Authors: @nsoman AKA @soaringdove in Discord, or Nick Soman in real life.
Summary
When defaulted loans are fully or partially recovered, reimburse stakers, not just lenders and the SAFU.
Current system:
When a default happens, stakers lose 10% of their stake (value transfer to lenders).
If defaulted loans are recovered, lenders are made whole first.
ALL excess is kept by the DAO/SAFU to compensate it for handling the default process.
Stakers have no way to get any of their slashed stake, even if all money is recovered. This means a successfully recovered default is a money transfer from stakers to DAO/SAFU - regardless of the amount, or time spent.
Proposed system:
When a default happens, stakers lose 10% of their stake (value transfer to lenders).
If defaulted loans are recovered, lenders are made whole first.
SOME excess is kept by the DAO/SAFU to compensate it for handling the default process (to cover all money and time spent).
Remaining excess is redistributed to stakers.
Motivation
The TrueFi model relies on stakers. I’m operating on the assumption that these are big entities and defaults have a >50% chance of being recovered. The current system punishes stakers for staking even when defaulted loans are recovered, and there is a significant risk that they will stop staking.
Hi @nsoman thanks for your post. Before addressing your proposal I think it is worth pointing out a few things about the way the current default system works that do not seem clear from your post. I have numbered my thoughts to align with the numbers in your post as best as I can.
The current system doesn’t automatically cause stakers to lose 10% of their stake if there is a default. It can slash UP TO 10%. This is a key point. If the value of the staked TRU on the system is $100 million, the maximum slash to stakers is 10% or $10 million. But if the defaulted loan is only for $5 million, the full 10% slash won’t happen. Likewise, if the defaulted loan is for $20 million the most the stakers can lose is still $10 million. This is a capped downside risk. I think when considering the value that stakers are getting for their staking activity it is important to recognize they are getting this capped downside risk while still participating in both the TRU rewards for staking, a percentage of the successful loan interest payments, and platform governance. Compared to a similar structured finance transaction in “trad fi” this is a much better deal. So to give up a portion of potential default recovery isn’t that big an ask (which I expand upon below);
If a loan defaults, the SAFU makes the lenders whole and also collects the slashed TRU from stakers. This activity acts as an internal bank insurance program to give the stablecoin lenders confidence to provide liquidity. It is necessary to attract the lenders. The other benefit of the SAFU managing this process is that rather than immediately slashing and selling TRU from stakers, and pushing the prices lower by dumping on the market, is that the SAFU can make lenders whole from reserves and manage the sale of TRU over time to protect the value of the stakers “un-slashed” TRU;
/4. - The point about “excess” recovery really needs to be explored. The TrueFI protocol is an uncollateralized protocol, meaning that if there is a default there is nothing to sell to immediately recover losses. Instead, the DAO and SAFU would work together to bring a legal action to collect the defaulted amount. I would recommend everyone reading this spend some time researching how uncollateralized debts are collected in “trad fi” because even though we are dealing in defi the same universal truths apply. Primarily, the likelihood of recovering 100% of an uncollateralized debt through legal action is almost zero. If the borrower has some funds to pay back the defaulted loan and agrees to a workout with creditors it is most likely that the DAO/SAFU is one of many creditors that will be standing in line to be paid. In a best case let us assume that the defaulting party pays creditors $0.80 on the $1 owed. This is a highly unlikely outcome but maybe it happens. That still leaves a 20% gap that the DAO/SAFU has to make up, before legal and administrative fees, just to get back to even after making the TrueFi lenders whole. The amount, if any, recovered is even less if the defaulter claims bankruptcy protection. There is no excess. The chances that there will ever be an excess recovered is so slim that voting on which party would have “first claim” to a theoretical excess is fine but ultimately pointless. And, I would say propose the staker’s capped downside risk, and the freedom from the administrative duties of actually collecting the debt justify giving this theoretical excess to the DAO/SAFU.
So while I understand your motivation, I think it lacks certain realities. Even if chances of some recovery is as you say “>50%” that is different than saying it will be a full recovery, inclusive of legal and administrative costs. Rather than punishing stakers, the current system caps their downside risk and allows them to vote on the terms of credit portfolio they are staking on. In return it asks stakers to be active participants. I have no problem if you want to have a vote but I think it would make sense to recognize the chances of an “excess” recovery are virtually nil.
Thanks @tim.welsh. I appreciate this thoughtful response. I still think we should change the system, and hope to use an example to illustrate why.
You’re right, I should have written “up to 10%” and that is an important point.
Agreed, and I don’t think we should change any of this.
I agree that the point about “excess” should be explored. Let’s use the real world example of a defaulting $19MM loan.
The DAO/SAFU is currently seeded with 5MM TRU (~$3MM). I’m assuming that gets paid out to lenders first, and they are still owed $16MM. Right now there is $161MM TRU staked in the system, so a 10% slash to stakers on default would cover the remaining $16 MM lost. Lenders would be fully reimbursed immediately.
Now let’s say that in the future 80% of the $19MM loan is recovered (~$15MM). Lenders have already been fully reimbursed. As I understand the current system, 100% of that $15MM would go to the DAO/SAFU. I am proposing that we change this. It certainly makes sense that the DAO/SAFU should be reimbursed for money and time spent to recover the funds. They should also get first claim on restocking their seed money, assuming that is paid out to lenders first. But if the DAO/SAFU spent $1MM on recovery, they should get $1MM + $3MM for seed $ = $4MM of the recovered funds, and the other $11MM should go to the stakers. It won’t fully reimburse them - they’ll still be out $5MM - but it will not enrich the DAO/SAFU at their expense.
I welcome further conversation on this. Please let me know if I’m missing something important above.
Again I understand your point and appreciate the illustration. I think what you were originally talking about isn’t correctly termed “excess.” That was the point that I was originally reacting to; that in almost every scenario, even if a recovery is made by the DAO/SAFU there will be a deficit between what the SAFT paid out and what was collected.
I believe you want to talk about what to do with an amount of money that may be recovered over and above the value of the liquidated balance of the SAFU at the time of the default and the legal administrative costs (~$3MM and $1MM in your example). Let us agree to call the original SAFU balance and the legal costs the “Net Costs” and any amounts recovered above those two amounts the “Net Recovery” (in your illustration, it would be $11MM). If participants in the DAO’s governance want to vote to send that back to TRU stakers they have that option. It can be proposed and voted on at any time.
Even though DAO governance could vote to return any theoretical amounts of the Net Recovery to TRU stakers, here is why they shouldn’t vote to do that - diversification of the insurance pool.
As your illustration outlines, if the SAFU sells the original seed assets and immediately dumps ~$16MM TRU on the market, the price of TRU will immediately plummet given today’s liquidity. As the recovery plays out, even assuming the recovery rates in your illustration, the platform will issue more loans that don’t default, and hopefully, the price of TRU rebounds. The smarter thing for the SAFU to do with the ~$11MM Net Recovery would be to restock the SAFU’s asset buffer with non-TRU and uncorrelated assets, and to manage an insurance portfolio that can grow over time and lower the chance of needing to slash stakers in the case of a second default. That is what would have the most overall benefit to the platform and to holders of TRU (both stakers and non-stakers). Sending the Net Recovery to the individual stakers who were slashed during the first default encourages a death spiral for the platform. Instead of growing a reserve in the SAFU and lowering the chance of needing to perform a 10% slash in the future, it keeps leaving the SAFU at a bare minimum level. This would mean each successive default will likely require another slash. That hurts stakers over time and they will unstake. It hurts the price of TRU over time as it continues to need to be dumped on the market. And it hurts the platform as a whole because it has a lower and lower balance of reserves to pay liquidity providers in the case of a default. If this was put to a vote, I would encourage a vote against it.
(An aside - I think an 80% recovery estimate is wildly optimistic and not correlated to recovery rates in real legal action by an unsecured creditor, which is where the recovery will take place, the real world court system. I would highly recommend everyone do their own research into the recovery rates of unsecured creditors who bring legal actions. Based on my experience it is much more likely that the DAO wouldn’t recover the original balance and legal fees, i.e., it would recover less than $4MM in your illustration.)
Thanks @tim.welsh. I appreciate your thinking and I believe we’ve clarified where we disagree.
We are talking about a scenario in which a loan defaults, all lenders have been fully reimbursed, and later a recovery is made over and above the value of the liquidated balance of the SAFU at the time of the default and the legal administrative costs.
I think some or all of the incremental dollars should be returned to stakers who lost their money on the default, since they were harmed directly by this default.
It sounds like you think all of the incremental dollars should be kept by the DAO/SAFU (which could vote to do anything with them, including reimbursing stakers), in order to reduce the negative impact of future defaults.
There’s a fair amount of systems thinking and assumptions going into both of our recommendations, but I think that’s the crux of our disagreement.
LMK if you think that categorization is incorrect. I’d be curious to hear thoughts from others in the community too.