Decentralised lending protocols are often presented to policymakers and the public as a categorically different kind of financial infrastructure — one without the maturity transformation, the discretionary risk committees, or the implicit government backstop that define traditional banking. In one important sense that is true: there is no branch, no loan officer, and no board meeting. But strip away the terminology and the underlying economic function is strikingly familiar. Depositors supply assets expecting to be able to withdraw them on demand. Borrowers draw down those same assets against posted collateral. The protocol sits in between, matching the two, and setting a price — an interest rate — to keep supply and demand roughly in balance. That is a bank's core function, running on a smart contract instead of a balance sheet.
How the mechanism actually works
Most major lending protocols use some version of a utilisation-based interest rate model. Utilisation is simply the proportion of supplied assets that has been borrowed out. As utilisation rises, the borrow rate rises with it — gently at first, and then, past a defined "kink point", very steeply. The logic is elegant: rising rates should discourage further borrowing and encourage new deposits, pulling utilisation back down before the pool runs dry.[1] Depositors, meanwhile, earn a share of that borrow interest, scaled by utilisation — the fuller the pool, the more each depositor's capital is actually working, and the higher their yield.
The mechanism is transparent, auditable on-chain, and applied identically to every participant. That is a genuine improvement on the opacity of some traditional credit decisions. But transparency of mechanism is not the same as safety of outcome.
Where the analogy to a bank run becomes exact
The vulnerability sits at the point utilisation approaches 100%. At that threshold, there is, by definition, no more liquidity left in the pool for a depositor to withdraw against — regardless of how attractive the interest rate has become. In traditional banking, this is precisely the scenario a lender of last resort exists to prevent: a central bank can extend emergency liquidity so that a solvent institution facing a temporary funding mismatch does not have to fail purely for want of cash on a given day.[2] A DeFi protocol has no equivalent facility. If utilisation locks at 100% during a period of stress — a large, correlated withdrawal, a collateral price shock, or simply borrowers who are slow to respond to a rate spike — the protocol's own code offers no relief. The rate curve can climb as high as its parameters allow; it cannot conjure liquidity that was never deposited.
International bodies monitoring the sector have increasingly flagged exactly this feature — liquidity and maturity mismatches embedded in DeFi's technical design, with no discretionary backstop — as a structural, not merely operational, source of fragility.[3] Analyses of historical stress events in major lending protocols have shown utilisation pinned at or near 100% for extended periods during sharp market downturns, precisely when depositors most wanted to exit.[4]
Why this matters for supervisors, not just users
None of this is an argument that DeFi lending is inherently reckless, or that traditional finance has solved liquidity risk — the 2023 banking turmoil is a recent reminder that it has not. It is an argument that the framing of DeFi as fundamentally risk-free because it is "trustless" and rules-based obscures rather than resolves the underlying question: who bears the loss when a liquidity mismatch materialises, and does anyone have the discretion to intervene before it does? A protocol's rate curve is a risk management policy with no committee behind it and no ability to pause, waive, or override itself in a genuine emergency. Understanding exactly how that curve behaves under stress — not just that it exists — is where meaningful oversight of this sector has to start.
The interactive tool below lets you set the total liquidity supplied to a simple lending pool and the amount borrowed against it, and watch utilisation, interest rates and available liquidity respond in real time, including what happens as the pool approaches its limit.