Yield / Rewards on Reserve Collateral

The Mento Reserve consists of a diversified portfolio of different on-chain assets supporting the ability of the Mento Protocol to expand and contract the supply of Mento stable assets, in-line with user demand. The highest goal of the reserve is the stability of the outstanding stable assets. Currently (as of Nov. 15, 2023) the reserve overcollateralizes the stablecoins by a factor of ~3.3 (i.e. the value of the reserve in USD is approx. 3.3x of the value of all outstanding stable assets in USD) with some of these reserve assets to be returned to the Celo Community Fund following CGP102.

So far, none of the assets in the Mento Reserve have ever earned a yield and the reserve was focused on holding highly liquid assets (BTC and ETH) next to CELO and other stablecoins like USDC and DAI. This was very conservative and not earning yield on these assets has protected reserve collateral and therefore Mento stablecoin holders from the wide range of hacks, bugs and rug-pulls across Web3 over the last years.

Yield and rewards on on-chain assets are a developing field with more solutions becoming available on a weekly basis and veteran solutions becoming more battle tested. It therefore seems worth revisiting the topic of yield on reserve collateral. Generating yield and rewards on a part of the reserve assets could help grow out reserve overcollateralization further and/or, in the future, could be used to incentivize beneficial behavior on the Mento protocol. It would however come with some additional risk (in addition to the principal risk of the asset), mainly platform / counterparty risk and liquidity risk due to unlocking periods.

Weighing these arguments against each other, not earning any yield with reserve collateral feels over-conservative to me at this point. The Web3 space has progressed and there are now some yield and reward solutions that appear to provide a reasonable risk/reward tradeoff. I would therefore like to propose a first allocation of reserve assets into yield bearing assets.

Solutions that require a KYCed counterparty:

There is a wide range of options available and several teams have showcased their solutions to Celo and Mento community members over the last few months. Many of these solutions require a KYCed counterparty before engaging is possible. Some notable solutions (in no particular order) are for example:

  • T-Bill type yields:
    • Mountain Protocol
    • OpenEden
    • USDLR by Stable (withstable.com)
    • Matrix Port
    • OUSD by Ondo
    • tfBill by TrueFi
  • Tokenized bonds:

Mento aims to be a fully decentralized protocol which makes it non-ideal that these solutions require a KYCed counterparty, i.e. a real-world entity acting on behalf of the protocol. Additionally, these more centralized solutions are not easy to analyze in terms of their risk/reward tradeoff and proper due diligence, for example, around the entity structure and counterparty risk involved, is challenging. Having these solutions deployed natively on Celo in a way that they can be interacted with in a decentralized fashion would give additional options for a fully decentralized Mento.

Solutions that do NOT require a KYCed counterparty:

The decentralized solutions are more easily automated going forward, as everything can theoretically be triggered directly from smart contracts. They are also a bit easier to assess as with them, most of the action happens on-chain. Since there are no KYC requirements, the setup / operational overhead is low.

The following yield solutions are more decentralized in nature and do not require a KYCed counterparty:

  • Staking DAI to receive sDAI (i.e. earning the DAI Savings Rate via Spark Protocol)
  • Staking ETH via LIDO
  • Staking agEUR to receive stEUR (Angle Protocol)

To get started with reserve collateral earning a yield, I would propose to start with these non-KYC requiring solutions. More precisely, I propose to:

  • lock all of the reserve DAI in sDAI (effectively switching out DAI for sDAI in the reserve target allocation),
  • lock the reserve ETH that are currently not committed to liquidity pools via CGP121 in LIDO and
  • allocate a quarter of the EUROC holdings into Angle Protocol’s stEUR.

100% of all the rewards earned shall flow back into the reserve as additional stablecoin collateral.

It is important to account for the unlocking periods of the chosen solutions to avoid liquidity crunches. sDAI can be unlocked immediately 24/7. Same for stEUR. Unlocking ETH with LIDO takes “anywhere between 1-5 days under normal circumstances”, i.e. if there are no unusually large withdrawal requests made via LIDO. With this combination, liquidity crunches should be very unlikely.

Please let me know how you feel about this suggested direction. I am also looking forward to discussing this and more in the upcoming Mento Community Call on November 22nd, 6pm CET.




excited for this proposal! :raised_hands:

I wonder if a vault strategy (ie. ERC-4626) could simplify operations & help diversify the yield sources (eg. ETH could be staked across multiple providers similar to Yearn yETH implementation).

1 Like

What if rewards were used instead to purchase CELO?

Building a collateral base is a good first option, but strengthening CELO seems like a priority for the ecosystem given its massive drop from highs + high concentration in the reserve (would also position it to be more competitive with ARB, OP, other L2s…)

[I’m just brainstorming out loud here, once you have yield there’s a lot of potential things that can be done with it]

Disclaimer: I hold CELO


Happy to chat mare about this in the community call but my tldr would be: The core mandate of the Mento Reserve is to protect stablecoin holders and I believe that’s what the focus should be initially when thinking about yield income - hence have yield flow back into reserve collateral. The Mento Reserve has been a massive net buyer of CELO since inception in any case.

1 Like

I tend to avoid calls, so I’ll offer a quick rebuttal here – I’m not convinced by your argument:

The core mandate of the Mento Reserve is to protect stablecoin holders

Strengthening the demand factors of the largest collateral in the Mento Reserve – CELO – protects these holders. Above the threshold of high quality liquid collateral that already exists in the reserve – i.e. a minimum 100% backing of all cStables outstanding by other stables or blue chips, as is policy today – one could argue overprotection is in place. Maintaining this floor makes sense, but above that floor, protection is economically inefficient – there’s better things that can be done with these profits (e.g. Mento is today considering a governance token – why not evaluate different types of patronage earning mechanisms?)

The Mento Reserve has been a massive net buyer of CELO since inception in any case.

To my understanding, this isn’t really the case. Mento pulls CELO out of circulating supply to a degree, but it’s not “buying” assets; that collateral is still available for price-balancing arbitrage (folks can still trade cStables against CELO in the reserve). Again: to my understanding, existing CELO in the reserve is more like a liquidity pool than anything else.

If anything, it seems that Mento has been a large net seller of CELO ever since the rebalancing, and one could make the argument that pocketing excess yields as collateral above the threshold needed to be adequately “protected” means that the Reserve essentially is profiting from these previous trades vis-a-vis CELO holders. This being said, in order to understand this completely, it’d be helpful to understand the full history of the Mento Reserve and the winners/losers of its rebalancing exercises.

Sure, happy to discuss here instead.

I would argue it protects them less than additional outside collateral. But agreed that CELO is an important asset in the Mento Reserve setup so definitely worth considering what can be done to strengthen it.

It’s certainly a tradeoff and overprotection is not great either but with, for example, USDC depegging to 96 Cents in March after the SVB default, I would argue that being overprotective and economically inefficients starts quite a bit above 100%.

The Mento Reserve has received 120M CELO in the genesis block, has returned 25M to the community fund already and still holds 117.5M CELO right now. So it looks to me like it has bought significantly more CELO off the market than it sold into the market.

1 Like

Hey all,

as discussed, here are the details for the Mento Community Governance Call tomorrow where we will discuss discuss this topic as well.
:spiral_calendar: November 22
:watch: 6pm CET
:point_right: Join here: http://meet.google.com/ktw-iuvp-ddp

Current agenda items:

  • Mento-Specific Governance Token
  • Yield / Rewards on Reserve Collateral

Please share in “# general” on Mento Discord if you would like to add other topics to the agenda!

1 Like

I think this is great analysis and a really interesting discussion @roman. I wasn’t able to join the call last week due to the holidays, but I’d like to provide an alternative view point here that I think is very relevant.

As a note, I work for Centrifuge, and our team has recently received a grant from Celo to deploy Centrifuge on the Celo network.

I’d like to share the perspective of Centrifuge Prime, a product we developed to solve exactly the problems you mentioned in your post. In simple terms, Centrifuge Prime is a managed services offering that allows DAOs and protocols to setup a compliant and reliable infrastructure for investing in Real World Assets. This includes an out-of-the-box legal framework, risk-analysis from the Centrifuge Credit Group, and dedicated resources facilitating a diversified RWA portfolio. Today, Centrifuge Prime is onboarding the Aave DAO and treasury, is built based on our long-standing experience in MakerDAO, and is actively in engagements with the largest DAOs in the space, such as Arbitrum and others.

First, the premise of allocating the Mento reserve into yield-bearing assets is a critical need. Your approach in identifying crypto native assets as well as real world assets together is a step in the right direction. A healthy portfolio should include yields that provide broad exposure to uncorrelated asset classes, varying degrees of risk profiles (within Mento’s limits), and ultimately balanced against the protocols higher-level organizational objectives and constraints.

However, we think much more can be said on the challenges you’ve identified with RWA.

We strongly believe that KYC/AML shouldn’t be a blocker for a decentralized protocol. The Centrifuge Prime legal framework is a battle tested means of safely solving this challenge.

In a nutshell, the legal framework sets up a conduit structure, integrated with DAO governance, that provides a tool for the DAO to comply with and participate in real world legal agreements. This conduit can be used for KYC/AML needs as well as handling situations of defaults and/or liquidation proceedings in the real world.

This same structure has been developed from our years of experience in Maker,where it today supports >1B USD in assets deployed under Andromeda and the Blocktower MIP6.

We agree that this is a significant challenge for crypto native organizations and even their sophisticated management teams. The analysis and assessment of credit opportunities is simply a different capability that most teams are unable to support.

Over the past year, the Centrifuge DAO has established the Centrifuge Credit Group as a resource to serve our own community. The Credit Group is a set of traditional finance professionals who can be engaged to provide bespoke recurring risk analysis for portfolios, assets, and deal structures and credit arrangements. With a diverse network of experienced and capable resources, the Credit Group can provide communities an additional layer of risk analysis necessary when dealing with credit investments.

Centrifuge Prime brings the relationship with the Credit Group into other communities in a dedicated and ongoing fashion. This capability can be used on an ad-hoc or regular basis, directly to the community or integrated with a treasury management team, and ultimatley can provide a significant value-add to a protocol’s risk management practices.

We agree and we’re happy to report that Centrifuge Liquidity Pools will enable investments to be made directly on the Celo network :slight_smile:

Ultimately, Centrifuge Prime is built to support a diversified portfolio of assets tailored to a protocol’s needs. The platform has available today treasury bills, asset-backed securities, and has a broad pipeline of deals across investment grade and high-yield opportunities. We would be happy to work with the Mento team to originate deals that meet their needs more specifically, relying on our ecosystem of high-quality asset managers and partners.

Bucket Description Target Return Liquidity
Liquid Lowest risk while maintaining US Treasuries, MMF, AA+ rated short term bonds 5-6% Daily
Preservation Investments in senior secured loans such as real estate, corporate bonds etc. 6-8% 3-6 months
Acceleration Trade finance, non-bank originators, emerging markets 8-20% 3-24mo

If the community believes bolstering and diversifying the Mento reserve is in its best interest, and it should, we would love to explore opportunities to onboard the protocol to Centrifuge Prime. This will deepen our ties with Celo and the community, and provide a critical service for bringing RWAs into the reserve, safely and soundly.


Hello Celo & Mento Community!

We’ve been following this proposal with great interest. The Flux Finance (Flux) Team is excited to contribute to the dialogue and believes that the Flux protocol aligns very closely with the goals outlined by @roman, with an emphasis on risk minimization and yield generation.

This proposal is provided for solely and exclusively informational purposes. This proposal does not make, and neither Flux Finance nor any other person makes, any representations or warranties of any kind, express or implied, at law or in equity, in connection with this proposal or the transactions contemplated hereby, including but not limited to warranties of merchantability, fitness for a particular purpose, title and non-infringement, and any and all representations and warranties are hereby disclaimed to the fullest extent permitted by applicable law.

Why Flux?

Flux is a decentralized lending protocol supporting permissionless assets like stablecoins alongside permissioned assets like tokenized US Treasuries. Flux was the first of its kind and is the market leader with over $48M in TVL since launching on Ethereum Mainnet in February 2023.

An Aligned, Non-KYC Solution

Lending stablecoins at Flux is KYC-free, and existing Mento treasury assets are supported. Flux is a Compound V2 fork and inherits the strength of a battle-tested protocol that managed over $10B in TVL. Compound V2 has been audited by Trail of Bits and OpenZeppelin, and has been formally verified by Certora. Changes to the Compound V2 open source codebase relate only to permissioning and Flux has undergone an independent Code4rena audit. Flux enables stablecoin loans collateralized exclusively by tokenized US Treasuries (OUSG from Ondo Finance).

Opportunity for Community Consideration

Mento currently holds over 5.8M USDC which isn’t earning any yield. These tokens could be lent at Flux with the position over-collateralized exclusively by tokenized US Treasuries. At the target utilization rate of 90%, this currently equates to 4.43% APY.

We are prepared to delve much deeper as per the community’s preference – this response serves as an invitation to open discussions. We look forward to exploring how Flux can contribute to the growth of Mento’s reserve over-collateralization, and we are happy to answer any questions that may arise.

Project Links:

Is there any mature DeFi protocol that hasn’t been breached or hacked at least in part over the past few years? What yield percentage is needed to make the smart contract and layering risks worthwhile here?

Will it still be a “Reserve” (semantically speaking) if it’s an actively managed DeFi account?

I’m not against this proposal in theory, but the risks feel almost incalculable.

Cicada Partners put together some research on this topic that I think is relevant. They calculate a ~3% risk premium (i.e. 3% annualized losses) for DeFi smart contract lending:

They also note that undercollateralized / credit lending has some different considerations from typical DeFi smart contract lending.

Collateralized lending Dapps represent massive honey pots of on-chain value, compared to undercollateralized Dapps that largely rely on smart contracts to track, move, and tranche value before it is lent out to borrowers.

In the case of the [undercollateralized pools], the design prioritizes capital efficiency, which has the added benefit of reducing the economic returns to nefarious actors. The expectation is that [a small portion] of deposited USDC will remain in the smart contract at any given time, with the majority of funds held off-chain with regulated custodians.

1 Like

Hi @roman,

Marcus from Adapt3r Digital here, the team behind tfBILL on TrueFi. Appreciate the insightful post and the direction that Mento is headed - we agree that the opportunity cost of not investing the stablecoin’s collateral is difficult to ignore. One thing we would also highlight is, counter-intuitively, the safety that bringing assets off-chain into regulated counterparties can provide.

We have designed tfBILL to be blockchain and stablecoin-agnostic, and as a result Mento can diversify its exposures away from both smart contract risk and issuer-specific risk (such as Circle, Maker, or Frax). The oversight that US regulators provide is often underappreciated in the Web3 community and is set up to provide a high level of transparency, investor protection, and operational security. By engaging with regulated US broker-dealers and qualified custodians, tfBILL has the capacity to scale with Mento while retaining an ultra-conservative stance on asset protection and legal recourse.

As @0xKhan so eloquently discussed, we do not believe KYC should be considered a blocker. While technical and logistical hurdles can be overcome, the DAO having real world legal recourse to underlying collateral is critical if it is going to scale materially from here. Ultimately what the DAO needs to decide is whether it cares about preservation of its capital or its utility and DeFi-compatibility. This is an embedded tradeoff today, and as a stablecoin it would logically follow that Mento should optimize for stability across its balance sheet. We note that most on-chain RWAs today have optimized for the latter (composability) over the former (stability).

Unlike other options, tfBILL is designed to be extremely simple and to be primarily used as an on-chain store of value - not a leverageable asset. While we believe the urge to make treasury tokens DeFi-compatible is alluring, a tokenized asset is only as strong as the regulatory compliance of its issuer (regardless of whether it is “bankruptcy remote”). This may be a hot take, but the more “permissionless” and “composable” a treasury token is perceived today, the higher the risk of failure once any semblance of institutional scale is reached. Additionally, I’m not sure if sDAI is truly decentralized. With Blocktower and Monetalis, both regulated asset managers, providing the underlying treasury yield, one is left to wonder how it is really different from the others.I guess the difference is just that they don’t KYC - which objectively raises the risk profile of the asset considerably.

As a result, we would posit that Mento should seek to diversify exposures away from purely decentralized collateral as a hedge against the inevitable regulatory crackdown that is coming.

Here are our key differentiators:

  • tfBILL is fully backed by actual treasury bills held by regulated counterparties in the US – not an ETF wrapper.
  • tfBILL is only available to qualified investors that can fulfill KYC obligations.
  • tfBILL is structured as a BVI Professional Fund to maximize investor protections while enabling on-chain access.
  • tfBILL is structured as an ERC-4626 to simplify operations and integrations.

Excited to see the conversation evolve and we’d love to hear from the community with any feedback or questions.