Product Discussion: NFTs as MakerDAO Collateral

This post gives an overview of opportunities for Maker to begin serving the NFT lending market.

Market Opportunity

The NFT market has seen incredible growth over the past year. Sales activity has dipped somewhat over the past 3 months, but the NFT space retains huge user attention, narrative tailwinds, and overall momentum. Despite this, NFTs are still second class citizens within the defi space with sparse lending and collateral support. I believe this presents an opportunity for Maker to capture market share in a lending sector likely to see high growth in the coming years.

Source: The Block

Large Addressable Market

The aggregate market capitalization of top NFT series (eg Art Blocks, Cryptopunks, etc) can range into billions of dollars, rivaling some of our other key collateral assets. If Maker captured even a few percent of the floor capitalization of top collections, this could allow for minting tens of millions of DAI.

Source: nftpricefloor.com

The below tables show some simulations of potential DAI generated and annualized revenues based on varying risk and demand inputs. Calculations are in this sheet.

Bear case:

Base case:

Bull case:

Strong Demand Potential and Tax Treatment

NFT owners have seen rapid price appreciation over the past year, with many sitting on large unrealized capital gains (which in some jurisdictions like the United States face higher collectible tax rates vs regular cryptocurrencies). The desire to avoid realizing taxable gains could drive significant borrowing demand among high end NFT owners. These individuals may also wish to rebalance their portfolio, increase exposure to other assets, or generate yield through farming activities, without selling their existing holdings.

Low Competition

NFT backed lending is currently not well served by the market. Cream Iron Bank underwrote a $3 million loan to PleasrDAO, but this was structured as a one time transaction and has not yet been repeated. Cream’s current struggles with their main money market’s insolvency may also hamper further NFT loan underwriting.

A few other platforms are also launching NFT lending mechanisms, including PawnFi, NFT Fi, jpeg’d (a NFT collateralized stablecoin project), and Drop Loans. Most of these additional platforms are still in development or private beta and are not accessible to the market.

NFTX offers tokenization of NFT series, and has recently created a pool on Rari Capital to allow borrowing against NFTs (currently just floor CryptoPunks). But this NFTX pool has relatively low available capital and currently only supports PUNK NFT tokens. Furthermore, users must deposit their NFTs into the NFTX vaults before being able to borrow via Rari, which means they may not be able to retrieve the same specific NFT they deposited and may also end up realizing taxable gains on their NFT (depositing NFT to NFTX vault may be considered a taxable exchange).

Natural Liquidity Advantages

Competing lending protocols need to bootstrap both borrowing demand from NFT owners and stablecoin supply from lenders. This limit’s available liquidity for borrowers, and large loans could have a significant market impact on interest rates (or impact on price peg in the case of NFT backed stablecoins). Additionally, for term lending based mechanisms lenders are forced to periodically roll over their positions, which further fragments liquidity and leads to worse user experience.

As a stablecoin protocol, Maker is not impacted by these challenges. NFT vault users would be able to mint DAI and access all of the market’s available liquidity, without additional lender friction or liquidity fragmentation. Maker would be free to offer either fixed term or revolving credit facilities without impacting DAI holder UX.

Trustless Collateral

High end NFTs are (usually) permissionless and crypto native assets. This is a benefit to Maker as we wouldn’t need to accept additional trust assumptions (eg. custody risk with WBTC, or credit and blacklisting risk with USDC) while lending against these assets. The value of these assets is generally also intrinsic to the item, and doesn’t depend on performance or actions of a DAO or other organization which helps reduce governance risk.

Challenges to Serve This Market

NFTs are less liquid than fungible tokens; they trade less frequently, often have higher volatility, and have internal variation based on relative rarity or aesthetics. This likely requires conservative risk parameters (e.g. high liquidation ratios).

The lack of consistent trading volumes, particularly for non-floor price assets (the more highly desired pieces in a set, or 1:1 pieces), makes it difficult to construct reliable oracles for liquidation. Even for floor assets from top sets, liquidity is potentially vulnerable to manipulation.

Certain highly valued NFTs are also unique, representing either top rarity pieces from a set or 1 of 1 pieces. Because these assets trade only very infrequently, it is impossible to use a regular oracle mechanism for live pricing. Appraisals and other valuation methods may be needed instead.

Projects like Abacus are working on sophisticated NFT appraisal mechanisms to help address these pricing difficulties. These mechanisms are still nascent but could improve appraisal accuracy in the future.

NFT Lending Frameworks

There are a few potential paths to serving NFT backed loans at Maker, each with different addressable market segments, advantages, and risks. We break this down into fractional token loans, floor asset loans, repurchase agreements, and loans to other NFT lending projects.

Fractional Token Loans

This type of NFT lending most closely resembles MakerDAO’s existing vault system. Maker would accept applications for collateral onboarding for liquid NFT set tokens (eg. the NFTX vault tokens for top sets such as Crypto Punks or BAYC). Each token represents the right to redeem an NFT out of the underlying NFTX vault, so in case liquidation is necessary keepers would be able to take advantage of both the fungible token’s liquidity pools and underlying NFT market.

Buyout based tokenization platforms such as Fractional Art may be less suitable for collateral onboarding, because liquidators would only have assurance of using any liquidity of the fungible token and may not be able to trigger a buyout of the underlying NFTs if they have a minority share of tokens.

NFTX tokens are traded on Sushiswap (some with relatively high liquidity), so it may be possible to construct a price feed using Sushi’s TWAP functionality even though the tokens are not currently available on other trading venues or CEX. And the planned rollout of Floor DAO (an Olympus style fork focused on accumulating NFTX tokens and LP positions) may significantly increase liquidity and market size for some of these NFT fungible set tokens. NFTX also plans to roll out single sided staking which would allow users to earn yield on NFT exposure without impermanent loss risk. And introduction of scalable collateralized lending could also help improve adoption of these fungible NFT sets.

Benefits:

  • Easier to integrate into existing MakerDAO systems
  • Relatively good liquidity (including both NFT and fungible token liquidity) - potentially less risk and more competitive borrowing terms

Drawbacks:

  • May have tax drawbacks for users (if depositing NFT into NFTX is taxable event)
  • More potential for competition from other lenders in the short term - eg Rari Capital already offers these services with several supported stablecoins
  • Relatively low market cap for NFTX vault tokens

Competitors:

  • Rari Capital Fuse
  • Other lending or stablecoin platforms could easily offer support

Floor Asset Loans

For floor asset loans, MakerDAO would directly accept NFT tokens as vault collateral and offer standardized valuations and borrowing limits based on floor price and liquidity of a set. Vault owners retain ownership of their exact NFT items (instead of a share in a pool of NFTs as per NFTX and similar tokenization platforms). This offers some intangible benefits to vault users (e.g. they could continue displaying their NFTs as an owner even while borrowing from Maker), as well as potential tax benefits of not needing to exchange their NFTs for fungible set tokens.

Due to NFT trading dynamics and the ability to manipulate average prices with private and wash sales, it wouldn’t be possible to use a typical price oracle for managing vault collateralization levels. Instead, Maker would conduct periodic appraisals of floor price items from each supported NFT series, and then update prices and risk parameters manually through governance. This could also potentially be handled through an “optimistic” approval flow where a mandated entity is given update authority on NFT prices, and Maker governance would have veto authority within a delay period. As an alternative, Maker could use a structured price feed including NFTX floor token liquidity and NFT sales, with additional safeguards against outlier trade prices or attempted manipulation.

Benefits:

  • Retains tax benefits of collateralized lending (no conversions or taxable events)
  • Relatively high liquidity, better terms and less risk for Maker
  • Vault users still have a claim on individual NFTs, so they can continue benefiting from utility value (eg. using a pledged Cryptopunk as their profile picture)
  • Large addressable market

Drawbacks:

  • Requires work to adapt Maker system to allow NFTs and individualized collateral accounts for vaults
  • May increase governance overhead to make price and risk parameter updates
  • Less capital efficiency for non-floor items within a supported series

Competitors:

  • Startup NFT lending projects
  • Genesis Trading and other centralized lending desks

Repurchase Agreements

For truly unique NFTs (eg high rarity items from NFT series, 1 of 1 art pieces, etc), or larger collections of NFT items, Maker could offer tailored lending facilities structured as pawn shop / repurchase transactions. Essentially, Maker would purchase NFTs from a collector in exchange for DAI (essentially granting a DAI loan), while giving the collector the option (but not obligation) to repurchase the same item(s) before a set future date for the original sales price plus a borrowing fee - this could include an up front origination fee as well as a continuously accruing interest rate. The initial purchase price would be only a fraction of the collateral’s appraised value, so the borrower will have a strong incentive to repurchase their assets (repay their loan) before the end of the loan period unless the NFTs face significant fall in value. If the borrower does not repurchase their assets within the term, Maker can sell the NFT to recover funds (and potentially even earn a liquidation surplus).

This mechanism is appealing for borrowers, who pay predictable financing costs and have no risk of liquidation from NFT price volatility before the end of their loan term. It also has advantages for MakerDAO, with no need to maintain price oracles for linked collateral - only an appraisal and risk assessment are needed to initiate the repurchase deal.

Benefits:

  • Can serve non-floor assets, 1:1s, heterogeneous collections, etc
  • Borrowers have no liquidation risk from valuation changes during the loan term
  • No need for up to date pricing information

Drawbacks:

  • Less liquidity and trading data - potentially more risk of loss to Maker and less favorable risk parameters for borrowers
  • Requires governance and risk work to review and approve each individual deal
  • Fixed borrowing rates create duration risk, which would negatively impact Maker’s profitability in a rising interest rate environment
  • Liquidating collateral when loans are not repaid could require more hands on management and marketing to find the right buyers - using a typical dutch auction like other Maker collaterals may not work very well for these sort of items

Competitors:

  • Cream Finance
  • Startup NFT lending projects
  • Genesis Trading and other centralized lending desks

NFT Lending Platforms

As more NFT lending mechanisms come to market, MakerDAO could also pursue an “indirect lending” strategy, where market professionals and specialized NFT lending protocols are empowered to price and manage collateral pools, with Maker providing senior financing. This could be accomplished either by accepting LP tokens from the NFT lending platforms as collateral, or by implementing a D3M type mechanism to provide liquidity directly into lending pools.

Benefits:

  • Potentially lower risk to MakerDAO (Maker may have seniority of repayment or additional capital buffer versus the issuing platform)
  • Less in house expertise or review needed
  • Can potentially serve both floor, fractionalized, and non floor assets

Drawbacks:

  • Greater governance risk - Maker relies on underlying platform to maintain appropriate risk management

Competitors:

  • Stablecoin protocols such as Fei, Frax, and Abracadabra
  • Several projects are working on NFT pricing algos to move away from floor pricing, these orgs could be competitors or participants in Maker lending mechanism

Business Strategy: Serving Large Volume Users

With access to the lowest possible cost of capital through the DAI holder base, Maker has a unique opportunity to form the foundation of the NFT lending market, similar to our role within crypto lending markets.

Just as some of Maker’s largest minters for fungible collateral assets are centralized lending platforms (e.g. Nexo and Celsius have large ETH and WBTC backed vaults), there is also an opportunity to provide pass through NFT financing to centralized lenders. For example, Genesis Trading has recently announced their entry into the NFT lending markets with a $6 million loan to an NFT focused VC fund.

As these CeFi lending businesses develop, they could engage in rehypothecation of client collateral - they would take their clients’ pledged NFT assets, and provide them as collateral to MakerDAO in order to finance their operations and lending business. This is a potential win-win for all parties:

  • CeFi platforms get access to lower cost capital from MakerDAO
  • Savings are partly passed on to clients in the form of lower rates, and CeFi platforms can also offer additional features like higher leverage or cross margining with clients’ other assets
  • Maker gets reliable, large volume vault users (CeFi platforms) who have more sophisticated vault management capabilities and backstop capital

In theory, DeFi NFT lending platforms could also rehypothecate collateral with Maker in a similar way, although this may be more complicated from a technical perspective. Maker’s NFT lending strategy should take account of these external platforms to ensure we can meet the needs of high volume, high impact users.

System Requirements

Segregating User Collaterals

Vault users will expect to receive the exact NFTs they deposited back after they repay their vault debt. This may require modifications to Maker’s collateral adapters to ensure that users are able to withdraw the correct assets, rather than just the correct number of tokens as is the case with fungible collateral assets. Possible solutions include segregated collateral accounts for each NFT vault owner, or mapping between vault IDs and token IDs for non fungible collaterals.

Price Data and Oracles

Maker needs reliable price information for NFTs to be able to offer vaults. This information would be used to determine maximum credit lines, trigger liquidations, and settle vault debt proportionally against collateral in the event of global settlement. There are a few possible approaches to evaluating and relaying price data to meet these needs.

First, Maker must choose between pricing NFTs directly in USD, or pricing them in ETH and then making the final conversion to USD price in combination with our ETHUSD price feed. Considering that NFTs have historically had strong correlation with ETH, it probably makes sense to price NFTs in ETH and convert to USD, but this decision should account for changing market dynamics and correlations.

After determining reference currency, Maker would also need to choose an appropriate oracle mechanism. Some potential options are listed below.

  • Appraisals with multisig oracle

A team or set of teams within MakerDAO will carry out periodic appraisals of supported NFT collateral assets. These assessments will consider NFTX / fractionalized token trading activity as well as direct NFT trading volumes. Outlier prices and private sales can be manually filtered out of these figures.

The price updates will be queued into a timelock by an appointed multisig, but Maker governance would be able to cancel the price update, submit an alternative price update with no time delay, or remove multisig authority over the price oracle. This is essentially an optimistic security pattern - the multisig is assumed to provide valid data, but Maker governance can step in as a backstop if necessary. This likely requires at least 2-3 day timelock delay to give Maker governance time to challenge any invalid price updates. The high latency in turn will require relatively high collateral ratios.

  • Sushiswap TWAP oracle

For NFT sets covered by NFTX floor fractional tokens, Maker could utilize Sushiswap v2 liquidity and corresponding time weighted average price (TWAP) oracles to determine fair market value of assets.

This mechanism should only be used for sets with sufficiently deep liquidity on Sushiswap (Cryptopunks are probably the only set to meet this liquidity criteria currently). And the time period used should be relatively long - Maker faces arguably higher risk from short term price manipulation (from potential pump and dump or liquidation hunting attacks) versus delayed price updates.

Liquidations

After a vault is liquidated, Maker typically returns any remaining collateral to the vault owner. This may not be possible for NFT backed vaults (excluding fungible tokenized options like NFTX tokens), as the entire asset would be sold during the liquidation process to a single bidder. Instead of receiving remaining collateral, vault owners would receive any DAI proceeds from the liquidation that exceeded their outstanding debt plus liquidation penalty. This would require modifications to our liquidation mechanism. Alternatively, Maker could simply claim 100% of liquidation proceeds (leaving none for vault owner), but this could scare away potential users.

Global Settlement

In the event that Maker governance or MKR holders trigger global settlement, all outstanding debts are immediately settled against a proportional share of vault collateral assets. This is challenging for NFT vaults, as the collateral assets are not naturally divisible to allow for splitting between vault owners and DAI holders.

NFTs could be preemptively fractionalized as an ERC20 while held in Maker vaults, which would allow for proportional settlement during an emergency shutdown. The fractionalization schema should use a modified buyout or redemption mechanism to ensure DAI holders can force a sale even if they hold only a minority of outstanding fractions and lack coordination. See explanation of RICKS for one solution to this problem.

Alternatively, vault owners could be given a certain period after global settlement (for example 1 week) to fully repay their outstanding NFT backed debt, or the item would be put up for auction with all proceeds benefiting DAI holders. This allows NFT owners the opportunity to fully repurchase their item at cost, while also simplifying the DAI holder redemption experience - DAI holders could receive a single additional token representing their claim on all NFT vault repayments or auctions, rather than a series of fractional tokens for individual NFTs. On the downside, this could result in a large number of NFTs coming to market at once in the aftermath of global settlement.

Example:

Alice and Bob both have open vault positions secured by cryptopunk collateral. The oracle value of floor punks is $200,000, and Alice has borrowed 30,000 DAI while Bob has borrowed 20,000 DAI.

Global settlement is triggered during a sharp crypto market crisis, beginning a one week period for Alice or Bob to repay debt and retrieve their collateral. During the crisis, the fair market value of a floor cryptopunk has fallen to 25,000 DAI.

Because the value is now lower than outstanding debt, Alice decides not to repay her loan, while Bob fully repays his debt with 20,000 DAI. Alice’s collateral is auctioned for 24,000 DAI, yielding a total of 44,000 DAI to be redeemed by holders of the tracker token.

In an alternative scenario where the price of cryptopunks doesn’t fall so much, or if collateral represents non-floor items, it’s likely that repayments or auctions would generate enough funds to fully settle debt without losses.

Go to Market

Here are some ideas for the simplest and fastest ways we can begin serving this market:

  • Provide DAI liquidity to NFTX Fuse lending pool (pool 31)
    • Accepts NFTX PUNK token with 200% liquidation ratio
    • Uses Sushiswap TWAP oracle
    • Also offers lending against NFTX governance token (out of scope for NFT lending goal but doesn’t introduce too much risk)
    • Additional NFT sets may be added in the future as liquidity increases
    • Could be structured as fixed DAI commitment to bootstrap market, or as D3M with interest rate targeting
  • Offer OTC repo lending against high value NFT collections
    • Deals can be made on case by case basis similar to institutional vaults
    • No continuous price data needed, no liquidations if loan is repaid within agreed term
    • High potential visibility and PR value
    • Downside: this approach requires a lot of workforce and governance attention, not as scalable
  • Offer cryptopunk floor asset vault
    • Requires oracle mechanism for floor cryptopunks (multisig, Sushi TWAP, or other)
    • Requires changes to collateral adapters and liquidations
    • Larger total addressable market, plus certain tax and utility advantages versus NFTX token collateral
17 Likes

Thank you Monet for the research and time you put into orchestrating this article. Very informative and well balance.

I was wondering if you have put thoughts into Moxie Marlinspike NFT concerns that the metadata is often stored at a centralized endpoint, and there’s no hash of what is stored? Meaning if the centralized endpoint goes down, and in the future someone acquires that metadata (NFT), they can freely change the metadata to whatever they want.

2 Likes

This is an important point, and something we’d need to determine for each supported collection or asset.

I know Cryptopunks and Artblocks are both fully decentralized projects, without this rug pull risk. Decentralized recordkeeping for linked NFT media should probably be a pre-requisite for becoming MakerDAO collateral.

5 Likes

I am personally astonished at the size of the NFT market. Whether NFTs have long-term staying power or not, the exploration of options for price oracles seems generally applicable to other markets. Great stuff!

2 Likes

Personally I don’t see a future for such a speculative market, as there are several cases where the price of a NFT or a whole collection can be inflated.

I think Maker should first concentrate on accepting Uniswap v3 collateral that is NFT.

This would give time for the NFT market to mature perhaps.

4 Likes

What do you think of JPEG’d DAO? Looks like they are 3-steps ahead into doing NFTs as collateral types. Have you looked into it?

1 Like

Thank you @monet-supply for putting this report together. I think many of us are astonished at how fast this space has grown and IS growing.

NFT Pawn Shop is a very intresting idea here.

My concerns here for Maker is the limited potential for DAI to be minted and the real limited amount of DAI revenues.

When I think about appraisers, oracles and management for adding these and how fast this market is moving I think it might be hard to isolate a couple winners here that could grow their markets and borrows sufficiently. I keep wondering if some communities wanted to guarantee Maker oracle, development and operational costs for such an experiment for a year (or even 1/2 of it) I might be more interested.

The other real issues here is this idea that we have NFTs priced in ETH. So we have a pricing model that is strongly correlated to two highly fluctuating base prices (the NFT in ETH and ETH in USD). Given the currently low liquidity I could easily envision oracle attacks. The other issue is going to be maintaining keepers on these. Each expansion into a new asset class stretches keepers to some extent and the dust levels are already high on many vaults (and not likely to go down). Precisely who is going to have 50-100K or perhaps 250-500K of NFT collateral to borrow given increasing dust limits here and will Maker end up being stuck with vaults of low size not being able to liquidate as dust limits increase?

I think it is important to follow this market but at least on L1 with Maker and real dust limits here may just not make this NFT vaults practically feasible.

Other point. If I have to look at applying limited MakerDAO resources to any single project rather than spreading them to two or more. I keep coming back to a MakerDAO manhattan type projec to get RWA DAI minted to 1B ASAP vs. spending any time on anything else. MakerDAO needs to keep an eye on these things to be sure, but we need a laser focus on getting the next 10B DAI here in large chunks not 1-10M at a time.

2 Likes

By only accepting collections that have deep liquidity and visible longevity i.e. Punks, Art Blocks, BAYC you can minimize downside risk.

There have been a few NFT lending markets for those pieces already for about ~6-8 months and they have been successful. Seems like a smart idea seeing how they hold/accrue value.

2 Likes

Personally, art has always seemed subjective to me, it doesn’t seem to me to be the right path to take.

And even less so in such a volatile era, where you can tell that we are more in an era of FOMO in the NFT market.

I would be wiser to work on the fly while the FOMO is decreasing.

As you said it’s better to work on recognised works and collections, I don’t think you get to see NFT tacos or NFT arepa to be collateral in Maker.

2 Likes

“Be fearful when others are greedy, and greedy when others are fearful.” :blush:

4 Likes

I have a doubt, in this case, we as Maker will work with NFTs kindda like with RWA? Like the example with NFTX PUNK token, we will work with tokens that represents the value of the NFT right? So the idea will be searching for the best dApps that offers this tokens? Sorry if I didn’t understand it quite good.

1 Like

Glad to be having this product discussion because NFTs as collateral seem like a logical future for MakerDAO. Users are already looking to turn these stagnant assets into productive ones through the magic of DeFi, and there’s no better place to lead the charge than MakerDAO. This can also evolve RWA into DWA (Digital World Assets) and an extension of Clean Money as digital assets are much better for the environment.

Coincidentally, last night I listened to Tetranode on UpOnly, and they brought up JPEG’d, so I did some digging. They are definitely ahead of the curve, but MakerDAO is more reputable than some new and somewhat anonymous team. They are doing what @monet-supply described as Floor Asset Loans, in which they are tracking the floor price to provide a base value of NFTs, so users don’t borrow more than the minimum value.

They are doing this by using a specialized oracle from Chainlink. It uses a “Time-Weighted Average Price (TWAP) of both sales and floor prices to create a blended price that will be used to value floor punks.” This is actually an intelligent approach because it filters out outliers and wash-sales to provide a reliable floor price that’s resistant to manipulation. They also have “safeguards” in case of liquidation, whereas a user can repurchase their NFT from the DAO with a penalty.

I’ll reiterate that this is an excellent opportunity for new use cases that we should explore in-depth. Also, from a business standpoint, there’s probably a nice product-market overlap between Maker users and Punk hodlers. Plus, I’m sure these users would rather receive good ol’ liquid DAI for their NFTs instead of some immature meme-stablecoin comically named “PUSd”.

3 Likes

Thanks for providing more details of Jpeg’d Mighty. Very interesting.

I believe in early 2021 during an G&R call I had foolishly brought-up the idea of using NFTs (back when only cryptopunks and haskmasks were the talk of the town) as collateral, and one of the major obstacles was the Oracle price feed of such. But today, I do believe there is more tooling available like rarity.tools, and I wonder if this one tool along with others, can help folks think of NFTs as a serious collateral type… We shall see.

3 Likes

Is definitely an interesting topic for for discussion. You have some very good points, however I think the “how to” is the detail. The use of art and collectibles in pawnshops is really common, so in principle it could work.

Currently the NFTs market is growing as well as its volatility, even one of the most stables nft games like axie does not scape from this fact. You can make very good profits, but with a good system capable of calculating the changing value they have on the market, perhaps a new form of oracles or as you said, a group in charge of monitoring the prices of these collaterals. I think it´s a great idea, but it´s still lacking.

1 Like