- The proposal increases reserve holdings of stablecoins, specifically DAI, in the short-to-medium term, and then phases stablecoins out as natural backed capital is introduced.
- This proposal increases the percentage of stable collateral in the reserve to better protect against cStable under-collateralization in the case of a crypto market crash.
- In the short term, this proposal has the effect of increasing the reserve target for stablecoins to 19.5%.
- In the medium to long term this proposal allows for stablecoin holdings to be reduced, and even eliminated, as natural backed capital is introduced in sufficient quantities
As the CELO community grows, there are increasing numbers of dApps, merchants, creators and community members making use of cStables. It is critical for these users that cStables retain their value relative to their fiat currency pegs. This is particularly true for community members for which the underlying collateralisation and security mechanisms of Celo are not apparent.
All dApps and communities built on Celo have an interest in the strength of and trust in cStables.
As of April 14th 2022, Celo has approximately $120M in issued cStables (CeloReserve.org). This is currently collateralised by a reserve value of about $580 million of which:
- 97.2% are volatile crypto assets (Celo, Bitcoin or Ether)
- 2.8% are natural backed capital (MCO2) or stablecoins (DAI).
The price of volatile crypto assets including Celo, Bitcoin and Ether can swing wildly and this puts collateralisation of the reserve at risk. For example, were crypto prices to revert to the levels of the 2018 crypto winter, cStables could become undercollateralised.
In the case of a crypto market crash, there is a mechanism to slow undercollateralisation of the reserve and depegging of cStables. This is the “Tobin Tax”, whereby a transaction tax would be charged for swapping cStables for CELO. However, the Tobin Tax is a) currently turned off (set to zero) and b) it is unclear whether - in practice - this tax could slow reserve depletion or instead hamper liquidity in a detrimental way.
This proposal takes a more straightforward approach of increasing the amount of stable capital in the reserve.
Quite simply, in a crypto market crash the value of the reserve may plummet (with Bitcoin, Ether and Celo falling in price), but provided that the price of stablecoins in the reserve (or later natural capital, once introduced in larger amounts) remains stable, reserve overcollateralization is better preserved.
Stable-value assets (currently DAI): 5%
Stable-value assets (specifically DAI): 20% minus the target percentages of natural backed capital for the reserve
Note: With cMCO2 at a 0.5% target level in the reserve, this would put the short term stable-value assets target at 19.5%
The numbers in this model should be considered as illustrative of just one possible scenario. By nature, reserve asset price movements are coupled in complex ways.
This spreadsheet provides a model for the CELO reserve i) today and ii) after this CGP assuming:
- cStables outstanding remain constant in a crypto market crash.
- I vary the parameter x, which is the % fall in BTC, ETH and CELO prices in a crypto market crash.
Crudely, this chart shows that (following the assumptions above):
- Today, the reserve could withstand about an 80% price drop in crypto (not an unprecedented fall, considering historical crypto data).
- After the CGP, the reserve could withstand up to a 95% price drop in crypto.
The chart above likely overstates the security of the reserve, mostly because CELO is less established than BTC and ETH, so CELO is likely to fall further in a crash than BTC or ETH.
There are a number of factors to balance in choosing a stable-value asset:
- Liquidity - can the stable-value asset be easily exchanged for other reserve assets with low slippage.
- De-pegging risks
- Regulatory risks - to what degree is it possible that funds could be confiscated or rendered immobile.
It is important to note that one bad scenario for the CELO reserve would be if the 20% stable-value allocation were to fully de-peg and be rendered worthless. However, there would still be the other 80% of the reserve. It is not just a matter of avoiding depegging risk, but balancing this risk against the overall risk of the reserve becoming undercollateralised, for example, the risk of undercollateralisation in a broad crypto market crash.
The choice of percentage is a judgement that aims to balance these two opposing risks (depegging of reserve stablecoins; undercollateralisation of the reserve to to a crypto market crash).
On balance, we recommend that the regulatory risks of USDC outweigh the liquidity benefits, and therefore recommend the choice of DAI. Further, DAI is already a CELO reserve asset and does not involve introducing a new asset.
Liquidity - Dai liquidity would most likely be obtained via ETH/DAI or WETH/DAI pairs. Pairs involving USDT (Tether) would be avoided in any trades owning to a lack of transparency around the collateralisation of Tether. Daily volumes on DAI pairs are as follows:
Depegging risks - DAI is an overcollateralised stablecoin (data here):
- There is more than $1 worth of crypto collateral for each $1 of DAI
- Notably, about half of DAI is collateralised by USDC, while the other half is collateralised mostly by ETH and WBTC type assets.
- MakerDAO (the protocol behind DAI) has an automated liquidation mechanism that results in collateral being sold if there is a drop in crypto market prices and collateral ratios drop below set limits. This liquidation mechanism can, and has in the past, failed, but the system has been around for multiple years and one might argue is reasonably tested.
- DAI is, besides the fact that it is backed to large extent by USDC, relatively decentralised and there is no central entity that custodies DAI. Collateral is kept in decentralised smart contracts.
- Possibly the largest regulatory risk to DAI is if USDC were to be suddenly withdrawn from the market by its issuers. As mentioned about 50% of DAI comes from USDC.
Liquidity - USDC liquidity would most likely be obtained via ETH/USDC, BTC/USDC or WETH/USDC pairs. Daily volumes on such pairs are as follows:
Depegging risks - USDC is a 1:1 fiat-collateralised stablecoin run by Circle:
- Collateral is audited and published online.
- USDC is custodial, i.e. collateral is held by centralised agents. The reputation and performance of these agents (Circle) influences depegging risk.
- Were regulators to undermine the current approach taken by Circle - especially in a manner that is sudden or unexpected - it is possible that USDC could de-peg. For example, collateral backing USDC could be frozen in bank accounts, reducing its value as collateral for USDC tokens.
- Were USDC to be rendered illegal, wallet addresses with USDC could be blacklisted by public and private organisations. This is a risk for any token, but perhaps especially fiat-pegged stable tokens because these most directly influence a government’s control of their own currency.
This proposal specifies that stablecoins as collateral should be targeted only to the extent there is insufficient natural capital in the reserve to meet 20% of the reserve’s value
As governance proposals are introduced to increase natural capital in the reserve, the target for stablecoins is automatically reduced.
For clarity, the challenge of finding high quality (liquidity and stability) natural backed capital is not an easy one. It will require significant energy from the community to implement safely for the reserve.
It is fundamentally a hard problem to provide safe collateralisation of one asset (here cStables) with other uncorrelated assets (here BTC, ETH and CELO).
MakerDAO (issuers of DAI) do this by invoking an overcollateralised approach plus a liquidation mechanism to provide protection of the peg during a crash in crypto prices. The investment and testing of such a liquidation mechanism is significant and not trivial to safely implement.
To date, CELO has in essence adopted an overcollateralised approach as well, but without a high-confidence mechanism to protect against undercollateralisation in a steep market crash. Thus, this proposal aims - in the short-to-medium term - to protect against this risk by adding stable-value assets to the reserve, i.e. assets that have a close correlation in value with the cStables issued by the Celo reserve.
All in all, DAI probably presents less regulatory risks than USDC (although DAI itself is 50% subject to USDC risks), whereas USDC provides easier reserve management owing to higher liquidity, plus more straightforward collateralisation, which likely reduces the risk of depegging. Either way, it will be important to move quickly to increase natural capital in the reserve (providing price movements with low correlation to BTC and ETH), in which case - under this proposal - stable-value assets would be automatically phased out.