The Impossible Trinity of DeFi Lending

Author: Anthony Bowman

Translation: Jiahui, ChainCatcher

Fixed interest rate lending has genuine demand on the blockchain. The obvious response is to issue fixed-rate loans, but there is no matching fixed-rate lending demand in the market.

The vast majority of on-chain funds are chasing yields and eager for immediate liquidity. Therefore, issuing fixed-rate loans merely shifts interest rate risk from borrowers to lenders. When the lender is a treasury committed to instant liquidity, asset-liability mismatch occurs.

In variable-rate lending, interest rates fluctuate with utilization and market conditions, and borrowers directly pay for this volatility. This is a real cost, but it is clear and transparent, ending when the position is closed.

Suppose a lender holds a fixed loan with a 3% interest rate for six months. If interest rates rise, the same loan now yields 5%. Mark-to-market (MTM), the value of the old loan shrinks. With equal risk, new loans offering higher returns are available, and no one will pay the amortized value of the old loan.

The MTM loss of an individual loan remains on the books because the lender can hold to maturity and receive full repayment. It only becomes dangerous when the loan is placed within a system that requires continuous pricing.

Morpho’s V2 treasury is currently the most representative design publicly available, integrating fixed-rate loans into a treasury system committed to instant liquidity.

Excerpt: Morpho Fixed Rate Market: Unlocking the Potential of On-Chain Lending

According to public information, this design includes three components:

Morpho Blue: The existing variable-rate lending protocol. Lenders deposit funds into an isolated market, borrowers pay interest that fluctuates with utilization, and positions can be opened and closed at any time.

Morpho Midnight: Fixed-rate, fixed-term lending achieved through zero-coupon bonds (ZCBs). Both parties are matched by an intent engine, with each loan being a bond with specific collateral, term, and interest rate. These zero-coupon bonds are permissionless, supporting any collateral, term, and parameter combinations.

Morpho V2 Treasury: Managed by curators, it allocates deposits between Blue and Midnight based on yield. Depositors deposit and withdraw according to the treasury’s share price.

Image from Morpho V2 Treasury documentation

Imagine two competing USDC-denominated treasuries: Treasury A allocates funds to both Blue and Midnight, while Treasury B only allocates to Blue. Treasury A allocates 30% to Blue (variable, 3%) and 70% to Midnight (fixed, 3%).

A rate shock pushes the variable rate up to 5%, while Midnight’s position remains locked at 3%. Treasury A’s blended yield rises to 3.6% (5%×30% + 3%×70%). The pure variable-rate treasury B rises to 5%. This 140 basis point gap creates conditions and incentives for a run on the treasury.

Depositors in Treasury A don’t need to calculate MTM losses, nor do they need to be aware of them. The yield spread itself acts as a coordination mechanism. Funds flow from A to B chasing higher yields, and exit through the only liquid part of A (the variable module).

This first depletes the most profitable on-paper gains, causing Treasury A’s blended yield to fall further and accelerating the run. What remains is a pool of illiquid fixed loans, below market rates, waiting only to mature.

Now, reverse the situation. When interest rates fall, Treasury A’s fixed positions are above market levels, and depositors enjoy MTM-based gains but cannot retain them. Depositors in Treasury B, sensing higher blended yields in A, rush in to deposit and share in the gains.

New funds enter at the current share price, proportionally allocated to existing on-book positions. This means new money enjoys the same proportional rights to those positions with yields above market rates. These gains are thus diluted.

Both scenarios are dead ends. When rates rise: depositors run, causing a treasury run. When rates fall: yields are diluted by new entrants.

The fundamental issue lies in bond valuation methods. Although accounting treatments for amortized zero-coupon bonds vary, regardless of that, the real problem is that external interest rate changes alter the actual value of bonds, while amortized-based pricing fails to reflect this.

Valuing bonds at amortized cost leads to asymmetries. A clear solution is to create a secondary market, which theoretically would allow treasuries to price bonds at their true market value.

However, for permissionless zero-coupon bonds with arbitrary collateral, terms, and parameters, a secondary market cannot form because each bond is essentially unique, lacking a liquid benchmark for valuation.

Even if a secondary market were to form, pricing treasuries based on it would only mask a worse problem. Share prices would depend on external trading data of customized, illiquid bonds. Anyone able to influence this data could manipulate share prices and arbitrage during treasury entry and exit.

Zero-coupon bonds with high expressive capacity and treasuries committed to instant liquidity are structurally incompatible. Perhaps some internal mechanism could resolve this within the architecture, but I have yet to see any such proposal, and I am very curious whether Morpho has a countermeasure.

Personally, I believe that directly issuing fixed loans is not a solution, at least not in the short-term outlook of over-collateralized lending. If borrowers want fixed interest rates and lenders desire instant liquidity, then interest rate risk must be transferred to those willing to assume this directional risk exposure.

If the underlying variable-rate benchmark curve becomes more efficient and robust, the rate risk buyers can offer better fixed rates. As I explore here, we are still far from the final form of variable-rate market design.