Managing borrowing costs in Liquity V2 requires dynamic interest rate adjustments, but running complex calculations on Ethereum Mainnet can be costly and inefficient. Gas fees make frequent updates impractical, limiting borrowers’ ability to optimize their rates in real time.
Rather than solely relying on manual adjustments, Liquity V2 integrates an autonomous rate management system (ARM) powered by DFINITY’s Internet Computer (ICP). By performing the rate calculations on the ICP while keeping Liquity’s core immutable and Ethereum-native, ARM helps ensure that the interest rates are more efficient and responsive to rate adjustments.
ICP’s performant architecture allows for a continuous rate monitoring and adjustment according to the current market conditions, reducing reliance on manual interaction, while ensuring a gas-efficient, trust minimized experience. Borrowers can delegate to an automated system that optimizes costs - all without compromising control.
Initially, the management system offers a single conservative interest rate strategy that is available for all three collateral assets (ETH, wstETH and rETH) with preset parameters.
The strategy aims to keep the subscribing borrowers’ interest rates in line with the market’s average (confined to a range of 0.5%-50%), monitoring it once an hour and making adjustments based on cost and redemption risk considerations.
Borrowers can subscribe to the ARM strategy by using the delegate feature on one of the Liquity V2 frontends.
View it on Github here: https://github.com/liquity/bold-ir-management
Based on user demand, the system could be extended by offering multiple strategies with varying risk levels.
Interest rate: decentralized delegation
One of Liquity V2’s key innovations is the ability of borrowers to set and adjust their own interest rates in exchange for their redemption risk (learn more here): borrowers with low interest rates relative to their peers have the highest risk of being affected by redemptions and vice versa.
To facilitate gas-efficient interest rate adjustments and a passive borrowing user experience, users can delegate their rate management to individuals or professional batch managers. As this requires a degree of trust in the delegated entities, borrowers may instead want to delegate to a smart contract managing interest rates in a fully autonomous manner.
However, due to its limited performance and gas constraints, running extensive calculations on Ethereum Mainnet would be prohibitive, if not impossible. Therefore it seemed prudent to build an autonomous interest rate management (ARM) system using DFINITY’s Internet Computer, a more performant and scalable L1 blockchain.
ARM initially offers a single conservative strategy for each collateral asset (ETH, wstETH and rETH). The system could be extended to multiple and more sophisticated strategies in a future release. Further strategies could also be developed by other project teams.
Interest rate adjustment strategy
In Liquity V2, borrowers incur two types of costs:
- Interest rate: a recurrent rate set by the borrower and charged on their current debt
- Premature adjustment fee: a one-off fee corresponding to 1 week of the average interest rate of the respective collateral market, charged on the debt whenever the borrower adjusts their interest rate within less than 7 days since the last adjustment (“cooling off period”).
The premature adjustment fee, which is also charged when a new Loan is opened or when its debt is increased, aims to prevent borrowers from unduly evading redemptions (hitting higher rate borrowers) just before they are about to happen, by increasing the rate on the fly.
Being hit by a redemption not only means an immediate reduction of ETH exposure, but also implies that the borrower has to pay the premature adjustment fee for reestablishing the original position.
While the adjustment fee is typically low (<0.4% for interest rates below 20%) in most situations, it can add up if a borrower adjusts their interest rate too frequently or if a borrower needs to regularly reopen the position after redemptions.
Therefore, an optimal interest rate strategy should minimize the costs of borrowing by striking a balance between the interest rate and its adjustment frequency as well as the redemption risk. Given that the implemented ARM pilot strategy doesn’t automatically reopen redeemed positions, it is designed to minimize the redemption risk in the first place.
Adjustment metrics
The implemented strategy takes the following metrics into account for triggering interest rate adjustments:
- Current total debt in front of the delegating borrowers
- Current redemption fee
- Time since the last adjustment
The debt in front metric used by ARM aims to keep a sufficient buffer of debt in front of the managed borrowers, targeting a default range of 42.5% - 57.5%, thereby roughly following the average rate for each collateral.
Note that the debt in front is calculated separately for each collateral according to the current “redemption split”, i.e. in proportion to the amount of BOLD kept outside of the respective Stability Pools (see explanation). As a result, a loan typically has redeemable BOLD as a buffer on all three branches before it gets hit by a redemption.
When the dynamic redemption fee exceeds 0.5% as redemptions occur, the target range is increased to account for the higher redemption risk and the possibility of other borrowers increasing their own interest rates.
See the Readme for the exact formulas.
Asymmetric trigger conditions
The trigger conditions depend on whether the current debt in front is lower or higher than the respective bounds:
- If the current debt in front drops below the lower bound, the rate is increased to ensure that all delegating borrowers have a debt in front corresponding to at least the respective midpoint (earlier joiners will benefit from a higher cushion for the same rate as borrowers with the same interest rate or manager are sorted in LIFO order). The premature adjustment fee isn’t taken into account given the urgency of rate increases.
- If the current debt in front overshoots the upper bound, the rate is decreased back to achieve the midpoint, provided that the last adjustment happened less than 7 days ago, or that the prospective savings from the lowered interest rate until the end of the cooling off period exceed the adjustment fee.
See the Readme for more details.
Architecture
The ARM system consists of three Liquity V2 batch manager contracts on Ethereum Mainnet (one for each collateral type) and a single management canister (smart contract) hosted on the Internet Computer.
Each batch manager contract has a pre-registered externally owned account (EOA) as its owner which is the only address authorized to perform interest rate adjustments on behalf of the delegating Loans. The EOAs have been generated by a specific subnet of the IC in a safe manner via a threshold ECDSA signature generation scheme as part of the IC’s chain-key technology. This means that nobody knows the EOA’s private key and only a qualified majority of the subnet nodes can collectively sign transactions for the respective address.
The IC canister is triggered by a timer every 1 hour to check whether the above-mentioned update conditions hold for any of the three collateral types. If yes, the subnet crafts an Ethereum transaction from the respective EOA and submits it to the Ethereum network through a variety of RPC providers, using HTTPS outcalls.
The canister has a built-in fault tolerance mechanism that may resubmit transactions that fail for various reasons.
Self-sustained via permissionless recharging
The ARM system is designed to be fully autonomous and self-sustained with regard to the Ethereum gas fees and the cycles charged by the Internet Computer. For that matter, all three batch management contracts charge a management fee of 0.1% p.a., gradually increasing the debt of the delegating borrowers.
As the management fee accrues in BOLD, the batch management contract allows anybody (e.g. professional arbitrage bots) to exchange the respective amount by providing ETH at a discount of 3% compared to the current market rate.
The obtained ETH is used to cover the transaction fees for the interest rate adjustment transactions and a part of it is wrapped into ckETH from time to time, an ICRC-2-compliant token that is backed 1:1 by ETH. Targeting a predefined reserve of cycles, the ckETH balance is converted into cycles when needed through a discounted permissionless swap mechanism.
These mechanisms collectively recharge both the canister’s cycle balance and the EOA’s ETH balance using the proceeds from the management fee. Due to its efficient O(k) design, interest rate updates performed by the batch manager incur the same transaction fee regardless of the number of delegating loans.
While the system aims to ensure that the strategies never run out of gas or cycles, it ultimately remains dependent on the amount of debt under management and the frequency of updates. If the collected management fees turn out to be insufficient, any interested party may send ETH to the respective EOAs or cycles to manually recharge the management canister.
For a comprehensive look at the technical implementations of ARM, check out this Github readme: https://github.com/liquity/bold-ir-management?tab=readme-ov-file#bold-interest-rate-manager
Disclaimer
It is important to note that despite the conservative design of the ARM strategies, there’s no guarantee that they can completely prevent redemptions of participating borrowers. Users should thus be aware of the risks and associated costs.
Internet Computer
The Internet Computer (IC), developed by DFINITY, is a highly scalable blockchain that runs decentralized applications at web speed. Its smart contracts, known as canisters, can handle large amounts of data efficiently, unlike traditional blockchains that rely on Layer 2 solutions. The IC uses chain-key cryptography for secure communication between nodes and to enable canisters to sign transactions on external networks like Ethereum, integrating directly with the public internet and eliminating the need for centralized servers. With low computational costs and robust security, the Internet Computer provides an ideal platform for scalable blockchain projects requiring cross-chain functionality.
