By Guest Author Sefton Kincaid (Cicada Partners)
The crypto credit crisis that loosely began with the collapse of Terra Luna in May, extended to the downfall of Alameda-FTX, and, arguably, remains to this day, should have taught market participants a number of important lessons. Viewed from the lens of first principles, the problems were not new or native to crypto, but the consequence of a nascent ecosystem with poorly implemented operational controls and risk management. Favorably, many of the solutions we believe to be fundamental to sustainable credit creation can leverage crypto building blocks in a more efficient and fair allocation of risk and reward.
Certainly, crypto lending has come a long way in recent years, from bespoke OTC lending and highly capital inefficient over-collateralized loans, but there are several areas that need to radically improve to build a more solid foundation for the next cycle and beyond. Moreover, the hypocrisy of most crypto lending not using crypto-native technologies like self-custody and smart contracts, is palpable. But there is a way forward…
The bull market proved we were not all here for the tech, but lets really use it!
DeFi Protocols enable a far more transparent user experience while improving the auditability of crypto liabilities. One of the constant battles our team had with crypto-native borrowers was identifying what centralized lenders were lending funds. We took serious issues with any company wanting to borrow capital but not willing to disclosure their counterparties and a detailed capitalization table. This must stop. The daisy chaining of risks only to find out, after the fact, that fraud was repeatedly committed could not happen with on-chain lending. Transparency reduces risk to lenders and drives down long term borrowing costs.
Auditing tools, like Credora, were early in their implementation phase during the prior cycle. We are big fans of the platform and view the quantitative asset and liability checks into any crypto-native borrower as a vital tool. However, accessing, using, and enforcing Credora integrations on borrowers are all equally important processes for a lending business that some lenders missed or struggled to implement. Creating industry standards for lenders are vital areas of improvement. We plan to more actively enforce borrower integration of such tools, while shortening loan tenors to facilitate more immediate recourse should borrower activities on the platform become incongruous with the underlying strategy.
Tracking tools to aid in monitoring wallet, exchange, and protocol transfers will become more important tools to reduce risk across the crypto-native lending space over the next few years. Tools like Elliptic and Nansen can help to track fund flows, detect misrepresentations/fraud, and monitor wallet activity. We are excited to integrate a rules-based flagging system by integrating off-chain financial information with on-chain fund flow data.
Moreover, smart contracts facilitate a trust-minimized approach to institutional lending. Lenders use self-custodial wallets to deposit funds into a smart contract, with a risk manager (”pool delegate”) operating a pre-defined mandate to lend funds to a borrower’s account. Trusting the risk management of a pool delegate is still needed, but importantly, the pool delegate never takes ownership of the pooled funds, removing the ability to ever misappropriate customer funds.
Incentives, Incentives, Incentives
Most CeFi (centralized finance) business models were doomed from the start. Take centralized models like BlockFi and Celsius as examples. These centralized businesses started out providing yield to customers by transforming bitcoin through the black box of a central institution into a yielding asset via the GBTC premium trade. After generating outsized customer growth on the back of BTC-yield products, the businesses were stuck in constant search of yield to drive customer growth, which then drove more lending growth, which in turn grew revenues. Consequently, these businesses began more as proprietary trading desks, with formal credit risk controls an afterthought. As the GBTC trade moved from premium to discount, these CeFi lending trading desks moved out the risk curve in search of yield. Had CeFi yield derived transparently on chain, with visible counterparties and transaction terms, fewer unassuming investors would be queuing in outstanding bankrupcy proceedings.
Sustainable lending models must start with better incentive structures from first principles, to better align interests of various stakeholders. An on-chain USDC lending pool has a defined risk mandate, separate from a BTC lending pool or another dedicated USDC mandate. Risks are isolated at the pool level, with customer funds deposited via self-custodial wallets.
The socialization of individual losses at a centralized firm level is problematic as well. While various DeFi pools were less than perfect in the fair allocation of loss, smart contracts are important tools to more efficiently tranche debt and liquidation in the event of loss. In stark contrast, CeFi institutions enabled reckless lending to clear ponzis but in opaque wrappers. Lenders were made to believe they were earning yield via over-collateralized lending, but losses on reckless lending were ultimately socialized across all lending verticals. If DeFi investors aren’t comfortable with the borrower or collateral or tenor or rate of a DeFi pool, lenders are free to withdrawal and reallocate. Importantly, a poor investment in one pool will not bleed into the solvency of another pool, which ultimately reduces risk to a lender creating a diversified strategy across a number of lending pools.
Risk Frameworks for Crypto-Native Lending Need to Radically Change
Responsible lending takes hard work. Crypto’s open, global, 24/7-traded marketplace adds unique complexity to crypto-native lending. Business due diligence takes time, rigor, and expertise. Unfortunately, many short-cuts were taken in the provisioning of credit. One such short cut repeatedly used was collateral.
Allow us to tangent into an illustrative example: Say a “sophisticated” institution is allowed to (a) borrow $100 USDC with $50 worth of bitcoin, buy $100 of bitcoin with the proceeds, (b) pledge $100 worth of bitcoin to get $200 USDC from another lender, and © proceed to swap for $200 more of bitcoin. What kind of system-wide restrictions on credit are lenders promoting? None. Such aggressive lenders justified such actions, inadvertently promoting reckless lending by constructing businesses that profit on collateral liquidation cascades. Even worse, these strategies were even promoted with illiquid, off-chain collateral like GBTC. In the above example, we took $50 of bitcoin and turned it into $300 USDC in debt and $350 in bitcoin assets. In this simplified example, the bitcoin is split across three different venues, with two lenders having recourse to 50% of their loan values worth in bitcoin, but with the majority of the assets ($200 in bitcoin) outside the recourse of lenders. Moreover, lenders had to trust borrowers to self-report their debt with other lender.
While the above is a highly stylized example, it happened, and it quickly demonstrates the need for unsecured credit risk frameworks to be integrated with collateral models as the use of funds borrowed is an extremely important driver to the creditworthiness of the borrower. No amount of collateral below 100% will change this. In fact, many of the largest and riskiest borrowers that are now in bankruptcy borrowed via collateralized deals. Collateral should not supersede good business practices, particularly collateral prone to bouts of illiquidity induced volatility.
Moreover, collateral models even above 100% collateralization need more human judgement and context, unfortunately. The multiple examples of lending to illiquid, reflexive, and or fundamentally mis-priced asset valuations highlights how collateral models need to be reworked. Critically, these issues weren’t specific to CeFi, with overcollateralized DeFi protocols like Venus, Mango Markets, and even Aave subjected to inappropriate risks as a result of poor asset selections for their automated lending pools.
So long as most crypto assets are risky, venture-like investments, with market pricing resulting from asset prices manipulated over short durations, market pricing as an indicator of asset valuation is not a sustainable approach. A more comprehensive and rigorous approach to collateral valuation is necessary to ensure the health and stability of the lending space.
In short, some lenders issued loans without proper underwriting given the view that pledged collateral made up for the risks. In contrast, lending businesses that performed proper operational due diligence and leveraged on-chain tools into a formal credit underwriting process would have identified bad actors early, resulting in materially lower default rates and preventing the systematic daisy-chaining of leverage.
Zooming out, crypto lending now has the potential to grow in a much more sustainable manner following the removal of many reckless market participants and business models. But without improvements in the key areas identified above, crypto-native lending is likely to repeat the prior cycle’s errors. People will be people, and cycles will come and go, but we are optimistic that the actual use of blockchain-native technology, well-designed incentive structures, and a focus on risk management will radically improve the lending experience for all stakeholders and bring more private credit onto transparent public blockchains.