Concerns about long-term RPL tokenomics

I’m very excited about Rocket Pool and just started running my first minipool. But I’m not very convinced that the insurance system is solid and future-proof. I’ll lay out my concerns and propose a different solution. There are still some gaps in my knowledge about the protocol so please let me know if I missed or misunderstood anything.

It’s basic economics that insurance costs money. Rocket Pool has adopted a clever system that currently allows node operators to earn money on their insurance. The gist of my concern is that this only works as long as Rocket Pool is in its growth phase, when new entrants are effectively paying existing node operators.

It seems that the primary demand for RPL currently comes from the need for node operators to provide RPL as collateral to insure new minipools. The primary supply comes from a 5% yearly inflation. There is currently rising demand due to new entrants wishing to create minipools. Hence, there is strong upward pressure on the price of RPL that will continue as long as Rocket Pool is growing.

This growth phase will end at some point, at which the number of minipools that are created and destroyed is the same. At this point, market supply of RPL (5% inflation + minipool exits) should be expected to exceed demand (minipool entrants), so one would expect RPL to depreciate slightly against the value of ETH. At this point node operators lose money on their RPL collateral, so one would expect node operators to reduce their collateral to the minimum of 10% (currently the average collateralization is 53%), further increasing supply and reducing the value of RPL.

If node operators are smart, they will reduce their collateral to 10% before a “run on the RPL” occurs. Since you can continue to operate existing minipools below the 10% minimum this might allow people to run minipools with very limited insurance. They will miss out on RPL rewards, but at this stage, it could be more profitable to have (say) a 5% collateral without rewards and a greater portion of your capital staked (which yields the normal ETH staking rewards and minipool commission).

So, unless I have missed something, it seems a depreciating RPL will become a problem at some point in the future.

A better system

Now on to what I think could be a better insurance system. Instead of a complex tokenomics, we should have a token that is fully backed by rETH in a single pool. It gets minted when people deposit rETH as insurance for their minipool, and destroyed when people liquidate their minipools. This way, the value of minipool collateral with respect to ETH will stay close to the initial value and slightly increase over time (say, >1.6 ETH per minipool). The only costs that operators are incurring for insurance in this system is the difference between minipool operator rewards and rETH rewards. Most importantly, this system remains stable over time regardless of whether Rocket Pool is growing or shrinking. I also think it’s a nice idea that node operators are (indirect) rETH holders. Having a stake in the system as a whole might induce more responsible behavior with respect to DAO voting.

Please let me know what you think.

1 Like

i like this idea, but it calls for a complete revamp even for RPL that are left out of the equation.
i concur that when the RPL APR will go under 8/10% the incentive to have more than 10% collateralization will fade.

i think the rocketpool team have some aces in the hole regarding this, but should be programmed to execute fast, at max right after the merge.

1 Like

It was pointed out to me on the Discord and in VVander’s blog series (Rocket Pool 2.0 Tokenomics, Pt 1: - HackMD) that the problem is what to do with RPL when switching to a different system. Once it’s announced that RPL will be phased out, its price will likely crash, depleting current insurance reserves. I don’t think there is a costless solution to this problem.

I think there should at least be the following change to the current system. Right now 30% of RPL inflation goes to the oDAO and pDAO. It would be better if the oDAO and pDAO are funded by an alternative system with a fixed cost per minipool. I think this could be achieved by reducing the current commission by say 5% (but I don’t know if such a change is possible with existing minipools). This has two big advantages IMO:

  • It increases incentives for node operators to keep their minipools collateralized at 10% or more. If all inflation is going to RPL stakers it is still possible that NO’s staked RPL value will remain stable even if RP is not growing (but not shrinking too much).
  • The current system in effect means that RPL holders are paying for the oDAO and pDAO, although these costs are currently offset by rising RPL prices due to the influx of new node operators. NO’s with little collateral are free-riding on these costs. It is fairer if the costs are the same for each minipool.
1 Like

5% per NO is steep. That would massively discourage new node operators from joining. You’re likely looking at somewhere between .25% and 1%.

Consider how much ETH is currently sitting in the insurance pools for RPL in addition to the amount of locked ETH from rewards. We have over 6000 minipools.

If we did something like .25% of NO rewards and .25% rETH rewards committed to the dao funds for funding oDAO and pDAO rewards. Until withdrawals are enabled, the funds would be locked, but would be released/freed up at that point.

Another option would be to represent that as either the same sort of rETH liquid staking token, but possibly just a rewards token that represents the amount of ETH that is locked up as the rewards mechanism.

Now, that could just be a modification to RPL to have it work like rETH and to use a small portion of the .25% or whatever small % taken for the oDAO and pDAO from the rewards.

In the longer-term, while I dislike removing some of the staking reward of a NO and/or the rETH holders, I think this promotes healthy rETH, RPL, and NO/minipool growth, while reducing/removing much of the RPL token value risk.

Essentially, it would be a strong asset-backing to RPL that would lower the amount of speculation.

1 Like

This is the core of your post. I vaguely agree with the premise, but (A) this may not crop up for a long time and (B) this is a gentle effect rather than a runaway effect.

(A)

  • If we currently used all RPL to make minimum-RPL minipools, the supply of rETH would be worth 44% of the current supply of stETH. In other words - even if we literally used every last drop of RPL perfectly, it’s still well within reason that there would still be demand for RPL. Worth noting - Lido is big, but we should be expecting quite a lot of LSD growth still.
  • This applies the same to the LEBs described in https://dao.rocketpool.net/t/variable-minipool-size-w-rpl-stake-based-on-protocol-eth/685/55 as well, which maintain the same amount of rETH to RPL value.

Keep in mind, that’s with both perfectly minimal stakes and zero RPL speculation. With current ~40% stakes and no speculation, we’d only be able to create 11% of the current value of stETH in rETH. Obviously the amount increases as the RPL:ETH ratio increases, but I still think we’re years away from getting to a steady state where inflation can play a significant role.

(B)
Let’s assume the RPL:ETH ratio has gone to .1 (about 10x current). We’ve made much more rETH with this RPL - in fact, we made 440% of the current value of stETH by having everyone have perfectly minimal RPL stakes.

  • Let’s assume that this is stable for a bit and then, due to inflation, the ratio drops a little bit to .09.
  • Ok. Now I decide I’m done with RP. I exit my minipool.
  • Now someone else goes to start a new minipool. Unfortunately, they can’t. I was only able to sell 90% of the RPL they need to start a minipool due to the ratio drop.
  • Ie, any drop in the ratio creates demand when we’re in steady state. Demand supports the ratio.

So you see - we have an effect that self-limits, rather than self-reinforces.

If this is how it goes, we’ll see the RPL ratio climb until RP reaches steady state. Then it will mostly follow rETH demand. If there’s more demand, folks will be incentivized (by arb) to create minipools which will make them more willing to buy RPL, which will increase the ratio (but not too much because new entrants will need increasingly less RPL as the ratio increases). When there’s less rETH demand, folks will be incentivized to exit minipools, which will free up RPL and decrease the ratio (but again… only to a limited extent because new entrants will need increasingly more RPL as the ratio drops).


It may be that the above is enough to make it stable. Or not, but we have quite a lot of time to figure out where to go when we have more information about the actual market behavior.

I think there are options. I think we have time to figure them out. I think we’ll be better informed in the future.

Plausible options

An example option off the top of my head: redirect a small portion of ETH commission to RPL buyback and burn to counter inflation.

Another more convoluted option: we could end RPL creation, and allow minipools to be created with either 16 RPL or 1.6 ETH of extra collateral – the pDAO could take its stream from minting and holding rETH using those “extra” ETH (ie, on exit the NO only gets 1.6 ETH). This would effectively make RPL an ETH stablecoin since it’s just being used as a substitute.

Another: Maybe we could swap to what you’re saying using some kinda RPL buyback system? Getting rid of RPL is very hard. It’s part of the system and removing it without either advantaging or disadvantaging stakeholders is extremely difficult.

Thanks for your reply. I grant your point (A), under the assumption that Rocketpool keeps growing for a long time. I think we should be worried though about a new decentralized staking system being introduced with a better business case.

I’m not following you in (B). A drop in the RPL:ETH price “creates” demand only if the price drop itself was not a response to changes in supply and demand. The issue is that supply outpaces demand in an inflationary currency.

  • Now someone else goes to start a new minipool. Unfortunately, they can’t. I was only able to sell 90% of the RPL they need to start a minipool due to the ratio drop.

If we suppose the price drop resulted from an RPL supply inflation, there should be additional RPL in circulation that new entrants can buy. The new entrant can buy 90% of the RPL from you and the remaining from others who received a part of RPL inflation.

Here’s something we should be able to agree on. Given the assumptions that
(1) we are in a steady state, that is, the number of minipools is constant for the foreseeable future,
(2) minipool collateralization is kept at 10% for every pool,
(3) the DAOs receiving minted RPL spend at least some of it,
the price of RPL should drop. This seemed intuitively clear to me, but since not everyone seems to be with me I’ll try to give a more structured argument.

Consider a single round of inflation in which NOs buy additional RPL collateral, there are some new entrants and an equal amount of exits, and the DAOs spend a little bit of RPL.

Let P1 be the price of RPL in ETH prior to the inflation round.
Let P2 be the price of RPL when the market clears.
Let N be the total number of minipools.
Let n be the amount of new entrants or exits.
Let S be an RPL amount spent by the DAOs.

Demand equals the RPL required by new entrants plus the RPL required by existing node operators in order to top up to 10%.
Demand = 1.6n / P2 + 1.6(N-n) (P1-P2) / (P1*P2)

Supply equals the amount of RPL supplied be exits plus the RPL spent by DAOs.
Supply = 1.6n/P1 + S

Setting demand equal to supply and solving for P2 gives:
P2 = 1.6N*P1 / (1.6N + S*P1)
Hence, if S=0 the price will not change. If S>0 there will be price inflation.

Consider what happens if RPL supply inflates by π and all of it is spent.
S = π * 1.6N / P1.
Solving for P2 gives
P2 = (1+π)^-1 * P1.

In the long term, we should expect price inflation to follow supply inflation, since eventually people will spend their unused RPL. Price inflation can temporarily lag behind if the RPL is not spent, but that means the RPL market cap grows, which it can’t do indefinitely. At some point people will cash out, given that RPL is only used for collateral and has no other use value.

The actual situation is worse than the above, since given RPL price inflation it is beneficial in the long run to let your collateral shrink towards 0. Hence, assumption (2) is unstable. It seems to me that to break this downward spiral we should at least remove the RPL payments to the DAOs. It would also be wise to remove RPL payments to NOs, since the transaction costs of repeatedly topping up could still make it a better long-run strategy to let RPL shrink towards 0 (in this scenario everyone would sell their RPL interest rather than use it to top up collateral, leading to inflation in the same way).

Maybe… governance should have non-zero value and scales with staked RPL (planned; obviously subject to change). Since we’re really talking about only the DAO spend needing to be offset, I think it wouldn’t take much perceived value to offset it.

Your concrete example was great, by the way - thanks :slight_smile:. Very convincing case that if the only value is keeping at 10% collateral then inflation will in fact make its way through like this.

In this scenario, people have decided it will have no value at some point. This shouldn’t be inflation, it should be a bank run. If everyone always sells at any price, it would become valueless overnight. A couple of things do limit here, including governance value and “humans are not game theory robots”.


In the end, I think we have (A) time before this problem shows up and (B) time to react when this problem has shown up. It’s a small leak, not a broken hull. As I mentioned in my “plausible options”, we could redirect some small amount of commission to RPL buyback (effectively aim to make S=0 overall in your nomenclature; or perhaps just a little above 0 if we see governance is giving perceived value), or we could stop RPL payouts and fund the DAOs directly from commission, or I’m sure there are plenty of other choices.