Thanks @Tobi , @0xLeo , @markbarendt - much appreciated on the input.
A few points to add:
1. Collateralisation Levels
@Tobi , earlier this year we had a time where reserve collateralisation dropped close to 2X.
2. I don’t see a basis for arguments of increased complexity
We already have 5% as a stables target (currently DAI) in the reserve. Increasing to 20% doesn’t substantially change operational complexity (provided there is sufficient liquidity in stables, see new new note below).
There is a tendency on this forum to use complexity as a reason against proposals. Indeed it is a good reason, and we should seek ways to simplify CELO. That said, arguments of increased complexity should be laid out clearly with the new approach versus the baseline. I think @Tobi and @0xLeo you did that well on DAI versus USDC, but I have a different view of how to compare baseline and new risks.
3. Evaluation of Risk
I agree that increasing stables in the reserve increases the exposure of the reserve to depegging of those stables. However, it is a matter of balancing new risk against the current risks.
Current Risks
The current risk is that we have a reserve that is ~97% crypto and 3% stable. That is a very crypto weighted portfolio. I know of very few people or organisations with such a one-sided weighting of their assets.
Being so heavy in crypto subjects the reserve to heavy drawdowns in crypto market crashes. Mathematically, this is highly punitive to long-run performance (see Spitznagel’s volatility tax: https://www.universa.net/UniversaResearch_SafeHavenPart4_VolatilityTax.pdf).
Side note: Such arguments are the reasoning behind 60:40 equity-bond portfolios. The complementarity of asset movements reduces drawdowns, which strongly pulls up average returns. Dampening drawdowns is even more critical to the Celo reserve because we are collateralising cStables.
New Risks
Under this proposal, we are putting 20% of our reserve subject to stable depegging risks. Yes, that is a risk, but it applies only to 20% of our portfolio.
BTW, DAI and USDC survived all previous market crashes. cStables would almost certainly be undercollateralised if we had a crash like 2018/19 today. So, even if we aren’t covering the most extreme scenario, we are covering a lot more scenarios than we are today.
All in all, this proposal brings the benefit of meaningful (but not complete) crash resilience via moving the reserve from an extreme 97:3 crypto:stable weighting to a target 80:20 crypto:stable weighting - and this is at the cost of putting 20% of our reserve at depegging risk in extreme scenarios (scenarios in which, btw, today’s Celo reserve would anyway certainly become undercollateralised).
Collateralising stables with other stables
@Tobi , I think this is an important question and there is a degree of truth. Some notes:
a. We are not collateralising stables with stables 1:1. Rather, we are collateralising with an 80:20 portfolio of crypto:stable assets. cStables will vary as a portion of the reserve.
b. While this may reduce decentralisation at the margin, we have to be realistic that a much bigger lever for decentralisation is to start putting the reserve directly under the control of governance (rather than a multisig). I plan to post shortly on a proposal for that (as you know, and thanks for the great feedback). I think this point is not apparent to the broader community and is under-appreciated.
USDC vs DAI
Good points are raised on this matter too. @roman and I are revisiting the choice of USDC vs DAI and plan to post shortly. There are meaningful liquidity benefits to using USDC over DAI (which means the reserve can be rebalanced faster with less slippage). Further, since the stable target is limited to 20%, we limit the portion of the reserve that is subject to depeg or regulatory risk - whether it is USDC or DAI.
Coupling
Worth bringing this proposal back to a bigger picture of what we are trying to do at CELO and how a lot of these proposals are coupled. Two specific points:
- Thinking about decentralising the reserve to governance is critical (proposal upcoming). If we were more decentralised with that, then regulatory issues (e.g. in considering DAI vs USDC here) would reduce.
- Over the next year we plan to bring forward many proposals such as a) expanding and/or launching cStables, b) finding credibly sustainable ways to give cStables more economic (and possibly voting) power. Right now the reserve is so crypto heavy (97-3 in crypto-stables) that in most new proposals, the forum asks the question: “But what if this proposal increases risks to cStable collateralisation?”. This will always be a good question to ask but, right now, by not having better stable collateralisation, we are slowing many of our other proposals. Adding a greater stable proportion to the reserve isn’t a complete solution, but I see it as a valuable step as we then move to ii) decentralise the reserve to government, and iii) add other forms of capital - especially natural capital - to the reserve asap.