We see overcollateralised lending as the more interesting, DeFi native primitive which, if appropriately designed, can satisfy a huge range of use cases. That said, there will always be demand for uncollateralised lending. This can broadly be split into two categories: (1) KYC-based uncollateralised lending - this is similar to traditional bank underwriting, requiring KYC and contracts but leveraging the transparency of DeFi (2) On-chain reputation based lending - this would be permissionless and rely purely on some kind of on-chain reputation or social graph.

While we see (2) as the ultimate end-game, it relies on a well-developed on-chain reputation system which is currently very far away. As such, we’ll focus primarily on (1).

While this isn’t a 0 to 1 innovation similar to other DeFi primitives, the relative performance of DeFi-based uncollateralised lenders like Maple/TrueFi/Clearpool compared to their centralised counterparts demonstrates that the transparency of the DeFi can lead to better outcomes. Most borrower demand on these protocols comes from crypto-related market-making firms and hedge funds who require spot leverage to facilitate their operations, and struggle to get credit from banks and traditional crypto institutions. Lender demand comes from higher yields, with the Ethereum based uncollateralised lending protocols consistently delivering yields 2-3x those which can be achieved on overcollateralised credit protocols.

All current uncollateralised lending designs involve some kind of KYC on the borrowers as well as contracts between borrowers and some centralised entity acting on behalf of the protocol. Some designs involve single asset pools with borrowers being whitelisted via governance vote (e.g. TrueFi), while others involve segregated pools each managed by certain governance whitelisted delegates (e.g. Maple).

Another way in which these protocols tend to differ is in how interest rates are set. On some protocols, such as Maple or TrueFi, interest rates are defined by the borrower or the pool operator. In these models, lenders can only participate at the yield offered by the pool and lenders within a given pool generally earn the same yield.

A different, more market-driven approach, has been implemented by Atlendis. Within this model, both borrowers and lenders get to decide at what yield to participate in a given deal. Specifically, whenever a pool is created, borrowers set a maximum interest rate and a given maturity for the loans. Potential lenders, on the other hand, can provide liquidity at any given interest rate below the borrower’s maximum limit. Notice that whenever a lender deposits liquidity at a certain yield, it acts as a de facto limit order that only gets taken if/when the borrower decides to do so. In this sense, this model allows for both borrowers and lenders to participate on their own terms. To reduce the opportunity cost to lenders associated with waiting for borrowers to take their liquidity, idle liquidity can be automatically deployed to a money market like Mars to earn passive yield.

While we think both approaches to interest rates have merit, as explained in the Fixed Rates Lending section above, we believe there’s an opportunity to move DeFi to a more efficient market-driven environment, where both lenders and borrowers have more flexibility to express their views.

Where: On a chain with a native orderbook such as Injective, as that infrastructure would facilitate all the operations involved in lenders providing liquidity (or limit orders) and borrowers taking that liquidity.

Mentor: Jonathan Erlich (Head of Research at Delphi Labs), Gabriel Shapiro (General Counsel at Delphi Labs)