I thought more about this and changed my mind. Because the DAO has a long time horizon and voting for gauge weights is a public good that should be provided by the DAO, I believe we should invest in a significant amount of CRV under the following conditions:
- We are long Curve protocol.
- We lock CRV for 4 years.
- We make use of the booster.
- We can buy CRV without too much slippage.
First of all, buying and vote-locking CRV isn’t that different from just incentivizing our Curve pool with TRIBE. The CRV reward is basically an inflation, so the future stream of CRV rewards is mostly reflected in the current price of CRV. In particular, the “APY” shown on Curve UI is not really an accurate measure of the profitability of investing in Curve. In this sense, the choice to reward the liquidity providers with CRV rather than TRIBE is to a large extent a bet on Curve. That is why the first order of business is to agree that we are not bearish on Curve - that we don’t think Uniswap v3 will end up killing Curve, for example.
However, the benefit of investing in CRV can be higher for the DAO compared to retail investors. There are two reasons. First, CRV holders who vote-lock for 4 years are essentially being subsidized by holders who vote-lock for a shorter period of time or those who don’t lock at all, in return for taking the risk of locking capital. So if we are more willing to lock CRV for 4 years than the average retail investor, then the PCV should buy CRV from them. But we know that we will use a certain amount of the treasury/PCV to incentivize liquidity for a long time, whereas retail investors may suddenly need liquidity to pay bills. As an institutional investor, it can be easier for the DAO to take a long term view without worrying about liquidity.
Second, retail investors who vote to increase a pool’s weight are subsidizing other investors who provide liquidity in the same pool. This means they do not have much incentive to hold more veCRV than is necessary to boost their own liquidity. On the other hand, we want to subsidize everyone who invest in our pool, so voting with veCRV benefits the DAO more than an individual holder.
Since the DAO can make better use of CRV than retail investors, it makes sense for us to buy from them, provided that they have CRV to sell. But CRV is being continuously minted and rewarded to retail liquidity providers, so we should buy this stream.
If we buy CRV, we should also make full use of the booster. If our veCRV is more than what is needed to boost the liquidity provided from the PCV (if any), we should find a way to use the remainder to boost user rewards.
Finally, a cursory search on 1inch seems to show that there isn’t that much CRV liquidity out there, probably because most have been locked. I think we should gradually build our position. In fact, it makes sense to continuously buy CRV from our LPs to maintain voting power. Perhaps we can partner with a protocol like Pickle to set this up, along with a way to boost user rewards using our veCRV.
If we agree that we can substantively invest in CRV, I can do some more careful calculations to figure out how much investment translates to how much rewards for users. But to be clear, that calculation will have no bearing on whether we should invest in CRV, since the rewards are mostly coming from the principal of the investment via dilution. The reason why we should invest is because the DAO has a long time horizon and voting for gauge weights is a public good that should be provided by the DAO.
Back-of-the-envelope formula for how much reward goes to users:
Assume the value of Curve protocol stays constant.
39% of circulating CRV (260 mm) is locked. Assume these CRV remain locked and the rewards that accrue to these CRV will be locked, and no additional CRV will be locked.
1.73 bn CRV is distributed to liquidity providers. So far, 260 mm has already been distributed, so 1.47 bn CRV will be distributed to this 39% of CRV holders (or the users of their protocols).
At the end of distribution, these holders will have 1.47 bn + 260 mm = 1.73 bn.
But the protocol only keeps the initial 260 mm. So the protocol only has .26/1.73 = 15% of the principal left. The users get 85% of the principal.
This means if we invest $100, we will be giving $85 to users over the course of the next few years and retain $15.
But in fact, the sum will be greater than $100 because we buy the CRV at a discount from retail investors for the reasons that I mentioned above, and also because there are swap fees accruing to veCRV holders. So it’s 85% + a going to users.
I want to emphasize again that this calculation does not depend on the APYs shown on Curve UI.