A modified IRM for Morpho markets

1

Note: The aim of this post is to initiate a discussion about the opportunity to launch a second IRM. A full proposition should encompass a deeper assessment of the current model, onchain analysis, comprehensive backtesting and paramater’s values optimization.

Introduction

Interest rate models in lending protocols have recently become an area of active research. While the Compound’s two-slope linear curve, or kinked curve, has long been the industry standard, more protocols are experimenting with new models.

Morpho’s Adaptive Curve, currently the only model applied to Morpho markets, introduces additional flexibility to the kinked model by allowing the interest rate (IR) to adjust according to both utilization rates (UR) and broader market conditions. Similar to static IR models, the curve has two slopes: a low slope for UR below 90% and a steeper slope for UR above 90%. In addition, when the UR exceeds the target, the curve gradually shifts upward, increasing the IR to incentivize loan repayment. Conversely, when the UR falls below the target, the curve gradually shifts downward, lowering the IR to encourage borrowing and thereby increase UR over time. This dynamic keeps the UR close to 90%, enhancing capital efficiency.

Current model assessment

The kinked model has been long known to enhance IR volatilty compared to a simple single base rate model. The problem is especially acute when the UR exceeds the threshold above which the slope increases.

A too volatile interest rate may also be an issue in the case of Morpho, at least for a subset of markets. Morpho’s isolated markets may result in thinner liquidity compared to monolithic pools. Borrowers and lenders may experience high slippage and interest rate volatility the same way traders experience high slippage and price volatility in low liquidity AMMs’ pools.

As an illustration, the Figure displays the interest rate dynamics in four markets, not necessarily representative but all with significant liquidity. It outlines the challenge posed to borrowers in their daily interest rate management.

SyHbtuJ7kg.png1104×562 139 KB

The instability of the IR is amplified by the interplay of the two mechanisms present in the adaptive IR model:

  • The two-slope curve, which instantly adjusts the IR in response to variations in UR.
  • The vertical translation of the curve, which aims to bring the UR back to its 90% target.

By pushing the UR back to the 90% target over and over again, the interest rate is permanently on the brink to take off following newly borrowed amounts or transfers of liquidity out of the market.

A modified pricing rule

It is suggested to create a second IRM in which the high slope of the curve is removed while the vertical translation around the 90% target is maintained. The curve becomes a straight line that gradually elevates as long as the UR remains below 90% and gradually decreases when the UR exceeds this threshold. This single base rate curve eliminates a large share of high-frequency volatility while the translation dynamics preserves the adhrence to the 90% target.

Potential risks

Risk of locked-in liquidity

The new curve trades off a more stable interest rate against a higher risk that the market’s UR reaches 100% during a prolonged period of time. However, the risk is mitigated by two mechanisms present in the Morpho design:

  • The speed at which the curve adjusts is determined by the distance of current utilization to the target. Hence, the curve will shift upward faster when the UR is 100%.
  • Most lenders withdraw available liquidity at the vault’s, not market’s level. Lenders’ inability to withdraw funds is only possible if all markets covered by the vault are fully borrowed.

While lenders are compensated by a high and increasing interest rate, the velocity of the curve could be raised to prevent excessive periods of full UR.

Risk of liquidity fragmentation

The same markets could be duplicated with a different IRM, entailing liquidity fragmentation. However, this is only a concern if curators split the liquidity between identical markets except the IRM. If liquidity concentration matters, they will supply in only one type of IRM. Also, curators are not expected to migrate existing markets to a new IRM, as the transition could be painful for borrowers. At the same time, the new IRM would expand the market’s parameters set for newly created markets.

Conclusion

By moderating the slope of the IR curve while retaining the dynamic adjustment around the target UR, the new IRM enhances the stability of the interest rate while maintaining a high level of capital efficiency. We believe that this modified pricing rule could usefully expand the set of IR models among which market creators could choose.

2

Hi Prevert,

Thank you for your thoughtful contribution. It’s great to see community members engaging deeply with such topics.

As a preliminary note, setting the IRM function is not a simple optimisation problem with a single, clear objective. Instead, it requires balancing capital efficiency and liquidity while accommodating the asymmetric needs of lenders and borrowers. There is no perfect setting—only a reasonable balance of trade-offs—and it’s difficult to prove that one setting is strictly better than another. More importantly, as you noted, introducing additional IRM models comes with significant costs—it could create duplicate markets that split liquidity and reduce user experience coherence for both borrowers and curators. For these reasons, the idea of adding an IRM should be considered very carefully. Nevertheless, critical assessment remains valuable.

If we summarise your argument correctly, the key idea is that because of vaults, lenders and borrowers benefit from the combined liquidity of multiple markets rather than being constrained by a single one. However, borrowers are directly impacted by rates at individual market level. This suggests that loosening liquidity constraints on individual markets to limit rate spikes would improve borrower experience without harming overall liquidity.

We think this perspective is sound, and adjusting the IRM in that direction could have benefits. However, several additional factors must be considered:

  1. Rate spikes are not harmful per se

While rate spikes might seem concerning, what ultimately matters to borrowers is the average rate over the loan duration, not short-term fluctuations. As explained in this document, vault curators quickly adjust for isolated spikes, so borrowers need not be concerned. Rather than focusing on IRM adjustment, addressing borrower perception through interface improvements and clear communication might be a more effective approach.

  1. Spikes serve a purpose in systemic movements

When price spikes are persistent and widespread, they play a crucial role in influencing the behaviour of borrowers and lenders, helping to maintain market stability. They can also trigger liquidations, which can help to keep the market healthy during adverse events. In such scenarios, when problems arise with loan or collateral assets, the compensating liquidity effect from vaults is less of a guarantee, making the liquidity safeguards of the IRM crucial.

  1. Deep liquidity is key for growth

A frequently overlooked but essential aspect of liquidity is its role in enabling the absorption of large borrowing demands. In several instances, substantial borrowers have expressed interest in engaging with Morpho but found available liquidity insufficient—even with the public allocator. While lowering liquidity requirements might seem beneficial for current user base, maintaining a deep liquidity cushion is crucial for onboarding major borrowers, especially during this growth phase.

  1. Liquidity enables effective vault curation

While liquidity may not directly impact end users, it is fundamental to vault curators’ ability to reallocate deposits. This flexibility is key to maintaining risk management operations and ensuring the system remains resilient.

3

Thanks for sharing your point of view on this subject, which is both interesting in many respects and very important given that we rely on an algorithm to price billion of dollars of liquidity lent and borrowed. The aim of this post was to initiate a discussion, and I hope other actors in the ecosystem will share their views as well.

Let me briefly extend the discussion in two directions.

First, the question of whether spikes in interest rates deteriorate the borrower experience is partly a quantitative one. It would be interesting to study whether some borrowers close their positions early during interest rate spikes.

Second, the user experience is also directly impacted by instant changes in interest rates following an additional supply of liquidity or borrowed amount —the “interest rate impact,” similar to the price impact experienced by traders in AMMs. Due to the high slope of the pricing curve when the utilization rate (UR) is above 90%, some users may refrain from borrowing or supplying if they see the interest rate changing instantly by a large margin. The same way traders dislike price impact, lenders and borrowers certainly don’t like interest rate impact.

In a related post, I show that the impact is negligible when UR are below 90% but becomes significant above this threshold, even for small variations of available liquidity. For instance, for a target interest rate of 7%, increasing liquidity borrowed from the market by only 1% of the total supply leads to a significant increase of interest rate by 2 percentage points and more. It is between 5 and 8 percentage points when the target interest rate is 15%