The Canto Lending Market (CLM) is an adaptation of Compound v2. While the Canto implementation has the same functionality its protocol parameters are different. This proposal aims to slowly move the Canto Lending Market towards more economic activity – both by enabling borrowers and incentivizing liquidators. If you are already familiar with the Compound v2 model feel free to skip to the last section.
Background
A decentralized lending protocol has to manage two objectives: The primary one is to remain solvent, i.e. ensure that at all times assets cover liabilities. The secondary one is to create a liquidity market that is attractive for both lending market users (depositors and borrowers) as well as liquidators. The primary goal of continuous solvency is addressed through a simple economic incentive: Unhealthy accounts can be liquidated by anyone while receiving a “bonus”.
When deliberating “optimal” parameters, one has to first understand the market risks the protocol is exposed to.
Market risks
A lending protocol’s market risks can be broadly classified into three categories:
- Volatility risk
Downside shocks to prices of collateral assets can lead to protocol insolvency if unhealthy accounts are not liquidated in a timely manner. - Liquidity risk
A loss of exchange liquidity can increase the cost of liquidation (e.g. slippage when selling collateral) to a point where the economic incentive to liquidate an unhealthy account disappears. Furthermore, the protocol may also compete against other lending protocols for liquidators – especially during times of market turmoil. - Liquidation Spiral risk
This risk describes the positive feedback loop that both of the previous risks can trigger: Liquidations can trigger a cascade of further liquidations and quickly dry up liquidity.
Parameters
The following three protocol parameters are the main controls to balance economic activity and safety:
- Collateral Factor
The Collateral Factor represents how much a user can borrow against the collateral supplied to the protocol. The parameter is asset specific. It value implies the maximum leverage a user can generate through the lending protocol. The parameter hence controls the riskiness of the borrowers: The higher these parameters are set, the more likely/frequent accounts will become unhealthy. These parameters primarily mitigate Volatility risk. - Liquidation Incentive
This protocol-wide parameter determines the amount of additional collateral a liquidator receives when repaying an unhealthy borrow position. Therefore it controls how attractive the protocol is to liquidators. The higher the parameter is set, the faster the unhealthy account will be liquidated. This parameter primarily mitigates Liquidity risk. Setting this parameter too high can also lower user demand as it may be seen as a potential cost of leverage for the user. - Close Factor
This protocol-wide parameter represents the maximum fraction of an unhealthy account’s borrow position that can be repaid in a single liquidate transaction. It applies to a single borrowed asset and not the aggregate of an account’s borrows. Setting this parameter lower will increase the account health more gradually as it may not be necessary to liquidate the full amount to bring the account back to a healthy level. Therefore this parameter can reduce the risk of a Liquidation Spiral.
Liquidation
The first order PnL of a liquidator can be estimated as liquidation “bonus” after subtracting the cost of selling the collateral:
PnL = LiquidationSize * LiquidationDiscount - Slippage(LiquidationSize)
Comparison
Below the parameters for a sample of the assets on Compound and Aave are shown.
Suggestions for the Canto Lending Market
The goal is to slowly allow for more borrowing activity on CLM. At the same time parameters are selected to rather err on the side of higher than necessary liquidation incentives to encourage market participants to develop tools of monitoring and executing liquidations. Similarly it is suggested to start with low collateral factors for stable-stable LPs and even lower collateral factors for risky-asset LP tokens.
Looking forward to any comments.