On May 1 Labor Day in 2023, Blur and Paradigm launched the peer-to-peer NFT lending agreement Blend. The latest data shows that Blend has lent out 18,528 ETH and brokered 1405 loans.
Paradigm also published an in-depth interpretation of the Blend protocol on May 1.
This article introduces Blend: a peer-to-peer perpetual lending protocol that supports arbitrary collateral, including NFTs. Blend has no oracle dependencies and no expiration date, allowing borrowing positions to remain open indefinitely until liquidation, with interest rates determined by the market.
Blend uses a sophisticated off-chain quotation protocol to match users who want to borrow against non-fungible collateral with any lender willing to offer the most competitive rate.
By default, Blend loans have a fixed interest rate and never expire. Borrowers can repay at any time, and lenders can exit positions by triggering a Dutch auction to find new lenders at new rates. If the auction fails, the borrower is liquidated and the lender takes possession of the collateral.
Blend is implemented by Blur core contributors. In their implementation, some protocol parameters, such as protocol fees, are controlled by BLUR governance.
What makes Blend unique
A lot of work has been done before when it comes to NFT-backed loans. Popular models include perp-like protocols (such as Floor Perps and papr), syndicated lending protocols (such as BendDAO and Astaria), and peer-to-peer protocols (such as NFTfi and Backed).
Blend is most similar to the peer-to-peer model, but has some important differences in improving the borrower experience. Instead of exhaustively examining the details of all NFT-backed lending protocols, we describe some common design decisions and how Blend differs.
Some of these protocols require oracles to determine when to liquidate positions or to determine interest rates. But the price of a single NFT is difficult to measure objectively. Even floor prices are often difficult to measure on-chain. Solutions usually either involve a trusted party or can be manipulated through trading strategies.
Blend avoids any oracle dependencies in the core protocol. The interest rate and loan-to-value ratio depend on whatever terms the lender is willing to offer. A failed Dutch auction triggers liquidation.
Some protocols only support maturing debt positions. This is inconvenient for borrowers, who need to remember to close or rollover their positions before maturity (or risk severe penalties such as forfeiting their NFTs). The process of manually rolling positions also consumes gas, which reduces the yield of borrowing.
Blend will automatically roll over the borrowing position as long as a certain lender is willing to lend that amount as collateral. On-chain transactions are only required when interest rates change or when one of the parties wants to exit a position.
Some protocols do not support liquidation before maturity. This is convenient for borrowers and makes sense for many use cases. But since this effectively provides a put option for the borrower, the lender needs to demand short maturities, high interest rates and/or low loan-to-value ratios to cover the risk that the position may become insolvent.
In Blend, NFTs may be liquidated whenever a lender triggers a refinancing auction and no one is willing to take over the debt at any rate.
peer to peer
Some protocols pool lenders’ funds together and attempt to manage risk for them. This often means relying heavily on on-chain governance or centralized administrators to set parameters. It also makes it difficult to support long-tail collateral without permission.
Blend uses a peer-to-peer model where each loan is matched individually. Instead of optimizing for ease of use by lenders, Blend assumes the existence of more sophisticated lenders capable of participating in complex on-chain and off-chain agreements, assessing risk, and using their own funds.
In this section, we gradually build the protocol, starting with a simple peer-to-peer fixed-rate lending protocol, gradually adding adaptations to allow gas-efficient rolling and floating-rate market discovery.
First, let’s imagine how our protocol would work if it had maturing loans instead of permanent loans.
We start with the lenders. Lenders sign an off-chain offer to lend a certain amount of ETH principal at a specific interest rate and maturity date for any NFT in the specified collection. They make it publicly available (for example, by publishing it to an off-chain quote repository).
Borrowers have an NFT that they want to borrow. They browse the available off-chain offers and select a compatible offer that matches the terms they are interested in. They then create an on-chain transaction to satisfy the lender’s offer, place their NFT in a vault with a lien, and transfer the principal from the lender to themselves.
Before the due time, the borrower can pay the lender the repayment amount (calculated as the loan amount plus interest), and the lender closes the position and lets them withdraw the collateral. After maturity, if the loan has not been repaid, the lender can take the collateral.
Note that the borrower can choose not to repay the loan if the value of the NFT is less than the repayment amount.
In the above mechanism, if the borrower forgets to repay the loan before the due date, they will lose their NFT, even if the value of the NFT is much higher than the repayment amount. This seems harsh.
In many cases, someone else may be willing to pay the lender the full amount of the payment in order to take over the loan before a later due time, although the interest rate may be higher.
So instead of simply providing collateral to lenders, the protocol can run a competitive process to extend the loan, using a Dutch auction in the interest rate industry.
At maturity, if the borrower does not repay the debt, the refinancing auction will start at 0% and the interest rate will rise steadily. Once the auction reaches a rate at which a new lender is interested in lending, the new lender can accept it by submitting an offer on-chain. The new lender pays the old lender the full repayment amount, calculated from the moment the auction is completed, and takes over the loan until the new maturity time (which can be calculated as the current maturity time plus some agreement-specified loan term), reached using the auction interest rate.
The Dutch auction may fail to find lenders willing to lend, especially if the value of the collateral has fallen close to or below the value of the debt.
Once the auction reaches some defined maximum rate (say 1000%) without any new lenders stepping in, the protocol deduces that the position is insolvent or not viable, and liquidates the borrower. Existing lenders can then send transactions to take possession of the collateral.
In some cases, the same lender may be happy to continue making the same loan on the same terms, and so may the borrower. We can even consider this to be the default. In this case, it would be wasteful to run an auction.
Instead, we can design our protocol to optimistically renew loans. At each maturity, the borrower and lender default to extending the maturity by some predetermined loan term, on the same terms. The aforementioned auction will only occur if the lender seeks to terminate the loan.
One problem with the aforementioned protocol is that during the life of the loan, if the price of the collateral drops dangerously close to the price of the repayment amount, it cannot be liquidated before maturity.
This is less of an issue if the loan term is short, because if the lender is concerned about the safety of the collateral, they can trigger a refinancing auction at the next maturity.
We can imagine reducing the loan period to infinitesimal. If at any point lenders become concerned about the safety of the collateral, they may trigger a refinancing auction.
This lets us drop the notion of maturity and loan duration. By default, loans continue indefinitely until some user interacts with the contract. Interest is accumulating continuously, and the repayment amount is calculated at any time.
Borrowers can repay at any time. If borrowers want to change the amount they borrow or get a better rate, they can atomically draw a new loan from the collateral and use the new principal to pay off the old loan.
If the lender wants to exit the loan, they can trigger a refinancing auction, as described above. All timelines and deadlines during a refinance event can be defined relative to when the refinance was initiated.
Alternatively, if another lender makes a compatible offer, the current lender can skip the auction by submitting the other lender’s offer to the vault to exit their loan.
The protocol does not rely on governance that evaluates collateral or sets acceptable loan-to-value ratios, reducing the need for extensive on-chain governance or centralized management. However, there may still be situations where tuning certain parameters can enhance the functionality of the protocol. These parameters include:
Fees: Borrower and lender fees charged by the protocol.
Maximum Rate: The highest rate the loan auction must reach before liquidation occurs.
Auction Formula: As the auction progresses, the equation that governs the interest rate on the loan during the auction.
In Blur’s Blend implementation, after a 180-day waiting period, these parameters can be managed through BLUR governance to ensure optimal performance and adapt to changing market conditions in a decentralized manner.
Blend is a flexible, permissionless floating rate lending protocol that can support arbitrary collateral without relying on oracles, and allows the market to bear any interest rate and loan-to-value ratio.