Celo Reserve Collateralisation

@asa , following the recent governance discussion and mention of 10X collateralisation levels, I’m doing a deeper post on collaterisation and ideas for metrics we may feel are valuable to track.


  • I’ve calculated the current collateralisation of cUSD based on BTC/ETH lows in March 2020.
  • In such a scenario, cUSD is under-collatateralised by 65%+ when one excludes the value of cGLD.
  • I propose three collateralisation metrics to monitor: a) BTC/ETH collateralisation, b) stablecoin collateralisation c) cGLD collateralisation - and I discuss how minimum values might be set for each.

Note: I own and plan to hold cGLD for the long term and I own some cUSD in Valora.

Collateralisation of cUSD based on BTC/ETH lows in March 2020 & cGLD price in June 2020
Stable Value Assets (Feb 7 2021): cUSD: $30,198,575

BTC Reserve (Feb 7 2021); Price (Mar 13 2020); Value: 1,349; $3,858; $5,202,706
ETH Reserve (Feb 7 2021); Price (Mar 13 2020); Value: 23,247; $90; $2,092,232
DAI Reserve (Feb 7 2021); Price; Value: 2,095,536; $1; $2,095,536
Total Reserve ex. cGLD = $9,390,474

cGLD Reserve (Feb 7 2021) & Price (June 2020): 122,023,439; $1.50; $183,035,159
Total Reserve ex. cGLD = $192,425,632

Collateralisation ratio (excluding cGLD) = 0.31
Collateralisation ratio (including cGLD) = 6.37


  • In a downmarket scenario such as March 2020, cUSD stability is heavily reliant on cGLD value.
  • Were DAI to fail (may face collateralisation issues at March 2020 BTC/ETH prices), collateralisation ex-cGLD would further fall to 0.24 .

Three Potential Collateralisation Metrics to Monitor
Suggested numbers are illustrative

*a. BTC/ETH collateralisation.

  • I would suggest there be a minimum of 1X collateralisation with BTC/ETH, assuming BTC/ETH prices at their previous 3 year low.

*b Stablecoin collateralisation.

  • I would suggest 1X collateralisation with a basket of non celo stables.
  • I don’t know why DAI is included over USDC (I’m guessing people felt that USDC is not decentralised enough for the mission?)
  • DAI seems risky because many DAI that are minted may be underwater at BTC/ETH 3 yr historic lows. I need to learn more about whether DAI’s liquidation protocol would be fast/stable enough in a rapid price spiral.
  • USDC does seem like it would diversify risk, at the expensive of decentralisation.
  • Side note: Stables provide cheaper collateralisation than BTC/ETC (at least if you collateralise with BTC/ETH off of previous lows).
  • Sorry to make these comments a mess, but I’m not sure whether including stables for collateralisation makes good sense. If the goal is to get the best security for end users (like Valora), I think it could make sense. If there is a purist goal to build a better and independent platform for stables, then I would think collateralising purely with BTC/ETH is a better approach (which would be a variation on DAI).

*c. cGLD collateralisation:
*I’ve left this until last because I think it is hardest to come up with a good metric here
*I think cGLD provides an interesting stabilisation approach, but I’m not sure what attributes make it good for collateralisation. I’m guessing it would behave like BTC/ETH in a market crash?

  • I don’t think this is the right answer, but based on my limited thinking so far it seems to me that cGLD in reserves should be set based on stability considerations (i.e. cGLD <> cUSD) rather than collateralisation considerations.


  1. In the very long run, which do we expect to be larger: the market cap of cGLD OR the market cap of all cStables? why? Could someone point me to any articles on this question? I think this question might have implications for cStable collateralisation because obviously if we want (or think) cStables to have a higher market cap than cGLD in the long run, it doesn’t make sense to use cGLD for collateralisation. OR is cGLD the long term goal and cStables are just for a transition period? OR we don’t know and we’re trying to move quickly while building a secure platform?

  2. Does the Tobin tax help with collateralisation? How so? I could see it would prevent high frequency traders so maybe it helps stabilisation and also prevents arbitrage attacks. Is there somewhere I can read about how - if cGLD is plummeting in value - the Tobin tax would mitigate that quickly enough?

  3. I’m sure much of the above has been thought over many times before and written about. I would appreciate sharing of any articles/analysis I should read beyond docs.celo.org to understand the thinking to date around collateralisation.

And as an aside I love the Valora App and the 0.001 transaction fees I’ve been paying.

Many thanks for your post! I am very happy to chat about these topics.
In general, the reserve asset selection is done through governance by the community, and therefore you are raising relevant topics for the community forum. Any reserve assets under consideration should be freely traded and settled 24/7 on liquid markets, and should be based on an open-source protocol.
Open-source, so that all transactions with the reserve can happen on-chain, in a decentralized manner, and in a fully-auditable way. As you know currently most crypto assets out there are highly volatile. Most stablecoins out there are less volatile, but come with a significant amount of downside risk. In the case of fiat-backed stablecoins, this downside risk comes with unquantifiable counterparty risk. In order to evaluate the impact of the reserve portfolio on the stability mechanism, cLabs took a simulation based approach you can find the outcome in a paper here: https://celo.org/papers/stability
Now on to your questions:
In the above mentioned stability analysis you can find several downside scenarios (similar to the markets in 2018 or 2020, and even more extreme) which led to an undercollateralization of the stablecoin. The good news is that the stability mechanism still works. And the protocol has several ways to bolster the reserve in these times. Namely a fraction of the epoch rewards can be distributed to the reserve during times in which the reserve ratio is below a certain threshold. Also for every trade with the Celo reserve exchange (Mento), the reserve charges a small percentage fee.
I would love your take on the stability analysis, and am also happy to discuss these topics further. We are planning to update the paper, and therefore all this input is super interesting. I will connect with you on Discord!

Thank you Markus! and thanks for connecting on Discord. Yes, let’s chat.

My short take on the stability analysis is that it is contingent on the price model for cGLD. In practise, I think the price behaviour of cGLD could take on almost any form. One baseline guess is that it may move somewhat proportionately to the general crypto markets (BTC) in a crash - although it’s worth considering scenarios where it does worse.

I think an important case to model is where you have an exogenous drop in BTC+ETH to 2018 or 2020 levels and you have an exogenous and proportionate drop (or worse) in cGLD value. I would think this is the case to be sufficiently collateralised for. If the platform gets undercollateralised, I would ask whether any mechanism can reliably maintain a peg.

I could be wrong, but I think the stability paper only looks at cGLD as endogenous, whereas I think stability needs to be modelled with exogenous downside prices for cGLD. I think the only way we could look at endogenously modelling cGLD in a downside scenario is if there is some large third party buyer of last resort for cGLD, like a VC. Looking forward to chatting further.