Discussing the Pool Delegate Role and Strengthening Pool Delegate Alignment

Author: ParaFi Capital (Josh Solesbury)

Thanks to members of the Maple Finance team, of the Maven 11 team, and of the Orthogonal Trading team for their review of and feedback on this post.

Maple Finance has witnessed a period of scalability and success over the last year, something the community should be tremendously proud of. Compared to CeDeFi and other DeFi undercollateralized lending platforms, Maple fared much better during the recent months of market stress, exhibiting both lower capital outflows and superior loan performance.

As tends to happen with successful startups, working hard to keep up with growth can leave some uncovered gaps that present future risks. For Maple, we believe one of those potential gaps lies in the relationship between pool delegates (especially future ones) and their loan portfolio. Up to this point, pool delegates have been a crucial part of Maple’s success with regards to (1) attracting institutional capital, (2) scaling platform AUM, and (3) creating an efficient underwriting engine for uncollateralized and undercollateralized loans.

Looking towards the future, the scaling of Maple’s platform places some pressure on the current pool delegate system, which incentivizes pool delegates to make responsible lending decisions through reputation and bespoke, off-chain agreements. This cannot be guaranteed at scale, and thus we propose a discussion on the establishment of a new pool delegate structure, among other changes, whereby pool delegates are better aligned with the provision of capital to creditworthy parties.

Before diving into the components of this post, it is important to highlight the key principles that should underpin the eventual implementations of its ideas and the discussions surrounding them:

  • Alignment - Overall, this initiative aims to build greater alignment between the interests of pool delegates and lenders. A successful Maple Finance is one in which pool delegates have enough ‘skin in the game’ to underwrite lending opportunities as if they were the lenders themselves. While Maple Finance has generally been fortunate to have great pool delegates in this regard to date, there must be an effort to codify processes that make sure all successive pool delegates behave in an aligned manner as well.
  • Scalability & Capital Efficiency - While this post urges the deepening of alignment, it is cognizant of the need to balance ensuring ‘skin in the game’ with a sense for capital efficiency and a scalable pool delegate experience. Growth of pool delegates / pools, along with the deepening of pool delegate relationships, is crucial to consider as changes are discussed.
  • Flexibility & Realism - All changes and derivations from this post should weigh in the need for flexibility. Different pools (permissioned vs. permissionless), different pool delegates (quality of team, experience, etc.), tenure / track record with Maple, and the like must be dynamic variables in proposed reforms. A one-size-fits-all approach will fail to drive value. On the other hand, a customized solution for every pool delegate does not make sense either.
  • Long-term Vision - This discussion has a significant focus on the long-term success, scalability, and evolution of Maple Finance as well as that of on-chain credit.

Current Status
Today, the provision of cover, or first loss capital, by pool delegates is small in relation to the pool size. A look at Maple’s documentation (found here) suggests that pool delegates must “stake a minimum of $100,000 worth of BPT when they establish a pool”. Note, BPT stands for Balancer Pool Token, a liquidity share of the MPL:USDC 50/50 pool. While this was a fair number back when Maple started, the tiny fraction it represents of typical pool sizes today reduces its ability to truly incentivize pool delegates to assess credit properly. Potential areas for exception lie in pools where either the pool delegate and the lender are the same party (as with the prior Celsius pool) or the pool is dedicated for a single borrower (as with the Alameda pool, with Orthogonal as the delegate).

Given the above, the Maple team has been motivating delegates to go beyond the $100,000 amount (both on and off-chain); however, as seen below, the on-chain quantum is still generally small relative to pool size.

*Sourced from Maple’s Application Interface (link)

A couple of important points to share here. Firstly, this is not meant to be construed as an indication that current pool delegates aren’t well aligned. In fact, Maple’s application interface understates the commitment of pool delegates to the pool, not accounting for how much they have in the pool itself as capital for lending (which some pool delegates have). Additionally, pool delegates are in many cases dedicated businesses with full time teams and allocated resources. Their investment with regards to operational expenses and time form another component of commitment. Finally, pool delegates today have other sources of alignment pertaining to reputation and long-term investment in the success of Maple.

Yet, the alignment that is harder to gauge with certainty is that of pool delegate 15, 30, 60, etc. One thing in crypto that we all spend more time thinking about than nearly any other field is incentives. Looking through these numbers, it becomes harder to completely assure that new pool delegates joining as Maple scales will act in the best interest of lenders when it comes to facilitating borrowing. And, given the criticalness of trust in the uncollateralized lending space, the Maple community must make sure to preserve a solid track record of enabling healthy, creditworthy opportunities.

Areas for Changes & Intended Benefits
In order to drive better alignment, we introduce several areas for potential change that seek to better nurture the key principles outlined at the start.

Increasing ‘Skin in the Game’

Starting at the foundation, we suggest a dynamic, tiered cover requirement structure, whereby pool delegates must contribute a minimum of a calculated amount based on the pool size they manage. As the pool size scales, the absolute quantum of first loss capital increases, however, the proportional amount (i.e., Min % Cover) decreases. Similar to U.S. tax bracket mechanics, the total cover amount could be the sum of the cover mandated at each incremental tier.

This suggested structure has firm grounding in the world of traditional finance. A poignant example lies in the post-2008 regulations enacted on collateralized loan obligation (“CLO”) managers. One of the major ingredients that led to the tremendous economic damage induced by the Great Financial Crisis was the Principal-Agent problem. Namely, financial firms building these complex financial products didn’t have enough ‘skin in the game’ to align them with the buyers of the products. When the returns of risks are removed from the consequences, the alchemy of finance can turn very sour. To mediate this, the Credit Risk Retention Rule was adopted by the Securities and Exchange Commission and five other U.S. regulators, which requires managers to retain no less than five percent of the credit risk for any asset that they securitize.

The main benefits of a tiered cover structure center on the provision of a stronger incentive to keep pool delegates aligned with lenders. While potential agreed upon tiers can seem to be a lot for certain pool delegates, note that (1) the capital required is small relative to the quantum being managed and is comparable to other alignment structures seen in the private and public capital markets and (2) the capital in the pool should be earning a compensatory rate of return on terms structured by the pool delegate (more on this below).

Note, there should also be agreed upon exceptions or modifications to this cover amount depending on certain predetermined qualifiers. For example, a permissioned pool structured by a pool delegate could have a different cover structure than a permissionless one. Additionally, pools where pool delegates and lenders are the same entity could be exempt from large cover requirements.

While this structure is adequate for the position Maple finds itself in today and for the near to medium term future, there could be rationale for modifying it later down the line to (1) accommodate new tiers when pools reach $b+ in size, (2) combine it with other qualifiers that have an impact on the Min % Cover (i.e. credit history on the Maple platform, which is discussed below), or (3) reflect innovation in incentive structures.

Balancing Requirements with Credit-built Relationships

While the structure above gets at the central issue of alignment, it may add friction to scalability originating from trusted pool delegates. The very essence of debt and credit, anthropologically, lies in trust, and there is a discussion to be had on whether the design of Maple Finance should more accurately reflect that.

As a large cover amount may be burdensome from a capital perspective, the platform could see a component of it funded with pool revenues earned by the delegate. In fact, some pool delegates already have off-chain agreements in place to do this. While both are forms of ‘skin in the game’, the former makes more sense for parties for which there is less trust (the tradeoff being more capital up front), while the latter makes sense for parties which have years of experience on Maple (the tradeoff being more capital flexibility based on a deeper and successful lending history). There could be ways to iterate on the cover structure above by factoring in time on the Maple platform as a pool delegate. More time, and consequently more lending history, can lead to more capital efficient cover structures.

Ensuring Resilient First Loss Capital

In addition to the above, we also urge changing the use of BPT tokens as the required form of pool cover. The rationale for this is founded on three key components. Firstly, the use of BPTs, and underlying MPL, as first loss cover could evolve quite negatively in a left-tail scenario where multiple first loss capital positions are being called in and effectively sold. This would create a negative spiral for MPL, affecting first loss cover positions across the board. It could also create unforeseen consequences for the health of the DAO and the process of governance. Secondly, while the BPT cover choice helped stimulate demand for MPL and, more importantly, built liquidity depth in the early days, it can be argued that this is no longer as valuable given the positive evolution of the project and the shifting tokenomics profile. Thirdly, the BPT cover choice leaves pool delegates potentially exposed to frustrating impermanent loss.

Given the above, it makes sense to require that first loss positions be denominated in the pool asset or in USDC, ETH, or wBTC. While the door can be left open to add in other assets as cover, it is important to balance this with the engineering resource requirements of building liquidation engines. Thus, starting with USDC and either ETH / wBTC (novel yield bearing opportunities for these assets) could make the most sense.

Developing a Tranche-based Model

Tying the three subsections above together, we could see the establishment of a tranche-based system, whereby the pool’s loaned out capital is segmented into a senior and junior tranche and a fair difference in lending economics exists between the two tranches in order to adequately represent different risk profiles. In this system, the cover from the pool delegate would come out of the junior tranche. Pool delegates, depending on the size of their pools, would have certain cover targets to hit as a percentage of the pool. Part of it would be required up front in order to have an immediate incentive effect, while the rest would flow in from future revenues earned on the pool (more capital efficient for pool delegates). Over time, these revenues would be used to buy out other junior lenders at par. All in all, pool delegates would be earning attractive rates of return on their capital acting as junior lenders while remaining aligned.

Conclusion & Next Steps

Note, the fundamental goal of this post is to ensure that Maple can scale in a sustainable and thoughtful manner, protecting its brand and trust among institutions and individuals, while also driving the ability to add new pool delegates and increase the platform’s exposure to new creditworthy opportunities. We thus see it as a conducive way to catalyze discussions across several potential avenues for change.

As importantly, Maple Finance can be seen as a pioneer, establishing practices and processes ahead of (1) other DeFi platforms, (2) CeFi lenders in the space, and, even, (3) TradFi lending firms. Maple Finance is on the frontier of DeFi-powered institutional credit and lending - it is important to not lose sight of this as both an achievement but also as its forward-looking mission.

Regarding next steps, we propose the following path for the community to take:

  1. Openly discuss the contents of this post, share views on each of its components, and gather feedback.
  2. Home in on areas where there is mutual desire for action and focus on further discussion there.
  3. As a general consensus for making a specific change emerges from the community, we would collectively move forward to enacting a new proposal aimed at providing a solution to vote on.

As always, we welcome any feedback and are open to incorporating new suggestions where needed.

Disclosure: This proposal is authored by ParaFi Capital LP (together with its affiliates, “ParaFi Capital”). ParaFi Capital is an investor in the MPL token and may from time to time provide liquidity to the protocol. This content is provided for informational purposes only, and should not be relied upon as legal, business, investment, or tax advice. References to any securities or digital assets are for illustrative purposes only, and do not constitute an investment recommendation. All information contained herein is obtained from sources believed to be accurate and reliable, however ParaFi makes no representation, express or implied, as to the accuracy of any such information. All expressions of opinion are subject to change without notice, and ParaFi does not undertake to update or supplement this post or any of the information contained herein.

1 Like

Thank you @ParaFi_Capital for this thoughtful governance post.

Maple and ParaFi are aligned on the 4 principles outlined in the proposal.

  1. Pool Delegate and Lender alignment.
  2. Scalability & Capital Efficiency.
  3. Flexibility & Realism.
  4. Long-term Vision.

Finding the right balance between those 4 principles increases Maple’s chances of long-term success. Pool Delegates perform a critical function on the Maple protocol and aligning their incentives with Lenders is key to solving the classic financial principal-agent problem.

If the Pool Cover size requirement is too high, it increases the barriers to entry. This will significantly hinder protocol adoption, growth and capital efficiency. Keeping flexibility and making thoughtful changes to the protocol are critical to maintain momentum.

To build on that, one important difference between the TradFi system and Maple is the transparency of counterparties and the loan book. Recent market turmoil has validated that, where CeFi Lenders faced losses north of $2 Billion, there was only a single default in one of the Pools on Maple, representing just over 1% of the $900M in outstanding deposits and loans at the start of June 2022. This transparency is a major source of accountability for Pool Delegates and when combined with Cover Reserve provided by PDs creates an aligned incentive model.

Summary of ParaFi governance Proposal

  1. Increasing “Skin in the Game”. Implement a tiered Pool Cover structure, where as the pool size scales, the absolute dollar value of first loss capital increases whilst the percentage amount (i.e., Min % Cover) decreases. (An exception for Pools where the Lender and PD are the same party).
  2. Balancing Requirements with Credit-built Relationships. More time with a strong lending history on Maple can lead to more capital efficient Pool Cover structures for PD’s.
  3. Ensuring Resilient First Loss Capital. Changing the BPT model for Pool Cover to a USDC-only model. This prevents the MPL price from spiralling down and provides a more stable Pool Cover reserve.
  4. Deploying a Tranche-based model. Establish a tranche-based system with junior debt and senior debt. The PD would provide junior debt as Pool Cover to increase capital efficiency. The PD would have a cover target %, if the current Pool Cover % is lower, the rest might flow in from revenues earned on the pool. Again, making it more efficient.

Maple values the thoughtful post and is open to exploring solutions with ParaFi and the wider community. Below some initiatives Maple is currently working on that cover some of the proposals outlined by ParaFi.

Maple 2.0

Q4 of this year Maple aims to release Maple 2.0. This will be a significant protocol upgrade that will make Maple more modular enabling more flexibility.

The launch of Maple 2.0 will have three consequences for Pool Cover in its current form:

  1. BPT will be removed as the asset for Pool Cover ****and replaced by the pool lending asset. If it is a USDC pool, this asset will be USDC and with a WETH pool it will be WETH. This assures increased stability and predictability in the size of Pool Cover, allows for easier liquidation and prevents the price of MPL from being impacted during a liquidation
  2. Pool Delegates will be the only parties able to provide Pool Cover.
  3. Maple 2.0 comes with a minimum requirement of Cover per pool set by the protocol for each Pool Delegate. This will include the ability to retain a percentage of the fees earned by the delegate for this purpose.

Post Maple 2.0 Launch

Due to the modularity and increased flexibility of Maple it will become easier to add or modify features of the protocol.

The potential implementation of a debt tranching mechanism will be assessed by Maple post Maple 2.0 launch. This is a similar concept to the concept that ParaFi suggested in the governance post, where Lenders can provide capital into different tranches within the pool. Each tranche is protected by the tranches below it in the event of default and enables Lenders to assess and deposit into different options with different risk/return profiles.

Next Steps

Maple puts a lot of value into feedback and suggestions by users, token holders and the wider community. Feel free to leave a response under this thread with new ideas or thoughts that might help Maple to find the right solutions to increase PD & Lender alignment, increase capital efficiency, maintain flexibility and scale towards Maple’s long-term vision.

Many thanks for opening the discussion on very relevant themes with a well-thought framework. Some comments below, starting with the easiest.

Pool cover asset

Agree with the arguments to remove BPT as a cover asset.

However, would stick exclusively to each pool’s lending asset (as proposed by the Maple team), to avoid the scenario of a diminishing USDC cover for a wETH denominated pool during an ETH bullish cycle, or vice versa.

Understand that the project to use xMPL also as pool cover will be abandoned.

Skin-in-the-game vs capital efficiency balance

Think it is worth keeping in mind that capital availability is not equivalent to operational proficiency (as seen with CeFi lenders) and the latter will become a key bottleneck if cover requirements become too stringent on a limited supply of quality credit teams willing and able to take the plunge into Maple’s platform.

Looking at how this played out for ETH staking protocols, Rocket Pool’s growth has been limited by bond requirements to node operators while Lido achieved exponential growth by relying solely on governance approval (based on qualitative assessment of node operator candidates).

Risk profile for unsecured lending is different from staking, and pool delegates should hold a relevant first loss tranche, but it is also different from securitisation (and CLOs), since the establishment fee represents less than 25% of pool delegate’s revenue (and more than 75% is linked to the pool’s continuing success in generating interest income).

Another key point to bear in mind is the need to supply relevant first-loss protection to senior pool investors, regardless of who is providing it. If we assume ~8% cover target, it will likely be impossible to reach if cover is limited to delegates (currently supplying 2-2.5%), so would strongly argue against removing the possibility for protocol users to provide cover alongside delegates (as I understand is going to be implemente in Maple 2.0).

Looking at a simplified simulation here, that assumes a positive mid-term scenario with delegates managing ~$500M pools and 90% utilisation, an 8% cover would require $36M. If we assume that a delegate would max out at 3% ($13.5M and almost 2x its annual revenue) this would result in an unattractive first loss cover for investors.

To balance the above, some (strawman) adjustments on the framework proposed could be:

  1. A requirement for pool delegates to top up cover provided by first loss investors with 25-50% of their revenue until it reaches 8% of utilisation. Delegates with a solid track record or capabilities pitch would see investors filling up the cover requirement themselves and would be free to take home all of their revenue.
  2. The above would motivate delegates to design (and pitch) attractive first-loss tranches. Introducing further tranching possibilities (like a mezzanine tranche between the first-loss and the senior) would add to delegate’s toolbox and is already in TrueFi’s roadmap.
  3. To avoid the scenario of successful delegates posting no cover, a $1M or 1% of utilisation minimum could be considered (again to be topped up with 25-50% of annual revenue).
  4. To keep the focus on qualitative screening of delegates, a formal recommendation on all submissions would be required from a committee made up by the Maple’s biggest lenders (along the lines of Lido’s Node Operators Sub Governance Group).

Off-chain wrappers

Looking at growth triggers, think a (huge) one would be to facilitate TradFi institutional credit investor adoption through a real-world conduit (SPV or similar) that allows lenders not yet prepared to deal with digital assets to access the credit sub-asset class through a familiar wrapper than they can show in their IRL balance sheet, along the lines of Compound Treasury.

This can also be (better) achieved by integrating with size custodians.

Happy to help out thinking this through and sounding with TradFi institutional credit investors.

Hope (any bit) useful.

bellonoff @kontxters (angel group with TradFi background).

1 Like