nobody can control it
2. when a new bad debt appears, don't trigger global settlement, instead, convert that debt position to a sell order which will never expire, sells its remaining collateral for its remaining debt, under the "bad-debt-holder" account
3. if the order get filled or partially filled, destroy the (debt) asset received.

In comparison to current mechanism, this new mechanism is much "softer" and more flexible. All state will be on the order book and easy to participate for all traders.

Thoughts?


Update (2018-10-21):

1. There is an edge case which slightly affect the implementation: according to BSIP35, when a limit order is too small, it will be cancelled, otherwise will lead to "something-for-nothing" scenario which usually messes up UI. If the limit order is owned by the bad-debt-holder account, we don't like it to be cancelled, instead, we treat it like a debt position (which will overpay when filling the last Satoshi of collateral).

2. In case when the bad-debt-holder owns several bad debt limit orders for a same asset, is it better to combine the orders into one limit order and average out the selling price? I tend to say "yes", because a) it frees memory, and b) it increases the chance that all bad debts get filled, although it may lead to a larger order hanging in the market which may add psychological pressure to traders.

Do you not think that global settlement as it's currently implemented should trigger if a majority of a bitasset's supply have defaulted on their loans though? Or should the proposed mechanism operate for the full 100% of supply?

IMHO we should operate the new mechanism for the full supply (don't trigger global settlement at all). However, perhaps we can have an option for asset owners to choose when there is no supply: which mechanism she would adopt when a black swan event occurs.

I think a penalty for such undercollateralized debt position holders will help keep them more attentive to their collateral and dis-incentivize traders from operating with too little backing.

Because bad debt appears when: collateral / (feed_price*MSSR) <= debt
So if MSSR is above 1 then there will be a penalty.

Perhaps we can another parameter here to replace MSSR though.