MIP13c3-SP5: Declaration of Intent: Maker to commence onboarding work of Centrifuge based Collateral

MIP13c3-SP5: Declaration of Intent: Maker to commence onboarding work of Centrifuge based Collateral

Preamble

MIP13c3-SP#: 5
Author(s): Lucas Vogelsang (@spin on forum.makerdao.com, [email protected])
Contributors:
Status: RFC
Date Proposed: 2020-09-09
Date Ratified: <yyyy-mm-dd>
---
Declaration Statement: MakerDAO supports Centrifuges effort in developing an onboarding process for their RWA as collateral to MCD.
Declaration to Replace: n/a

Specification

Context and Motivation

Onboarding Real Wold Assets to Maker is seen as one of the main priorities to scale MakerDAO. This is a goal Centrifuge has been working on for over a year. It requires adapting the current collateral onboarding process as these assets are different from the current collateral. Centrifuge has worked with both the domain teams as well as the broader community to discuss these differences and come up with solutions such as the MIP22 proposed in this cycle.

Onboarding real world assets to MakerDAO is a unique challenge, that - similar to the migration from SAI to DAI - might seem as an unsourmountable task at first. However, it is only the natural extention of the current MCD building on prior learnings. This declaration outlines a path towards onboarding those assets at a low debt ceiling to gain first experience within the community and kick of the next evolution of Multi-Collateral-Dai towards a diversified asset backed lending protocol.

With this endeavour come many questions that the community hasn’t had to address so far. Those need to be diligently answered by setting up a thorough process. This process should MakerDAO can gain valuable experience by experimenting at a small scale slowly growing the amount of DAI that can be generated from these assets as it gains confidence in the system and the system is improved to offer further protection.

This declaration of intent is the first step to developing an onboarding process specific to Centrifuge based RWA applications in MakerDAO.

Collateral Overview

Centrifuge has partnered with a number of different asset originators that can be onboarded to Maker ready to add demand for DAI and diversity to the system. These asset originatiors jointly have capital needs of well over DAI 50M and bring a diverse set of crypto-uncorrelated assets to Maker. All of these collateral types have followed the standard MIP6 process and have already been greenlit by the communiy (except KickFurther which is in the current cycle).

The following collateral applications are to be considered for this new process. They are listed with their capital demands and their current cost of capital as a benchmark for the Maker to use as guidance on the opportunity to generate DAI.

  • ConsolFreight, Financing freight invoices, Capital needs in DAI: up to 5M, Current Cost of Capital: they have a bank line of credit at 1.5% for $50k, MIP6
  • New Silver: Financing real estate backed loans, Capital needs in DAI: 15M, Current cost of Capital: NS holds a private fund credit line of $20M with a 3MO LIBOR +4.75%, MIP6
  • Paperchain, Financing music streaming invoices, Capital needs in DAI: 1M, Current cost of capital: sourcing from independent investors and funds at 6-10% range, MIP6
  • KickFurther, Financing consignment inventory for eCommerce & retail brands, MIP6 application to follow
  • Harbor Trade Credit, Financing short term trade finance transactions MIP6

Declaration Detail

The MakerDAO community declares its intent to onboard Centrifuge’s assets according to an onboarding process outlined below. Centrifuge is encouraged to develop and implement this process for its proposed assets resulting in a MIP template that can be used for this category of assets broadly based on the current MIP6-MIP12 process.

Given the proper due dilligence by the Maker community and the domain teams, Centrifuge is encouraged to use the developed MIP with collateral types as soon as they are ready.

Scope of this onboarding process

The process is designed to be safe for onboarding assets up to a certain limit. If these limits are hit, governance must decide if the debt ceiling can be increased gradually or how further protection (such as more due dilligence or technical improvements) should be added. Assets onboarding with this process should at the beginning not surpass 4M DAI in debt ceiling per collateral type. And the total for all collateral types following this process should not surpass 15M DAI to stay within a bounds that even in the worst case scenario would not endanger the Maker project and DAI.

Collateral Onboarding Form (MIP6)

The regular collateral onboarding process used for tokens like 0x, Loopring etc. is an inspiration for this process. And all assets to be considered for onboarding should start out with a MIP6 proposal. In the future a likely adaption of this might be to create multiple variations for MIP6 proposals depending on the collateral type. For now, this is is out of scope and the MIP6 and MIP9 Community Greenlight Poll should be used.

Onboarding Process Component 1: Liquidation Process

Real world assets require a different liquidation than using the existing keeper infrastructure. The technical changes are defined in the proposed MIP22. For the collateral type, the application should outline how the portfolio can be liquidated through the mechanism defined in MIP22 listing different offchain lenders. The community must exercise due dilligence on this process ensuring that in fact these assets can be liquidated off chain.

Requirements for onboarding:

  • Description of the liquidation process overall
  • Detailed model of the expected losses incurred at liquidation as well as the speed of which these assets can be liquidated
  • Supporting documentation/evidence of available off-chain refinancing options for the collateral if the loans can’t be held to maturity

Onboarding Process Component 2: Smart Contracts Work

Like any other collateral added to Maker, the technical integration must be audited prior to onboarding the collateral type. Centrifuge is developing the necessary adapters as well as the executive code (the spell) that can add onboard a Centrifuge-based asset to MCD. As part of the smart contract work that will go into this, Centrifuge will do a full dry run of this with a Kovan-deployment of DSS and work with the smart contracts domain team as well as other developers familiar with the project. In addition, the code will be audited by an audit firm that has performed audits of the Maker codebase before.

The resulting set of contracts that are to be used then only needs to be minimally modified for each new collateral type. Any modifications to the code would require the re-evaluation of the technical onboarding process.

Requirements for onboarding:

  • List out any modifications deviating from previously approved versions of collateral adapters.
    • If there are none, no external audit necessary
    • If there are changes, resent a thorough technical review and audit of any modifications

Onboarding Process Component 3: Risk Work

The onboarding process for risk will be in conjuction with the risk domain teams as their capacity permits. All asset proposals need to include the usual risk parameters for the system. This includes the parameters below:

  • Stability fee: fee on Dai that is generated from a RWA Vault should be competitive with the current cost of capital of the asset originator or the industry standard rate of the specific asset.
  • Debt ceiling: We propose the maximum amount of Dai that can be generated from Vault/RWA should not be higher than 4m in the first itereation. The debt ceiling for RWA as collateral should be determined by the capital need / origination volume of the asset originator.
  • Collateralization ratio: The ratio between the value of the collateral and the value of the generated Dai for the specific RWA vault.
  • Downside protection / TIN ratio: The ratio between DROP and TIN (Senior and Junior tranche). This is an additional saftey cushion for any RWA vault. It should be derived from the asset originators historic default rate and/or the asset’s historic default rate (e.g. Invoices as an asset type have a historic default rate of less than 1%)
  • Reporting: Present a Risk Assessment Report co-authored by active Maker community members; either by the current Risk Domain Team or longer term a Domain Team specializing in non-crypto assets.

Executive Vote

Following a positive community greenlight (MIP8) and completion of the work outlined in the three components above, the result should be an executive vote on the spell onboarding the asset to the DAO based on the risk parameters. A MIP template will be created for this.

Relevant links

MIP6 Proposals

6 Likes

Hi @spin,

I am a bit uncertain about this declaration of intent.
The Centrifuge collateral applications are as far as I know being worked on - there is no need for additional declarations of intent in order to start or expedite the process. Just look up other collateral such as MANA, it is application->discussion->poll->domain work. The domain work can take some time.

The Centrifuge collateral applications have so far received very little attention from the current domain teams and we’ve drafted this proposal talking to governance stakeholders and domain team members. Domain Work on Centrifuge assets has not started and the current domain teams are not well equipped to handle on-boarding these assets. Centrifuge has submitted some of the first MIP6 proposals receiving by far the most community greenlight votes as well as being widely received by many members and MKR holders as a key collateral type (RWAs) to be added. Our asset originators have the need to generate DAI in the millions and are willing to pay a stability fee >0 for that. Yet the community so far, following the MIP6-12 process has only onboarded relatively small crypto assets as collateral (ZRX, MANA and KNC) with currently less than 500k in DAI generated. All the while DAI is back at ~$1.03.

The reason for the slow progress was mostly due it not fitting into the existing collateral types and general uncertainty of how to deal with them. Recent departures at the foundation have not improved the lack of resources and attention to onboarding RWA and the process of onboarding new domain teams is in progress. With the plans to completely dissolve the foundation, it would be counterproductive to that goal for the foundation to start hiring and building out its own team leading that effort and thus this must come from the community. I am very excited to see the progress and work being done to onboard new domain teams and the MIPs that @LongForWisdom et al. have proposed to better structure these teams.

All this work will ultimately lead to a structure that will be able to deal with the diversity of assets Maker really needs. But I think this is not something that should slow the progress of addressing the DAI peg. With the launch of our third version of Tinlake (being audited right now) we have both the technical functionality on our end as well as capacity now to do a lot of the onboarding work by ourselves. Unlike other projects, we’ve worked with the multi-collateral-dai code base from the early on and are intimately familiar with it. We had a previous version of Tinlake integrated directly into MCD in pre-release version already last summer originating a loan with DAI minted with our collateral.

I obviously have my own interests in ensuring that there is a way for Centrifuge collateral into MCD. There’s no denying that but Centrifuge, a now 2.5 year old project, has proven again and again that we’re not short term oriented and share the vision of what DeFi should become with many of the Maker stakeholders. I believe it’s in the best interest for Centrifuge and Maker to start experimenting with these assets and make concrete steps for onboarding these assets. This is what we’re asking: the community’s to cooperate and support us in this endeavor. That’s the purpose of this declaration of intent.

5 Likes

This post I just made clarifies a bit what the intention of this declaration is and how it fits into the greater picture of Centrifuge assets in Maker: Centrifuge: Onboarding RWA Backed Collateral to MCD

1 Like

I have asked if a status update on this is possible for today’s Governance meeting.

1 Like

From a risk team point of view, I would not say this is accurate. The teams have members that have experience dealing with RWA. I would say, there are two points to consider. First, it’s a matter of dealing with a growing backlog of assessments and prioritising all the work (RWA and others), while still maintaining a certain standard of quality. Second, when dealing with RWA there will be a certain infrastructure to be built for collecting data, reviewing/adjusting models and monitoring performance. That will require some work for sure. But I would not say the team wouldn’t be able to handle these assets/activities.

Centrifuge and our asset originator’s interactions with the risk team have been very limited so far. We usually invited Cyrus, and by extension the entire domain teams to take part in the dialog around how to safely onboard these assets, for example in the community calls. There were good discussion concerning risk and we’ve summarized them in the forum and mentioned the ongoing work to the risk team.

It seems like it didn’t make the cut above the backlog of tasks that you mentioned. An issue that became apparent in the conversation on yesterday’s GNR call but also before already I believe was mentioned in a thread: there isn’t enough transparency towards the community on how domain teams are onboarding new collateral. Projects submit MIP6 proposals and then have no idea what the path forward is.

I’m glad to hear that there is interest & capacity to start figuring out reasonable risk parameters for these asset classes. I will propose to set up a call specifically to talk through these assets and figure out the next steps on the risk assessment.

3 Likes

Going to just read over and leave comments (later than promised, sorry, been busy!). :slight_smile:

This is awesome! Is there room for scaling up these capital needs or are they fairly fixed?

This should read ‘Maker Token Holders’ or ‘Maker governance’, imo. I know in practice the terms are often used synonymously, but it’s good to be precise here.

Great!

This would need to be fixed right? Can you maybe highlight that?

itereation -> iteration.

This is something we need to set? I assumed Centrifuge would be pushing for a specific ratio.

Community Geenlight is MIP9.


This reads really well, @spin, super clear and concise. Great work, looking forward to seeing this become a reality.

2 Likes