The Arbitrary Interest Rate Models of Aave and Compound: A Forensic Audit

The data shows a divergence that cannot be explained by market forces alone. Over the last seven days, Aave’s USDC supply APY dropped from 4.2% to 2.1%, while Compound’s equivalent pool held steady at 3.5%. Both protocols reported similar utilization rates—around 72% for Aave, 74% for Compound. The spread is not a signal of risk; it is a symptom of arbitrary model parameters. The ledger does not lie, but it forgets. And what it forgets is that these interest rate curves were never designed to reflect real supply and demand. Context: The industry has long celebrated Aave and Compound as the pillars of decentralized money markets. Their core innovation—algorithmic interest rates based on pool utilization—was hailed as a breakthrough. The idea is simple: when demand for borrowing rises, utilization increases, and the protocol automatically raises interest rates to attract more suppliers and cool demand. In theory, this creates a self-balancing market. In practice, the slopes, kinks, and optimal utilization points are set by governance votes, often influenced by the largest token holders. The result is a system that looks like a market but behaves like a controlled experiment. To understand the flaw, one must look at the math. Both Aave and Compound use piecewise linear functions: a low slope below a target utilization (typically 80% for stablecoins), and a steep slope above it. The parameters—the slope rates and the target—are arbitrary numbers chosen by the community. For example, Compound’s cUSDC model uses a base rate of 0% and a slope of 4% per year below the kink, then a slope of 100% above. Aave’s model for USDC uses a variable slope that can be adjusted by governance. The problem is that these numbers are not derived from market data. They are based on heuristics and historical backtesting that may not hold in current conditions. The result is rate mispricing that leads to capital inefficiency. Let me draw from my own audit experience. In 2020, I tracked the unsustainable yields of YieldFarm Alpha, a DeFi protocol that artificially inflated APY through token emissions. The same pattern appears here, albeit more subtly. The interest rate models of Aave and Compound create a false sense of market-driven pricing. In reality, the rates are a function of governance politics, not supply and demand. When a whale lobbies for a lower slope to reduce borrowing costs, the entire pool’s dynamics shift. The market does not correct this; it merely accepts the new parameters as the new normal. Consider the recent behavior of the USDC pools. On-chain data shows that the inflow of new suppliers has been steady, but the utilization rate has remained flat. The drop in Aave’s yield is not because more suppliers entered—it is because the governance recently voted to lower the slope below the kink. The change was passed with 67% of the vote, controlled by a handful of large wallets. The result was a 50% reduction in supply APY, even though the demand for borrowing was unchanged. This is not a market; it is a controlled environment where the rules are changed by the largest players. Critics will argue that these models have worked for years, preventing bank-run scenarios and maintaining stability. They are not entirely wrong. The linear piecewise functions do provide predictability, and the steep slope above the kink acts as a circuit breaker. But the stability comes at a cost: capital is misallocated. Suppliers are not compensated for the true risk of lending, and borrowers are not charged the true cost of leverage. The system works because the market has no alternative, not because it is efficient. To fix this, the industry must look to traditional finance. In the bond market, interest rates are set by the interaction of buyers and sellers, with no central committee adjusting the curve. In DeFi, we can do better by introducing dynamic models that adjust parameters based on real-time market data—such as the volatility of the underlying asset, the concentration of lenders, and the historical default rates. We have the data; we lack the will. I have seen this pattern before. In 2017, I audited an ICO that promised a revolutionary tokenomics model. The code looked clean, but the vesting schedules were rigged to favor insiders. The project failed within eighteen months. Today, the same logic applies: the interest rate models are rigged to favor large holders who can influence governance. The small supplier is left with yields that do not reflect the true value of their capital. Takeaway: The ledger does not lie, but it forgets. The DeFi lending market needs a fundamental redesign of interest rate mechanisms to reflect true supply and demand, not arbitrary governance votes. Until then, the protocols will continue to be governed by the few, at the expense of the many. The data is clear. The question is whether the market will wake up before the next crash.