Does 150% collateral actually stabilize the protocol?
So I had always the idea that the obscenely high collateral (150%) was inefficient/unnecessary but helped stabilize the protocol by preventing selling each time the price rises, but since @NonFungibleYokem made this post I’ve changed my mind.
My new thesis: paying rewards to NOs to have a high collateral increases volatility (initially to the upside, then to the downside) and destabilizes the protocol.
Evidence:
A overcollateralized node will benefit from volatility to the downside in 3 ways:
a) Highly overcollateralized NOs (>150%) become more capital efficient by being able to use all RPL for effective staking
b) Undercollateralized NOs (<10%) will be absent from the rewards denominator
c) Undercollateralized NOs (<10%) will buy additional RPL on the way down to try and keep up-essentially decreasing supply and stabilizing the market- even if this amount is insufficient to reach 10% collateral by reward snapshot.
overcollateralized nodes will see no benefit in RPL terms in buying more RPL unless ratio falls >93% (ie near total collapse)
undercollateralized nodes tend to be smaller, newer (see the nodes that have joined in the last 6 months) with (presumably) fewer external resources to purchase more RPL; this will be even more assured during LEB. They will not be able to stop a price collapse until it hits some of the whales at 50-80% collateral.
Conclusion:
The optimal scenario for a NO at 200% collateral is that ratio falls by 95% for several months before recovering, creating a perverse incentive that does not benefit the protocol.
The price stabilizing effect of the minimum collateral is damaged when more nodes are overcollateralized.
Possible solution to bringing down collateral, criticism welcome:
Use the carrot of LEB to lead people to a more capital efficient maximum collateral.
Currently, maximum effective collateral for a 16 eth node is 1333 RPL
Under 8E LEB as currently planned, maximum effective collateral for a 16E node is 4000 RPL
Under 4E LEB as currently planned, maximum effective collateral for a 16E node is 9333 RPL
If we made maximum effective collateral 50% on an 8E LEB, this 16 ETH node be allowed 1333 RPL. No node operator would lose effective stake by switching to LEB.
When we switch to 4E LEB, the maximum effective collateral can be 25%, or 1555 Rpl for that same 16Eth node. again no node operator loses effective stake.
Edit: space saving reply to valdorff below, probably read only by myself
The incentives I’m talking about is how the protocol incentivizes NOs, not accusations of malicious whales. And I’ll leave out RPL price and just talk about protocol benefit.
For instance:
The difference between 10% and 50% collateral might conceivably be useful for security. The difference between 110% and 150% is frankly not.
In the event of a 80% ratio downturn, the protocol strongly encourages NOs with initial collateral <50% to buy collateral to stabilize the RPL value and protect collateral; NOs with 150% will not be forced buyers until at 93% drawdown.
There is more yield in rewards alone from RPL at 150% than running more minipools, so RPL whales are disincentivized to run more minipools.
Having RPL locked until 150% discourages new NOs, who are largely buying in near 10%.
Any of us are welcome to accumulate 150% RPL because we have a bull thesis, but why is the protocol paying them to do so? What is the benefit to the protocol?
Essentially I think there are 3 takes:
FireEyes nailed it years ago and 150% is the perfect number.
150% is not the best number, but changing it now requires too much capital to be worthwhile.
150% is not the best number, and we should transition to where the best number lies.
I’m at #2 for vanilla minipools, but #3 for LEB because those tokenomics have not been finalized yet.
So… I agree that would be optimal… but I don’t agree that it creates much of an incentive. I don’t think any of us have a magic wand to make it drop and then go up as we wish. In reality, dropping 95% would be a catastrophe from which RPL may not recover, and the biggest holders would (obviously) be most damaged. I can assure you that I have never once heard anyone say “gee I wish RPL would drop so I would get more yield”. The closest I’ve heard is “gee I wish RPL would drop so I can buy more of it”.
NOs above 150% are true believers. If price drops a ton, I’d guess they’re quite likely to buy for speculative reasons - even if they don’t need to do so to keep their stake effective.
Regarding the solution, I’ve yet to be convinced there’s a problem. Minimum should scale with Protocol ETH to make sense as secondary collateral. To me, maximum makes intuitive sense to scale with minimum.
I’ve seen two main things in this thread:
We could reduce how much inflation goes to NOs and direct that revenue elsewhere
I’m clear on the potential protocol benefit here, at least
This is a tough call to make – we’re effectively trying to guess at future NO supply and rETH demand. My instinct so far has been that NO supply will stay ahead. Post-merge, I have definitely been surprised by how much rETH demand has rallied. I’m very unsure how the next 9 months will look with LEBs and rETH getting defi integrations. I still think NO supply will stay ahead, but I think it might take LEB4s. If that’s true, I don’t really want to reduce NO incentives before that.
We can take away yield from the largest holders, because it doesn’t serve a secondary collateral role
I’m not clear on a protocol benefit here, except insofar as this serves #1
I don’t love changing the game on folks, and this has a little of that flavor. That said, I think RPL’s value at this time remains more appreciation than yield, so it might not be too painful.
Interesting, so do you think we should revisit the conversation about minimum RPL collateral with LEBs? Arguments were based on the idea that we wont be NO constrained at all. Lowering collateral is an easy way to make NO side more attractive without relying on unsustainable levels of inflation.
Haven’t put a lot of thought into it, but potentially my suggested per minipool reward system would be helpful in removing the need for oDAO involvement in the rewards process, which would become relevant if we are looking to get rid of the oDAO eventually.
Just my two cents, but it seems to me like this could potentially be broken into two separate questions that don’t need to share the same solution. Heads up that I’m coming at this from the perspective of someone who wants to operate Supernodes in the future
What is the right collateralization ratio range to incentivize, and should there be a more complex/nuanced approach to issuing RPL rewards?
How much do we choose to arbitrarily restrict liquidity when it comes to RPL unlocks?
As an example, we could choose to allow RPL unlocks above 50% collateralization while retaining the existing reward structure that extends to 150% collateralization.
This would potentially help alleviate destabilizing liquidity crunches and subsequent mass unlocks/selloffs while simultaneously rewarding dedicated RPL holders. I strongly believe that liquidity crunch-induced volatility will negatively impact the general perception of protocol stability.
More importantly, it dramatically simplifies liquidity management for Supernode operators, who may in the long run be responsible for considerable ETH/RPL TVL. @knoshua 's solution for bucketizing Supernodes (An Alternative SaaS Design) exists because of the arbitrary friction imposed by the current RPL locking mechanism (as well as concerns about contract upgradeability).
Would lowering the RPL unlock threshold negatively impact RPL’s long-term value? Perhaps, but extending the RPL rewards window beyond the unlock threshold may mitigate negative price pressure in the near term (everyone who is heavily overcollateralized is a major RPL bull anyways). Down the line, implementing UEBs would solidify RPL as an asset worth holding in of itself, as it gives the holder the right to generate ETH cashflows.
RPL holders who’ve seen considerable price appreciation would be incentivized from a tax perspective to continue holding their RPL and utilizing it to run additional minipools instead of swapping over to ETH.
In summary, reduce RPL unlock threshold to improve UX and Supernode operator efficiency (it’s a bad look if the system’s design inherently drives validator churn), implement UEBs if they are sound from a security perspective (further enshrining RPL’s value in the system), and come to a reasonable agreement on RPL incentive range and any other unique reward dynamics (e.g. diminishing returns as you increase your collateralization level).
I think we should look after withdrawals with more data. I still think it’s most likely we are not NO limited, but postmerge rETH demand outpaced my expectations so I’m less confident than I was.
Is it? I could argue it’s a good look because it allows new participants. Obviously there’s a “too much” churn, but I think some turnover is probably healthy.
I think the largest benefit to reducing the max collateralization ratio is that NOs can obtain the optimal return with less total capital. Right now to get the highest % yield, you need to have a 150% collateralized node (due to RPL having a higher yield than ETH).
If you could get the same yield with 75% collateral (in terms of net RPL, since the 70% inflation to NO didn’t change), that frees up capital to deploy elsewhere, like minting rETH, joining RPL liquidity pools, or creating more minipools depending on which asset you prefer to hold/feel will appreciate more.
I think this ultimately is a benefit to node operators, incentivizing more minipools by making it require less capital for optimal yield, thus being more attractive. And there is potential that the capital which is unlocked could benefit the protocol if it is used on rETH, liquidity pools, or more minipools, though obviously not everyone will and it is unclear how much of an indirect benefit this would be.
I suppose my working assumption is that 90% of the time, those validators will be recycled back into the system anyways. Instead of allowing the user to withdraw RPL, liquidate, and spin up additional validators with their ETH, we’ve introduced a inefficient/expensive process that only serves to clog the activation/exit queues.
Even for an individual with a few minipools this is onerous and in a bull market, potentially very costly from a gas fee perspective.
I would only support a change to the rewards structure well after withdrawals are enabled, since everyone agreed to the current rules when they joined.
My biggest grievance with the idea of reducing the max collateralization rate is that the extra RPL liquidity privileges are not awarded to NOs with smaller collateralization rates. For instance, if the unlock % is set to 50%, a NO with 100% collateralization would be able to sell half their investment, but a NO at 50% would not be able to sell any. So there is a greater opportunity for sells at the cost of smaller % NOs which would not be able to do the same.
Worse, NOs that chose to put their ETH toward more minipools, perhaps closer to the “optimal” collateralization rate, would be punished, i.e. have more or all of their RPL locked, compared to NOs that invested half their ETH into Lido stETH instead for the same RPL amount locked.
Although I dont think we should do this as of now, I generally like the idea of this for one main reason:
Being potentially used as a tool to combat a lack of demand for minipools relative to demand for rETH (aka a full and stagnant deposit pool)- especially when LEBs are in place.
Idk how long a full deposit pool is acceptable, but obviously that can always be voted upon. I like that we would be able to easily calculate how many minipools would be freed up by lowering the max collateral level by X%.
Strongly against changing RPL collateralisation parameters.
Node operators who have committed capital previously, have done so based on their long term calcs with current parameters. Changing it changes their investment risk profile.
Also RPLs primary function is an insurance product to protect rETH holders, not primarily for yield. You get it back if you exit your minipool.
while this is a huge intervention into the tokenomics of the Protocol token, the hoped benefit to me isnt clear at all.
Tokenomics of the Porotocol Token itself should, if messed with at all, only be improved OR the protocol should evolve in a way that better makes use of the existing tokenomics like LEBs.
What should never be done, is worsen Tokenomics in retrospec. That simply destroys trust and conviction in the Protocol and will scare off new Nodeoperators even more to take on risk with a ERC20 Token like RPL that many are averse to anyways.
A lot of the discussion in the thread is along the lines of “Don’t fix what isn’t broken”
I would just like to point out that the status quo is a collateral of 24 ETH = 150% per minipool.
In the context of LEBs as currently described 150% collateral would be 36 ETH worth, or 72 ETH on 2x LEBs, so making no adjustment is still a significant change.
I am also strongly AGAINST this proposal for the following reasons:
The argument that high levels of collateralization causing lower collateralized folks to have their rewards diluted and that that is reducing the security of the protocol, I don’t find convincing at all.
RPL collateralization is changing a lot with Atlas release
There were concerns that the highly collateralized nodes are speculating. That might be so but that won’t change with reducing the maximal collateral. These are people that are waiting for RPL to gain massively on ETH, they are not stacking RPL for the staking APR.
Reducing the collateral is a big protocol change and doing it now without compelling reasons will bring risk.
The vote will very likely split the community as most of the whales and early adopters are against it.
Not in favor. Reducing RPL collateral requirements because it doesn’t contribute enough to protocol security is, in my view, the opposite of what needs to be done. Instead, we should look at how the existing RPL collateral can do a better job at its intended purpose.
Currently RPL doesn’t do a good enough job at being collateral for protocol security, being used only as a last resort vs. a catastrophic correlated slashing event. We could, for instance, be applying it to MEV penalties as well (being liquidatable fast than minipool-locked ETH is) or for contributing to a baseline APR (rather than all penalties above the 32 ETH balance being socialised between the NO and rETH.)
These are real flaws in the protocol which RPL could help alleviate, if it did a better job at what it was intended to do. This maintains RPLs social contract functionality-wise and investment-wise.
This would limit the max yield for operators at a time where we are NO-limited. This would directly harm our growth. The idea that “well, overcollateralized NOs will just sell the RPL for more minipools” is a very unlikely outcome IMO.
Proposing big economic changes while we’re also pushing ideas like LEBs is just multiplying the risk to the protocol. In the future it’s very likely we can make more productive use of this excess collateral (UEBs?).
More generally, this is changing the terms of engagement for node operators who made decisions about things like ETH vs. RPL asset allocation. It’s very bait-and-switch if we suddenly make a portion of their RPL non-yielding.
Just a question, which others have talked around to some degree–does the launch of LEBs present a one-time opportunity to reduce the collateral limit more fairly than if a change was made after-the-fact? If so, I think it should definitely be part of the LEB planning (if it’s not already).
I’d argue the opposite. Introducing LEBs adds protocol risk, so it’s specifically an inappropriate time to adjust other economic variables since that a) multiplies risk, and b) makes it hard to disambiguate results.
One could also argue that we’ve already “spent” the capital efficiency benefit by reducing commission.
I’m more inclined to support a reduction in the withdrawal of RPL below 150% (e.g., any collateral over 100% is unlockable) for the sake of capital efficiency than an absolute reduction in the maximum RPL collateralization. When it comes to the maximum RPL collateralization my preference is to let market participants sort it out on their own.
Nonetheless, I support coupling neither change with the inception of LEBs and, instead, studying such a change following the implementation of LEBs.