The future of real-world-asset composability in the Gensler era
Product lessons RWA builders should take away from the last 12 months as they adapt to 2023's regulatory setbacks
In a pre-2023 world where web3 peacefully coexisted with major global regulators, tokenization of real world assets (RWAs) seemed to make more and more sense with each passing day, for both the real world and DeFi.
Tradfi securitization is inefficient, error-prone, and opaque. Fiat-backed stablecoins seemed to offer a much cheaper railway for securitization than the finance industry.
Over time, blockchains could bring significantly more transparency to these products at all levels than tradfi does today.
On the DeFi side, L1 cryptocurrencies can be thought of as digital equivalents of Hong Kong or Switzerland — digitized communities which, besides collecting taxes on activity internal to that network, can accrue significantly more value by by servicing assets far beyond their borders. RWAs offered the fastest way to scale this value proposition.
DeFi also needs a deeper tethering to the real world for both risk-diversification and branding reasons.
Risk: It does nobody in DeFi any good for all their digitally-native assets to be high-volatility cryptoassets. Lots of money in crypto would like to “come ashore” so that crypto can gain exposure to more diversified cash flows.
Branding: crypto is tired of being seen as a purely speculative instrument of degenerate gamblers, and wants to be able to point to real differences it makes in the world, underwriting virtuous activity.
On the technical side, RWAs, imo, have another 1-2 years of development to go before they can really leverage blockchain efficiencies.
The long road ahead to RWA DeFi composability
Getting a real world asset “on a blockchain” is theoretically easy. Many protocols already claim to do that: Centrifuge, Goldfinch, Credix, Maple (all on Ethereum), Provenance/Figure (Provenance is a Cosmos chain), Intain (on Avalanche), and others. You basically create an NFT that references an off-chain legal agreement, and assign a token or tokens to that NFT which rebase according to interest payments as they’re made on the off-chain debt, reflecting the asset’s change in value as it accrues more interest. The underwriter is required to upload an offchain data tape, usually on a monthly basis, that reflects changes in the underlying loan pool. This requires trust, and thus legal liability and/or first-loss exposure for the loan underwriter.
Different RWA protocols have already productized this, and results so far are respectable.
However, creating an RWA that truly leverages blockchain programmability is much harder.
For an asset to make the most of DeFi’s strengths — transparency and financial interoperability — it needs to be “composable” as opposed to merely “ledgered.” Composability means being composed of a number of standardized parts which foreign DeFi protocols’ smart contracts can consistently make sense of and independently interact with. The RWA gold standard will be a composable RWA that a foreign protocol like MKR or FRAX can lend against at an X% LTV. This asset would have enough decentralized on-chain inputs to create an accurate third-party picture of what’s going on with a securitized asset, such that the foreign protocol could create automated risk-mitigation rules to respond to, e.g., a rise in nonrepayment rates which degrades the first-loss tranche of an on-chain pool of securitized loans.
A composable RWA, e.g., a tokenized multitranche loan pool, needs every important element to live in its own smart contract and interact with the RWA’s other components according to a predefined set of rules. This is the only way that an RWA can be accepted as collateral by a yield-seeking DeFi counterparty like Aave or Compound or Frax: their own protocols need to be able to reference genuine, relevant values for different aspects of the asset which send different signals about the credit performance of that asset’s constituents.
This creates a number of underappreciated challenges distinct to web3:
Smart contract immutability: Because a smart contract can only be executed by network consensus, smart contracts are generally immutable once they’re deployed. Even if a SC has built-in upgradeability features, those features can only encompass changes which were envisioned when the upgradeability was created. This makes modifying SCs according to evolving product requirements much more difficult.
Contract design limitations: In TradFi, asset design is limited only by the imagination of the issuer and underwriter. The key parameters of a structured loan are negotiated over a period of months, and then the “coders” of structured credit (lawyers) draw up the contract negotiated by “product and engineering” (bankers / deal underwriters). At no point in the process does the lawyer say, “hmm, that provision of the loan sounds too complicated to include in the debt agreement.” But in a DeFi-composable loan, the parameters are limited to the components envisioned by the product lead and the smart contract engineer. So flexibility is much less.
In tradfi structured credit, mutability is unlimited, which makes interoperability virtually zero, but USD isn’t programmable in the first place and the structured credit market is highly fragmented and sophisticated by nature, so there is little upside to interoperability.
In the short term, the smaller wrinkles can be punted off-chain. However, the larger wrinkles (is it a bullet loan or an amortizing loan? what’s the grace period of the asset if a loan repayment isn’t made on time?)
Design complexity. TradFi folks in RWA are always rudely surprised once they understand how immutable smart contracts are, vs. the flexibility that structured credit borrowers & lenders are accustomed to in TradFi. This has major product implications. In practical terms, if you want to create a DeFi-composable pool of (say) auto loans, it needs to be programmed quite differently from a DeFi-composable pool of mortgages, or one of SMB invoices. In other words, your product is less scalable than you expected. Meanwhile, at the same time as an RWA’s TradFi folks are learning about scalability limitations on the technical side of things, the DeFi folks are learning that different kinds of loans, in fact, have very significant differences, and there are real reasons why private structured credit is so fragmented. This feels like something that has to be learned the hard way unless your product and engineering practices both have substantial experience building DeFi as well as TradFi products. Which is very rare.
Ethereum’s structural limitations. For a structured credit asset to live on-chain, its smart contracts need multiple interactions with the blockchain for every credit event, so that its individual components can interact enough to make sure the different risk parameters are being followed. This creates the need for multiple smart contract calls for every credit event. But the more SC calls you have, the more your ETH gas fees blow out, thanks to ETH’s never-ending problem of MEV sniping (frontrunning smart contract executions). This was a key reason in my opinion why Centrifuge, whose product vision seemed most influenced by DeFi composability, decided a long time ago that the next iteration of its Tinlake product could not exist on Ethereum.
Beyond these technical hurdles, there are some US-specific issues which may be underappreciated.
US legal restrictions on stablecoins. Even before the latest regulatory rampage by FDIC/NYDFS/OCC/Treasury/SEC, crypto had already been substantially de-banked from the US financial system: US banks, for example, were not allowed to hold stablecoins on their balance sheets (USDC was considered ‘more risky’ than a 30-year mortgage, lol), which made US banks’ scope of cooperation with RWAs extremely limited, and made RWA<>fiat interactions unnecessarily capital-inefficient.
An RWA product practice needs to find ways around these problems. (This area was one where I was particularly pleased with our internal results at END-Labs).
Provenance/Figure, among others, has invested a ton of time into lobbying US regulators to change this. As with anyone else who has tried lobbying regulators to be less hostile to crypto in the past 2 years, I doubt they’re happy with the ROI on their investment.
Compliance requirements of institutional investors. Beyond the under-penetrated universe of crypto-native customers (crypto-native hedge funds, stablecoin balance sheets, and DAO treasuries), the TAM growth opportunity lies in 3 distinct customer types for this product: Tradfi credit funds, fiat/crypto hybrid funds, and “experimentalists” (sovereign wealth funds, fiat megafunds, and fiat megabanks). All of them have very large regulatory attack surface & are very sensitive to regulatory issues. When the FDIC is calling up banks every single day (not an exaggeration) demanding their latest list of crypto counterparties, NYDFS nationalizes a solvent bank that happened to possess key RWA infrastructure, and Gary Gensler behaves as if slandering the legitimate part of crypto is his full-time job, it sends a message to all potential institutional RWA counterparties that touching the sector in any way is “not consistent with a culture of compliance.”
Building out offchain infrastructure to encompass the custody / title / RegFi event chain. Truly blockchain-native RWA is dependent on building out decentralized physical infrastructure networks (DePIN), in which oracles will have the ability to ping the location and/or status of physical objects in the real world, is a 5- to 10-year challenge. In the meantime, though, some protocols have invested in technology tailored to specific asset classes. I understand this to be a particular competitive strength for Provenance (re: home loans) as well as M^ZERO Labs (re: loan asset interactions with ECB lending facilities). Maker and Blocktower also supplemented the Centrifuge platform with a ton of time, legal fees, and brain damage into the offchain legal/custodial scaffolding of the Maker/Blocktower/Centrifuge facility, which is probably the most scaled onchain-offchain-hybrid RWA product operating at the moment.
In the short term, the RWA product offering will have to get over a “worst of both worlds” hump. RWAs will have to replicate all the legal/compliance/red tape headaches of tradfi products while at the same time adapting to the product inflexibility created by SC immutability.
Building an RWA marketplace seems to parallel building a standardized equity options in the 1970s, not that I was around then. There was huge, 2-sided demand for a protocol (a set of standards) as well as a venue (to pool liquidity) for equity options trading, which before then was an extremely fragmented, OTC, unstandardized market. It was obvious to all practitioners that electronification of that market would slash costs, increase volumes, increase liquidity, and grow the pie for all participants.
Private credit today feels like it’s in a very similar place, even though it’s harder to standardize. I think the first really successful DeFi-composable RWA tokenizer will accomplish something on the scale of what the CBOE accomplished for US options trading, standardizing and disintermediating what is today an extremely fragmented, OTC, bespoke market, lowering the cost of capital for mid-tier borrowers and reducing fees in the process.
Of course, the regulatory attitude towards options then was night-and-day from what on-chain RWAs face today.
Industry adaptations thus far
Different RWA protocols have taken different roads to optimize this challenging set of issues. Most RWA protocols aside from Centrifuge put off DeFi composability for later, in favor of optimizing UX (a sensible decision, given that most of them have been around for far less time than Centrifuge has, and thus aren’t as well capitalized). I think there’s a growing consensus that DeFi composability is the next big challenge for all these protocols.
Adaptation #1: Offshore from the US
This applies to borrowers, investors, and underwriters to whatever extent possible.
Adaptation #2: Offshore from Ethereum
Based on my own experiences with Centrifuge, I am very bearish on Ethereum’s long-term viability as a venue for DeFi-composable finance. And before you say What about rollups?, the Eth rollups / sidechains (ARB, OP, MATIC) are each controlled by a single multisig wallet. These multisigs are one court subpoena away from being forced to financially quarantine the assets of someone on their networks. This is an especially big risk for Optimism (SF-based) and Arbitrum (London-based). It seems like less of an issue for Polygon (based in India), but there’s no reason to give themselves the power to delete assets unless they wanted to be able to do it in the event of a government order. Also, India’s government is more hostile to crypto than most.
ZK rollups seem a long way away from scaling, and ZK engineering muscle is also extremely expensive.
I feel like Cosmos, whose number of recently-updated forks seems to have overtaken Solidity-eth’s but is well behind geth’s (the golang implementation of ethereum), and Avalanche both deliver usefully functional decentralization at a much more attractive price than Ethereum, and Cosmos has the largest buy-in from the global dev community.
Adaptation #3a: Start with extremely standardized assets like Treasuries
There are 2 ways to go with tokenized assets: go super-standardized, as Ondo has, or go for more specialized assets.
The standardized assets are much easier products, but also have much less to gain. Treasuries, for example, are already an extremely liquid and efficient market. Putting a blockchain wrapper around a highly standardized tradfi process will satisfy a short-term hunger for yield among DeFi depositors but is less likely to deliver a more efficient product than what’s available in TradFi today.
Adaptation #3b: Stick with more fragmented asset classes, but leverage greater engagement with TradFi structured-credit specialists
At END-Labs, even though we had plenty of internal credit expertise, there was a constant tug-of-war between tradfi and engineering over what the product should prioritize. A RWA product needs to be closely tailored to TradFi needs, but the TradFi-focused side of product leadership needs to have its expectations very clearly set on the flexibility side, or its product requirements will quickly diverge / it’ll quickly get discouraged.
TLDR
To graduate from basic “ledgering” to a product that is truly superior to what TradFi RWA offers, blockchain RWAs need to get around an unfavorable design flexibility-vs-composability tradeoff. To make the product more flexible, and thus more relevant to a wider institutional audience, tokenized structured-loan assets need to be “templatized” according to parameters frequently used by sophisticated credit funds. Creating these templates (probably 10-20 financial parameters per NFT asset) will require hand-in-glove participation and guidance by DeFi/tradfi hybrid funds with significant credit experience.
This composability also needs to be delivered in a package that doesn’t carry Ethereum’s unacceptable MEV-driven transaction costs, which other L1s have already addressed. Eth rollups are the easy shortcut (which we took when we decided to build on Polygon), but if I had to do it again, I’d accept the reality that the Ethereum community has prioritized rent over product, and does not underappreciate the systemic risks embedded in today’s rollups.