Lending Platform Landscape

Stablecoins have grown to represent many tens of billions of dollars in value in the DeFi ecosystem. The dominant lending model for stablecoins (and crypto assets in general) is the liquidatable over-collateralization lending model. Collar's goal is to offer an alternative lending model specifically for stablecoins that is more capital efficient. What follows is a brief history of the current lending paradigms.

Peer-to-Peer Lending with Over-Collateralization

Decentralized lending platforms grew from peer-to-peer lending models, which allow users to create loan requests containing the terms such as the amount, the requested interest rate, and the duration of the loan. Lenders can browse through loan orders to choose those that look attractive to them. The entire process is executed via smart contracts and no centralized custody is involved. These platforms are prototypes of an on-chain money market and represent a key phase in the growth of DeFi. But without repayment enforcement, loans are mostly backed by 200% collateral. The over-collateralization offsets the default risk for lenders. ETHLend was one of the earliest Defi protocols that followed this paradigm.

Algorithmic Pool-Based Money Markets

The next lending model to be developed was the algorithmic pool-based one, with the first being launched by Compound. Lenders deposit their assets into a pool and their share of the pool is tokenized. Borrowers supply collateral to withdraw directly from the pool and pay interest to the pool. The accrual of interest increases the amount of underlying assets a lender’s tokens entitles them to. In Compound, the interest rate paid by borrowers is determined for each asset pool by the total proportion of assets currently borrowed. This interest rate is therefore floating and can change significantly due to large deposits made into or withdrawals taken from the pool. Loans are also still backed by 200% collateral in this model. The advantage of the algorithmic pool-based model is that it provides users with a simple way to monetize their idle assets and take short positions.

Using Composability to Increase Liquidity in DeFi

The success of algorithmic pool-based money markets illustrated the need for decentralized protocols that increase liquidity either by replicating a part of CeFi, or by offering a completely new lending functionality. By making these protocols composable, the possibilities within the DeFi ecosystem grow with every new protocol launch. A notable new lending functionality, the flash loan, was popularized by the AAVE and dYdX protocols. This functionality enabled any user to lever up to any desired amount of capital when arbitraging. In a similar spirit, protocols were developed to build liquidity bridges between the different lending protocols themselves, such as CREAM’s Iron Bank, which uses a credit model for protocol-to-protocol lending. Most protocols developed in this era incentivize liquidity in the form of governance token rewards that entitle holders to portions of the fees generated by the platforms.

Attempts to Lower Collateral Ratios

With composable, yield-generating lending platforms in place, the demand for borrowing increased, and protocols tried using various methods to drive down collateral ratios in order to increase capital efficiency. Wing introduced credit evaluations for borrowers in order to lower the ratio (in select cases, for low amounts) to 80%. Liquity lets users borrow its native stablecoin against ETH and incentivizes borrowers to pool those stablecoins back into the platform to act as guarantors of last resort in order to perform efficient liquidations and lower the ratio down to 110%. Alchemix maintains the standard 200% ratio, but applies the yield generated by the collateral to the debt so that the borrower’s leverage decreases over time. The most recent trend has been the development of interest rate derivative protocols that deposit assets into multiple yield-generating platforms in order to tokenize the yield and bundle it up into tranches having varying risk profiles, such as BarnBridge and Saffron.
Throughout all this development, the underlying paradigms of over-collateralization and pool-based algorithmic lending have never been seriously challenged by a more capital efficient alternative.

Flaws of Current Lending Models

1. Floating Interest

Borrowers are burdened with estimating the capital cost of borrowing and the risk of liquidation. Algorithmic pool-based lending protocols use capital utilization to decide interest rates; this mechanism is vulnerable to manipulation by deposit whales as the following example shows:
Whales can manipulate the interest rate by taking advantage of liquidity premium
This example shows a lending pool with insufficient liquidity that is vulnerable to being pumped dry by a whale. The whale has a 50M USDC saving position with 9.32% lending APY. But when withdrawable liquidity in this lending pool is only 10M USDC, this whale can withdraw all of it to raise the lending APY to 38.2%. Before this withdrawal, his 50M deposit is rewarded 532 USDC/h, and after the withdrawal his reward rises to 1774 USDC/h. In effect this lender provides a lesser deposit to receive greater earnings.

2. Risk of Frozen Assets

Lenders cannot accurately predict when capital utilization will be high enough to prevent them from withdrawing their assets in algorithmic pool-based lending protocols.

3. Low Capital Efficiency and High Risks from Liquidation

Borrowers must accept greater liquidation risk to achieve greater capital efficiency when they are required to over-collateralize. Over-collateralization puts a low ceiling on LTV and borrowers risk being penalized by liquidations and their associated fees.

Benefits of the Collar Lending Model

Collar's lending model overcomes the previously stated flaws and offers further benefits to users that are briefly described here:

1. Capital Efficiency

When a borrower deposits assets, they receive tokens that can be sold to a lender for 100% the value of their deposited assets (minus the interest paid for the loan).

2. No Liquidations

Borrowers always retain access to their collateral before the expiry time of their loan, even when the value of the borrowed assets change. There is also no reliance on oracles for price feeds to determine the collateral value and therefore fewer exploit vectors.

3. Predictable Costs and Interest

By including operations intended for arbitrageurs to use and expiry times for loans, market forces will put pressure on interest rates to only reflect the perceived risk of de-pegging, the demand, and the time remaining until the expiry time. For established stablecoins this will lead to a consistent price curve and predictability for both borrowers and lenders.

4. Flexible Repay

Borrowers can default their loan with no penalty and can repay it at any time before the expiry time.

5. Predictability for Lenders

Lenders can sell their debt if they find the price is acceptable to them, and if not they can wait until the expiry time of the loan to collect the collateral/repaid assets.