This document reflects solely the author’s opinions and is provided for informational purposes only. The content of this document should not in any way be relied upon as legal, business, investment, financial or tax advice. Although the document has been prepared based upon information shared by some of the parties mentioned, the quality and completeness of such information has not been independently verified and this document makes no representations about neither its accuracy nor appropriateness.
This document intends to offer the outside-in perspective of an experienced structured credit investor with regards to the onboarding of the FFT1-DROP token as a new RWA vault type. The onboarding followed the MIP6 application dated 11th of January 2021, the approval of the related executive proposal on the 4th of August 2021, and the consequent execution on the 6th of August 2021. The document will touch, although briefly, several aspects of the matter, including:
- Commentaries on the underlying credit exposures
- Key risks and mitigating actions
- Notes on the underwriting process
- Outside-in perspective on MakerDAO’s unique characteristics
The document has been independently completed by Luca Prosperi (@luca_pro, [email protected] – you can find a summary of Luca’s relevant experience at the bottom of the document). Any reaction and commentary from members of MakerDAO’s community will be most welcome.
Fortunafi is a financial intermediary (the intermediary) that partnered with Centrifuge (the structurer) to launch a financing pool with the technical abilities to attract on-chain funding from partners such as MakerDAO (the senior investor). The financing pool behind FFT1-DROP would focus on revenue-based financing (RBF) as underlying credit collateral. In accordance with Centrifuge’s construct, the financing pool, tokenised through the Tinlake platform, has been subdivided into a TIN (junior) and a DROP (senior) tranche. DROP will be the collateral deposited in the MakerDAO’s vault.
As mentioned, Fortunafi acts only as an intermediary by (i) identifying and partnering with credit originators, (ii) sponsoring the structuring of the exposure, (iii) retaining part of the exposure, and ultimately (iv) marketing the remainder to third parties. At the time of the MIP6 application, Corl was the only asset originator constituting the financing pool. Since then, a new arrangement with Pipe, another originator, has been reached. Pipe is now expected to originate the vast majority of the financing pool. Both those originators, however, are not the ultimate borrower of the credit issued, acting themselves as tech-enabled intermediaries.
Although facilitating diversification and scalability, the long value chain borrower → asset originator → intermediary → structurer → senior investor presents several risks that should be carefully addressed in the due diligence phase and, especially, through the definition of the underwriting criteria and structure. We will touch on all those points below.
The underlying exposure of the FFT1-DROP is expected to be entirely constituted by RBF.
In RBF, investors inject capital in a business in return for a percentage of originated revenues, with either definite or indefinite maturity, with the possibility for the business to terminate the arrangement by buying out the outstanding loan at a pre-agreed multiple. The credit investors might in addition ask for the issuance of equity warrants, not to improve the underlying credit risk but rather to further enhance returns. Based on the information provided, underlying borrowers may or may not offer personal or hard asset guarantees on the credit; given the start-up nature of the ultimate borrowers, the presence of those guarantees does not constitute a game-changer for credit robustness.
The loan, de facto, is sustained only by the future cash flows (both operating and investing) of the business and has a risk profile similar to that of preferred equity – given the existence of contractual interest payments senior to dividends and other distributions to shareholders. In practice, this type of financing has been used by founders to bridge their needs between venture financing rounds, making it a specific type of venture financing. The business and equity investors are incentivised to buy the loan out using fresh funds gathered in each round – typically when revenues are expected to improve significantly. RBF is therefore inherently pro-cyclical and might not offer beneficial dynamics to investors: in best case scenarios of solid revenue growth and available financing borrowers are incentivised to pre-pay the loans and consequently reduce the effective time to generate interest, while in phases of stagnant growth and lacking venture funding credit performance will be impacted. This pro-cyclicality also suggests the risk of high correlation among individual exposures, making tranching less effective.
In their Collateral Onboarding Risk Evaluation, the RWF CU anchored expected probabilities of default in the range of 1-4%. This estimate seems extremely optimistic based on observed performance – information protected by NDA. Given the nature of the credit, and the potentially sudden effect of macroeconomic dynamics on the credit quality, any covenant package might have limited effect in protecting MakerDAO. Although this type of risk-return profile might still be attractive for venture funds and similar investors, it might not have the same level of attractiveness for lenders like MakerDAO, focused on providing a positive return but with the preservation and expansion of the DAI as the core ambition.
Fortunafi is a fund manager focused on alternative credit, represented in forums by its CEO, Nick Garcia (@_nick, [email protected]). The firm is focused in generating yield from tokenising real-world cash-flowing assets, and it was founded in 2020. Based on the original MIP6 application, Fortunafi was created specifically to partner with Centrifuge & Corl and launch an on-chain credit fund. At the same time, following the partnership with Corl and now Pipe, Fortunafi intends to expand the roaster of originating partners and become one of the key players in connecting real-world borrowers and DeFi liquidity providers. Centrifuge has built a tech solution to tokenise collateral assets and be utilisable on-chain. Through their on-chain securitisation platform, Tinlake, Centrifuge has currently c. USD 32m of total value locked. Based on the conversations had with Fortunafi, the firm expects to fully finance the TIN junior tranche of the financing pool – more on structuring below.
Corl serves as an onboarding platform for businesses seeking funding. Businesses can apply for financing through Corl’s web infrastructure that acts at the same as a financing platform and a business dashboard with access to the underlying payment, banking, and customer data. Based on the documents provided, Corl does not participate in the financing of the businesses originated and scored through their platform, acting as a pure distribution channel for credit investors. Fortunafi’s stresses Corl’s ability to leverage new alternative data sources and machine learning techniques to assess credit quality; based on the author’s experience, and on analysis performed by credit risk specialists like Oliver Wyman, the ability of those techniques to enhance human credit underwriting is yet to be proven.
Pipe is the next originator that is supposed to constitute underlying exposure of the vault. Based on the limited information accessed, the credit originated by the firm is supposed to be significantly lower risk compared to Corl’s, with a focus on 12-month term loans for large SaaS providers with at least USD 5m revenues. Pipe is expected to represent 90-95% of the pool, making the initial due diligence performed with a focus on Corl largely redundant. Although understandable in such early stages of development, Fortunafi’s shift of focus in the very initial stages is a cause of concern that should be monitored.
At the time of writing, no DAI has been minted through the FFT1-DROP vault. However, based on the conversations had with the parties involved, it is not clear what proportion of the DROP tranche will ultimately be allocated to MakerDAO, and who would be the other DROP co-investors. This information will be crucial to both provide a proof of concept and to offer operating leverage to take remedial actions in case of need, given the peculiar nature of MakerDAO as non-legal entity.
Fortunafi has shared with MakerDAO the most updated (end of July 2021) dashboard of the underlying pool composition – composed only by loans originated by Corl at the time of writing. A high-level analysis of the loan development provides a mixed view:
- Small size compared to debt ceiling, with levels of collateralisation in check
- Size (by annualised revenues) and growth trajectory of the financed business is below target
- Although single investees don’t represent a significant portion of the debt ceiling, the outstanding loans seem quite concentrated (40% represented by one borrower); this seems physiological in ramp up
Interestingly, the dashboard developed by the RWF CU doesn’t seem to include a satisfactory view of credit quality development. This could be due to the nature of RBF – and this brings us back to the point that it is quite difficult to preside over credit quality in such asset class. Based on the author’s understanding, it is Fortunafi’s responsibility to provide an updated Net Asset Value (NAV) of all credit positions – given the early days of the partnership, this methodology hasn’t been investigated, but the RWF CU should monitor it alongside observable events.
Fortunafi acquires RBF assets from the originators via a bankruptcy-remote SPV. The SPV, via Centrifuge’s Tinlake infrastructure, will tokenize the title of each underlying asset into NFTs that will BE added in Tinlake as collateral. Those tokens will be utilised to issue FFT1-DROP (representing 90% of the pool, pledged into MakerDAO’s vault) and FFT1-TIN (representing the remaining 10%, fully purchased by Fortunafi). In customary manner, the SPV will have authority over the underlying collaterals. The investor will be represented by an independent director within the SPV, but MarkerDAO won’t have authority over the underlying collateral in case of adverse events – given its non-legal nature. Based on the information shared, there is no contractual obligation on the participation of Fortunafi and other co-investors into the DROP and TIN tranches. The same applies to a randomised vertical participation, typical of off-chain collateralisation deals.
Based on the FFT1-DROP parameters, MakerDAO will have an additional margin of safety. With a minimum vault underlying collateralisation ratio of 116.6%, MarkerDAO will mint DAI representing only c. 86% (i.e. 1 / 116.6%) of the underlying collateral, meaning that the implicit attachment point of the senior tranche would be c. 12% rather than 10%. Based on the conversations held with Fortunafi, it is understood that the SPV presents an excess spread between the contractual interest payments from the borrowers and the interest paid – this excess spread will function as a tool to replenish the junior tranche in cases of losses; it is our understanding that the full residual value of the structure will be of benefit of the junior investors – this is not uncommon in the market. The documentation of the underlying SPV hasn’t been provided, but it is expected that the junior investors will be obliged to always maintain the minimum junior tranche requirements also per SPV terms, with the ability to distribute at least part of the residual value before the end of the vehicle.
The RWF CU has proposed a set of soft (i.e. non-enforceable) covenants, mainly covering:
- Allowed underlying investments
- Investment concentration
- Fortunafi junior participation – minimum 60% or USD 1m
- Senior co-investment – minimum 25% or USD 1m, and 10 co-investors
- Monitoring guidelines
However, it is our understanding that, given the non-legal nature of MakerDAO, the ability to take remedial actions in case of unsatisfactory performance of the underlying collateral or business partners might be extremely limited. The co-investment requirements, in this context, become of extreme importance.
In case of insufficient junior collateralisation (due, also but not only, to unsatisfactory performance), the liquidation mechanism will follow Centrifuge’s model, as per the MIP22. Interestingly, the MIP22 analysed states that the liquidation mechanism would work well for loans “[…] of very short maturity and waiting for loan maturity is economically viable” or “[…] The loans can be sold off (refinanced) by the issuer off chain and there is a process in place for the liquidation of the pool in case any DROP token holder (such as Maker)”. The type of engagement behind FFT1-DROP doesn’t seem compatible with those requirements. The diagram below represents the liquidation mechanism when triggered.
Underlying pro-cyclicality and correlation of RBF
- Commentary: RBF is a specific type of venture financing with inherent pro-cyclicality and correlation, impacting the effectiveness of tranching in adverse phases of the credit cycle; based on the author experience, this type of financing is sub-optimal for MakerDAO’s risk-return profile
- Available mitigating actions: increase collateral requirements, introduce mezzanine financing, maintain/ reduce debt ceiling, avoid RBF for future exposures
Fortunafi’s focus seems shifting from the original intentions stated in the MIP6 application
- Commentary: the FFT1-DROP vault would have been constituted initially to provide on-chain financing to the credit originated by Corl only; with time, also given an apparently unsatisfactory performance of the originator, another originator, Pipe, has been added to the pool – this changes of focus, happening in the early days of the proposed partnership, should be monitored
- Available mitigating actions: although not much can be done to mitigate this risk now, the need for an approval by MakerDAO (governance or RWF CU) to add new originators with the impact of changing the risk profile of the financing pool is expected to be a soft covenants
Separation of origination, risk underwriting, and financing
- Commentary: this separation, typical of a pure originate and distribute model, and the consequent misalignment of interests (with the originator incentivised to intermediate more and more volumes, and the financier focused on net yield and credit quality) could have negative effects on credit performance during the adverse phases of the credit cycle – this is particularly important given the nature of the underlying exposure
- Available mitigating actions: originators should be required to retain a super-junior exposure on the originated credit, throughout the life of the loan; best practices would require the originator to offer a buyback guarantee of non-performing loans – e.g. following 60 days of delinquency
The implied IRR of the investment, and of the junior tranche especially, might not be consistent with MakerDAO’s conservative nature and mandate to preserve the DAI
- Commentary: based on the estimates provided by the underwriting partners, the internal rate of return (IRR) achievable by RBF-types of exposure is extremely high, in the 20-40% range, similar to that of venture investing; this IRR could be significantly enhanced for the junior tranche investors – currently, FFT1-DROP Stability Fee stands at 4.5% for an implied attachment point of c. 12%; given the nature of MakerDAO’s interests the risk-profile could be enhanced via more prudent structuring or higher stability fee
- Available mitigating actions: increase stability fee, increase collateral requirements, introduce mezzanine financing
Given the nature of the underlying credit and the allowed concentration, the senior attachment point seems aggressive
- Commentary: FFT1-DROP has an implied attachment point of 12%, meaning that following defaults constituting 12% of the pool the vault would start suffering losses – this seems aggressive considering the risk profile of RBF; in addition, with the maximum allowed concentration as per the soft covenants at 30% of the debt ceiling, the default of one single borrower (with recovery rate expected to be very low considering the nature of RBF) would wipe out 20% of the vault, i.e. > 4 years of stability fee for the whole pool
- Available mitigating actions: reduce maximum exposure concentration limits, increase collateral requirements, introduce mezzanine financing
Given the limited ability of MakerDAO to enforce remedial actions, the co-investment requirements might be insufficient
- Commentary: it is understood that MakerDAO does not have contractual obligations in place with the SPV nor any other party, and this could dramatically impact its ability to act in remedy of unsatisfactory performance of the underlying credit or partners; in this construct, the existence of large and reputable senior co-investors becomes key
- Available mitigating actions: increase the soft covenant requirements for the presence of senior co-investors, ideally transforming MakerDAO into a minority senior investor; those co-investors, if institutional, would themselves require enhanced representation and control
The liquidation mechanism might not be compatible with the underlying nature of the credit
- Commentary: the liquidation mechanism within the Centrifuge construct might not be compatible with the bulky, asset-light, low-churning, nature of the underlying credit – this could expose MakerDAO to losses due to the expected low recovery rate of non-performing loans
- Available mitigating actions: although not much can be done to mitigate this risk now, increasing the seniority of MakerDAO’s exposure might be a partial solution
Given the indirect nature of the credit origination flow (borrower → asset originator → intermediary → structurer → senior investor), it becomes clear that MakerDAO’s due diligence and underwriting should focus on the direct contact point – in this case Fortunafi; of its history, capabilities, shareholding, financial structure, and incentives. MakerDAO seems to have had extensive conversations with Fortunafi in order to understand their modus operandi. Based on the materials reviewed, however, the analysis of the underlying credit scoring ability of the originators, and of the asset classes to be underwritten, seems unsatisfactory on MakerDAO’s side. Fortunafi’s analysis of the platform’s credit underwriting ability doesn’t seem profound enough, and the same applies to the monitoring of credit quality. Based on the author’s experience, one credit risk analysist should be entirely dedicated to shadow rate the portfolio of the originators based on the data tapes provided – usually one analyst can cover 2-3 originators and c. USD 100m of outstanding loans, and it isn’t clear whether Fortunafi has currently those skills in house.
Where structuring is concerned, an aspect of paramount importance in structured credit, MakerDAO’s accepted terms and soft covenants do not seem to be fully aligned with the conservative nature of MakerDAO’s mandate. The agreed implied attachment point, the maximum concentration limits, and the co-investors requirements do not seem fully aligned with MakerDAO’s main priority, that it is in my understanding a scalable yet sustainable growth of the DAI footprint. At the same time, alternative legal constructs should be investigated – e.g. providing delegation to director(s) acting in representation of MakerDAO, informal yet transparent agreements with back-up servicers, forward flow agreements with non-performing credit investors, etc. All those aspects seemed absent in the negotiated package.
In addition, the liquidation mechanism applicable within the Centrifuge construct does not seem compatible with the type of underlying exposure. In a standardised, off-chain, loan agreement, such a circumstance would have been mitigated by a pledge of assets by the direct intermediary (in this case Fortunafi), beyond the intermediary’s junior exposure within the SPV. Although this cross-collateralisation might be an effective mitigant in TradFi, it is my understanding that MakerDAO wouldn’t be able to independently enforce such remedial action if negotiated. This is another reason the presence of a large, institutional, co-investor would be extremely beneficial for MakerDAO. The bankruptcy remoteness of the SPV vehicle, as well as the presence of an independent director, acts only as a partial mitigating factor.
Providing financing to real-world assets via MakerDAO’s DAI is an ambitious, yet extremely challenging, task. On-chain securitisation and pseudo-securitisation has a series of challenges that go beyond traditional structured credit. In addition, the extremely low cost of capital of institutional investors these days doesn’t reduce the adverse selection issue of originators seeking financing on-chain – adverse selection that wouldn’t translate only in a low quality of credit, but also in an extremely short corporate history of the intermediaries. The difficulty of the challenge the RWF CU has tackled shouldn’t be underestimated.
At the same time, even in a bootstrapped start-up setting, sufficiently solid governance, structuring, and underwriting practices should be implemented. This in order to preserve the DAI, as well as to build the foundation of an increasingly ambitious real-world financing plan.
Defining the role of MakerDAO regarding financing projects in the real world goes way beyond the scope of work of this audit report. However, a few high-level suggestions already emerge. MakerDAO might benefit, for example, from a more passive position in relation to the underwritten credit. This would translate, in practice, into:
- Clearer underwriting criteria published ex ante in MakerDAO’s forum for potential counterparties to measure themselves against
- More robust structuring requirements, potentially discussed and agreed with Centrifuge and other similar service providers, also published ex ante in MakerDAO’s forum
- More stringent co-investment requirements, both in terms of quantum, relative exposure, and nature of the co-investors to partner with
- Clearer internal governance rules, in order to create effective multiple lines of defence that go through RWF CU, independent risk management, and ultimately governacne and the larger community
- A replication of the centres of excellence within MakerDAO that currently deal with real-world assets, in order to avoid the existence of a single point of failure and further protect the DAI
Those suggestions need to be considered only as indicative. It is the opinion of the author that a more extensive and deeper effort to clearly define MakerDAO’s role in the real world of credit should be considered.
Luca Prosperi, MakerDAO believer and senior finance professional with 15 years of experience in banking and credit investments:
- Publishing Dirt Roads, a Substack newsletter on the evolution of banking and DeFi
- Originated and financed c. EUR 200m of credit through alternative lenders via both a banking platform (Citadele banka, ECB-regulated bank where Luca was Board Observer representing Ripplewood as the majority shareholder), and a credit fund (while at Partners Capital, a c. USD 40b AuM outsourced investment office)
- Completed deals totalling > USD 5b USD throughout 5 years as a private equity (Ripplewood Advisors) and private credit (Partners Capital) investor, with focus on financial institutions and fintech companies
- Advised on several high-profile transactions throughout 4 years as M&A banker in Morgan Stanley’s London FIG team
- Advised financial institutions on the topics of risk management, strategy, and business growth for 5 years as strategy consultant at Oliver Wyman
- Master in Business Administration at London Business School, MSc and BS in Mathematical Economics and Econometrics at Bocconi and Tel Aviv University