Green assets as collaterals for stable coins

Hello everyone,

Collateralised stable coins need to be overcollateralized with quality, stable and liquid assets. In the case of Celo’s stable-coins (cStables), there is a desire to include nature-based assets in the reserve backing up cStables. This post discusses a range of green assets (which include nature based assets) according to their properties: impact, liquidity and yield.

Green assets’ trilemma - Choose 2 out of 3: impact, liquidity or yield

Impact against climate change: is the impact high and transparently measured? A rainforest offers both a high impact against climate change and an ease of measurement (e.g via satellite or drone survey). On the other hand, the impact of a themed bond issued by a corporation is arguably lower and more difficult to measure.

Liquidity: Is there a liquid secondary market whereby the assets can be sold without affecting its price? A share in a listed renewable energy company (yieldcos) is more liquid than a rainforest in the example above.

Yield: Is the asset offering yield? This is financial yield as the impact referred to above can be considered to be environmental yield. A green bond issued by a corporation offers tangible financial yield compared to a nature-based asset such as peat land.

Based on empirical evidence (see more in the table below) there are trade-offs among impact, liquidity and yield. There are no assets that can fit all three criteria i.e. high impact, high liquidity and also high yield. For this reason, there is a concept called greenium within the green bond market - given the impact that green bonds supposedly bring, investors have been willing to accept lower returns (yield) in exchange for impact and liquidity.

To be included in the cStables reserve assets will first need to be tokenized. The process of tokenizing a green or a nature-based, real-world asset requires not just creating a token but also a legal representation of the asset. This is something that real world asset tokenisation protocols such as Untangled focus on.


The following green assets are classified according to their corresponding properties. These are based on empirical observations and not theoretical possibilities. They are also investable assets that could, in principle, form part of cStables reserve. The list is not exhaustive so the community can add more.

ReFi/Green Assets Impact Liquidity Yield
Solar and wind energy “yieldcos” H H L
Operating renewable energy projects developed markets H M L
Renewable energy construction loans in developing markets H L H
Green receivables such as home solar systems H L H
Natural assets such as REDD+ natural schemes H L L
Green supply chain receivables M M M
Green bonds H M L
Carbon Credits H/? H/M M

The above assets can be classified further by their functions or instruments:

Function: in terms of CO2 emission, there are nature based assets such as forests or regenerative agriculture land that are good at CO2 sequestration whilst renewable energy assets such as hydro, wind or solar are good at preventing further CO2 emission to the atmosphere. We believe that cStables should be backed by both of these functional assets.

Instruments: The above functional assets can back different instruments such as private credits, listed bonds or equities in renewable energy companies. Depending on the instrument, the impact of the assets might be more difficult to verify. For example, a debt instrument issued by a SPV whose assets are solar home systems in Africa could be easier to verify in terms of impact compared to, say, an ‘use of proceed’ bond issued by a large corporation where there is a lack of transparency on the impact measurement.

Carbon credits are a special asset whereby the impact is highly dependent on the quality of the underlying project. It is arguably more liquid as its trading is more of a ledger entry rather than a complex change of legal ownership often associated with other nature-based assets. Its yield or gain is determined by supply and demand, other things being equal. In the current environment, supply tends to be lower than demand creating an upward pressure on prices. Since there is no such thing as a unified carbon credit, there is no single price for it. This creates a further challenge to impact or quality verification for carbon credits. Further, if carbon credits are meant to be retired and not speculated then an increase in prices might not be desired especially from a perspective of a buyer who needs to offset their CO2 emissions.


Onboarding real-world assets to cStables reserve introduces a new risk dimension: credit/counterparty risk. Not all assets will be repaid when due or it is not easy to liquidate them given a lack of a secondary market. Tokenization in itself does not significantly change the liquidity of an asset.

That’s said, some assets are self-liquidated. For example, green supply chain receivables hived off in an SPV could be both short term (e.g. more liquid) and self liquidating. When a receivable such as an invoice becomes due, the invoice receiver/buyer usually pays. In the context of a supply chain it is important to keep your suppliers happy by paying on time so there is no disruption to your business.

For other types of assets, liquidation is necessary when there is a trigger or a breach of covenants. It is important therefore to track the assets beyond the point of tokenization. Having reliable oracles (centralised or decentralised) is crucial in ongoing net asset value calculations (NAV calcs). When the assets are to be liquidated they often follow standard procedures within existing securitisation structure e.g. The Trustee instructs a predetermined liquidation agent to sell the assets and return the proceeds to the reserve or token holders.

Since a DAO may not have a legal status, formalising off chain real-world legal agreements represents unique challenges. Legal structures could be set up to facilitate this. Celo could also learn from other DAOs such as MakerDao.

Asset allocation strategy for cStables reserve

Understanding the tradeoffs among the properties of green/impact assets is important in reserve asset allocation strategy. Unlike crypto native assets, these real-world assets are not correlated with the rest of crypto but they are subject to different types of risks such as liquidity and credit risks. Extensive stress testing and backtesting are needed to construct an optimal basket of collateral assets backing up cStables.

Onboarding assets, particularly unlisted assets to the reserve, requires strict procedures. We are therefore welcome onboarding policy discussion as in the article.


Hi Muntangled

First, This is a great framework.

When you say “we” which is the entity you represent? is it the ClimateCollective?

1 Like

Hi @aaronmgdr,

Thanks for your comment. We are Untangled Protocol - decentralized lending and liquidity protocol for real world asset collaterals. Please see our intro post here Untangled Finance.

We work with the ClimateCollective team on various fronts.


Oh hence your username

1 Like

Cross linking to a relevant discussion.

The stuff in the high liquidity asset class would make nice candidates.

One thing to think about though is that the choice of reserve assets is and should be nearly pure economic decisions. If an asset meets the economic criteria great, if not there should be no shoehorning.

What I’m getting at here is that the reserve money can’t be considered as idle capital that can be allocated for mixed priorities.

The other thought here is that if Celo were to use fiat as the reserve currency, the reserve size required drops a bunch. Most of the ‘money’ current held as over-collateralization could be reallocated.



The stuff in the high liquidity asset class would make nice candidates.

But if you look at the table most high-impact assets are low liquidity. The earlier discussion made a case for forest and regenerative agriculture land tokens. Whilst we would advocate their inclusion in the reserve to an extent we would tread carefully there.

I spent 7 years at the front line of green field agriculture project development in Africa. By front line I meant we were surveying lands in some of the most remote areas in West African countries from Nigeria to Senegal and every countries in between. We therefore can say that land-based investments in these high impact areas are extremely challenging. The land tenure is mixed between state, local authority and tribal systems. Securing legal titles is challenging but even after title has been secured (and tokenized) there is no guarantee that the land will be used or preserved for intended purposes. The local population surrounding the land/forest need alternative livelihood such as intensive farming, eco tourism or they will resort back to old ways of living. This is our experience in Africa. Other areas of the globe might be different.

So we advocate a balanced asset allocation strategy that takes into account the inherent trade-offs, after extensive analysis and screening. For nature based assets we think it is easier for Celo to partner with NGOs or forest trusts who have many years of experience securing preservation areas, working and learning with local communities. Once this part is sorted, tokenisation is a much simpler exercise.


Morning @untangled-finance

I believe in your mission and the motivation, but the reserve isn’t idle capital.

The reserve is essentially ‘money that is held in trust’ to protect the value of and provide liquidity for the Celo stable coins. Each cUSD, cEUR, cREAL coin is essentially a debt that Celo owes to the holders of those coins. The reason for over-collateralization is that all crypto assets are risky, even stable coins.

The holders of Celo stable coins can demand payment of that debt at any moment and the reserve needs to be able to liquidate every single coin in short order if needed to pay all those debts.

Now, there are ways to make the reserve less risky, which in turn could reduce the need for over-collateralization, and that could allow for money to be reallocated out the reserve and put to use elsewhere.

Reducing the reserves risks is the key to being able to use that money for less liquid projects. That faces some real regulatory and philosophical challenges.

Here’s one place where that is being talked about, there are other places too.

1 Like

We agree that liquidity is a desirable property for reserve assets @markbarendt. To the extent that trade offs have to be made to include some illiquid but impactful assets they need to be made clear for the Celo community to vote on. Like a bank Celo needs to develop capabilities for asset liability management - ALM (Liabilities = cStables, Assets = Reserves). All collateralised stablecoins need to be good at this.

1 Like

Thank you for posting this thoughtful summary - very useful! It’s also great to see that you mention the legal challenges around tokenizing RWAs, looking forward to learning more about your approach with Untangled.

Regarding the liquidity-impact tradeoff, I spent some time thinking whether a tiered liquidity setup would help. On a high level, it could work something like this:

  • We (logically) split the reserve into a core reserve and an impact reserve
  • The core reserve holds highly liquid assets (e.g. the current crypto assets) and has sufficient collateral to cover all outstanding cStables.
  • Let’s define two collateralization thresholds for the core reserve x and X with x < X, let’s say x = 1.5C and X = 2C where C = USD value of stables in circulation, i.e. we aim for 150%-200% overcollateralization with the core reserve (numbers are just examples, TBD)
  • If the core reserve value R exceeds X, the difference R-X is ‘transferred’ to the impact reserve and invested e.g. in real world assets
  • As long as x < R < X, the impact reserve is not touched
  • if R falls below x (trigger event), impact reserve assets are liquidated to boost the core reserve

Would love to hear your thoughts on this


This actually seems like a good first step towards a multi-collateral Mento design. Collaterals can be batched into different reserve buckets based off of risk scoring with interactions between buckets. Something to think about.


Good morning @Slobodan

So with regard to the collateral percentage there are more considerations than just a generic percentage of the total.

It’s time for a reality check.

One of the things that has come into focus because of recent events and discussions is that the Celo assets in the reserve poses the same risks to this protocol that Luna posed for Terra. In a de-pegging situation the value of Celo could go to essentially zero.

So, it’s the non-Celo assets that actually act as the reserve that could repay the coin holders.

Given that risk I have suggested in another discussion that the Celo assets not be counted as part of what we consider the reserve that is backing the stable coins.

At this minute the total reserve value is $268,632,850
The Celo portion value is $104,220,663

That means the non-Celo reserve is $164,412,187
The outstanding supply of stable coins is $100,592,110

The current ratio of non-Celo assets to outstanding supply is at 163%

That’s not the end of the assessment though.

In the current market conditions it is not unreasonable to fully expect BTC and ETH to drop considerably. Estimates for BTC during this bear cycle suggest that we could see prices for it in the neighborhood of $20,000, so roughly a one third drop from today. It is reasonable to expect ETH to drop more than that but for today I’ll keep the math simple.

Non-Celo $164,412,187
Minus 1/3 $54,804,062 to consider value at a reasonable probability of market price in the near future
Net probable value of non-Celo assets available to cover stable coin debt $109,608,124

The realistic colateralization percentage after correcting for market conditions and real risks is roughly 109%.

So in reality, Celo isn’t over collateralized.

1 Like

Some supporting info regarding the risks of using/considering the native coin as collateral.

A report released Thursday by Terra backers Jump Crypto, the crypto arm of investment firm Jump Capital, shows based on recorded blockchain transactions that big investors with several million UST each liquidated their stocks early in the depegging process, driving the price ever downward and leaving smaller holders with practically nothing.

1 Like

More on liquidity and liquidation

We also think the x<R<X framework is a good starting point. It has been tried and tested with protocols such as Maker. We call X collateral ratio and x liquidation ratio. By definition x should be more than 100% as we would need to provide some incentive, essentially a discount to ‘fair value’ of underlying assets, to enable a quick sale during a liquidation event.

If cStables are collateralised by crypto natives then there is plenty of data within the Celo network and other similar protocols to model the value of x, X. The UST debacle reminds us of Black Swan events even though, strictly speaking, it was an undercollateralized (algorithmic) stablecoin that was built on an unsustainable foundation.

if R falls below x (trigger event), impact reserve assets are liquidated to boost the core reserve

As R fluctuates with underlying crypto collateral and the value of cStables in circulation, boosting R through liquidating impact (RWA) assets could be problematic. Typically RWAs can’t be liquidated quickly but celo could employ a combination of strategies:

  1. Creating a safety margin: say liquidity need for a time horizon of 1 week, when the most liquid RWA can be liquidated
  2. Creating a liquidity strategy within the RWA portfolios: essentially making conscious tradeoffs among the impact/liquidity and yield
  3. Working with real world asset funding protocols to create highly liquid exposures to impact assets

1 above could be implemented with the value of the Celo token in the treasury. However it is much more sustainable to use protocol profits or surplus to fund safety margins. For this reason, celo needs to consider assets yields when making asset allocation decisions.

2 above has been covered in this post. Again all three properties and their tradeoffs should be considered.

3 above starts with selecting quality assets to form the impact reserve. The asset onboarding process needs to follow good practices as being proposed in this thread. It is important to note that once the asset has been onboarded the on-going monitoring is robust enough to ensure all asset risk parameters are in check. Having real/near real-time, on chain asset tracking, net value calculation and reliable oracles is crucial.

Even AAA assets have a non-zero probability of default. A proposer of impact assets (to be included in the reserve) needs to have a pre-approved liquidation process. Typically, these assets are sold under ‘true sale’ to a Special Purpose Vehicle (SPV) whereby an independent administrator or trustee is instructed to liquidate the assets (perhaps with the help of a professional liquidator) when certain conditions are met. This sort of liquidation procedure is well functioned within traditional finance/securitization. The challenge here is to create a legal structure that can work with a DAO as a DAO may not have a legal status.

Legally speaking, impact asset tokens come in two main forms:

  • token representing a claim on the actual impact asset
  • token representing a claim on the proceed of the impact asset in the event of a liquidation

Both of these are securities: equity in the former and debt in the latter. An equity token might be subject to registration with various authorities before being included in the reserve. Price discovery would need to happen on a licensed secondary market, and in the US, subject to various transfer restrictions. There is also an issue regarding custody of the token, establishing a control location for sovereignty over the custodied token and compliant liquidation mechanism.

Under the second form, assets would be sold to an SPV in the manner described above. Each of the assets are tokenized as an NFT with adequate metadata to enable oracles to price assets on an ongoing basis and in the event of a liquidation. This second form is arguably simpler as it involves less parties, at least in the US. The liquidation process is still manual as described above.

Solving for the trilemma of impact tokens

Both of the above are securities so unless they are registered or exempted and listed there are no secondary markets in DeFi. What if there is a token with exposure to underlying impact assets but are not subject to securities’ restrictions? This token would then be traded freely on DEXes such as Uniswap.

This could be done by separating between Liquidity Pool and Asset Pools. Under this construct, liquidity providers would deposit their funds into Liquidity Pool and receive an LP token in return. A permissioned asset originator would sell/upload their asset NFTs to an Asset Pool (a smart contract) to which Liquidity Pool automatically allocates funds under certain conditions.

Liquidity providers can timelock their LP tokens to receive extra allocation of LP tokens and a governance token. The payment waterfall of Liquidity Pool would follow the timelocking of LP tokens e.g., no time lock/immediate, maturing in 1 month, 3 months, 6 months etc. With a suitable design, these time locked LP tokens can follow the ERC20 format, making them composable with the rest of DeFi. A secondary market for these tokens e.g. pairs of LP token/timelocked LP tokens could be enabled.

Neither the Liquidity Pool nor LP token holders are lenders of record. The Liquidity Pool participates in the economics of Asset Pools by virtue of the protocol design (i.e. the smart contract). This would be a leveraged and structured product and in normal financial services the dealer managing this would be regulated. But if the entire mechanism is automated and controls are at the end points - at the originator validation/auditor node and investor onboarding nodes - the internal mechanism can be completely DeFi.

If cStables reserve holds the above LP tokens, not only it is exposed to underlying impact Asset Pools, it could quickly sell the LP token on any DEX when needed. If the LP token also provides yield this could be a solution to the impact asset trilemma.


So for RWA collateral, how do I flash loan stuff and get that liquidated in a single transaction onchain? Or does that liquidation process require someone to hold principal risk?

For RWA collaterals is there also oracle risk and liquidators need to have their own assessment of the true value of the underlying?

Cus as soon as you introduce principal risk and the oracle pricing risk into the equation, you will need to add in some huge buffers / liquidation bonus.

It also severely restricts the number of people who will even bother to become liquidators.


There are two types of liquidation:

-liquidating asset tokens takes time and can only be triggered in certain circumstances.
-liquidating Liquidity Pool LP tokens are subject to prevailing secondary market conditions and are unrelated to the asset tokens above.

Hence why the construct that separates the 2 types of tokens while allows users to take a view on the ‘price’ of liquidity and credit risks through secondary market trading.

Oracle risk could be minimised through ‘market to model’ or a transparent asset pricing mechanism. Many RWA assets have predictable cash flows and observable discount rate. These can be inputs into on chain net asset value calculation.

As to principal risk, this can be managed through off chain legal agreement with reputable counterparties. Note that here we refer to liquidator of asset tokens.

1 Like

Do you have models/spreadsheets that demonstrate the concepts you are talking about so that the general population of Celo users can understand what’s happening.

I feel like these RWA liquidations will become a exclusive club and not really accessible to ordinary users.

Mark to market is also tricky cus what if I don’t even trust the financial data reported by the auditors? (Even the big four accounting firms have been caught fluffing up their reports)

Not against this idea and there are a few crypto startups doing this, just not a very open and transparent game to play.


Type 1 liquidation does require professional liquidators as here we are talking about distressed asset liquidation. That’s said, depending on asset classes, there are many participants in the distressed asset market within TradFi. Also as we mentioned above, the type of liquidation should not happen too frequently.

Type 2 liquidation, under the above construct, is accessible to any ordinary user - here we are talking about selling ERC20 tokens on DEXes. Also this type of liquidation or sale will happen more frequently.

Re mark-to-market/mark-to-model - Agree, any system relying on human actions might be subject to abuse. The only way for Celo to completely alleviate this risk is not to be involved with RWAs. That means no exposure to impact assets whatsoever but isn’t this against Celo’s own differentiation? a layer one at the forefront of ReFi?

RWAs bring counterparty risks, no doubt about that. This risk is however can be minimised with staking or dealing with reputable counterparties. An increase in counterparty risk is however compensated for by a decrease in market risk, compared with crypto native assets. RWAs are more stable and arguably less correlated with the rest of crypto. They are many orders of magnitude bigger than the crypto market. With what happened to crypto market in recent weeks, one does wonder what use cryptos actually bring to humanity?

You, the Celo community and we are not against this idea and this sort of discussion is helpful for us in making conscious choice in spearheading ReFi.

1 Like


If an arrangement can be made for one type of financial asset, then there is no reason it couldn’t be made for another. The legal challenges would seem to be exactly the same.

That makes fiat a viable option for the reserve.

1 Like

That seems to be a significant point of centralization.

Is there a way that the accounting can be on chain?