By Crypto Loop · Updated 2026-10-06T20:47:44.583Z
What DeFi lending actually is
DeFi lending is usually an overcollateralised system: a borrower posts crypto collateral and can then borrow against it up to a limit set by the protocol. The core logic is simple. If the collateral is worth enough relative to the debt, the system can absorb ordinary price moves. If the value of the collateral falls too far, the protocol tries to protect lenders by liquidating the position before the shortfall becomes too large.
This structure is often presented as a source of passive yield, but the yield framing can hide the more important question: what risks are being paid for, and who absorbs them when the system is stressed? In most DeFi lending markets, the lender is not evaluating a company balance sheet or an underwriter’s credit judgement. Instead, the lender is relying on collateral value, liquidation rules, oracle accuracy, and the code that enforces both.
A practical way to think about DeFi lending is as a rule-based risk machine. The protocol sets terms in advance: what collateral is accepted, how much can be borrowed, how often rates change, how close a position may get to liquidation, and what happens if liquidation fails. Understanding those rules is more useful than focusing on the label attached to the expected return.
Collateral and overcollateralisation
Collateral is the asset a borrower locks in to support a loan. In DeFi, accepted collateral is usually more valuable and more liquid than the asset being borrowed, because the protocol needs a buffer against price swings. That buffer is the point of overcollateralisation. If a borrower deposits value of 100 units and can only borrow 50 to 80 units, the remaining margin is meant to absorb losses from adverse price movement.
The key metric is not simply how much collateral exists, but how much headroom remains after accounting for volatility, concentration, and liquidity. A collateral asset that appears strong in calm markets can behave very differently when trading depth thins out or when many users try to deleverage at once. In those situations, the nominal buffer may shrink faster than expected because the market price used for accounting can move quickly while execution prices worsen even faster.
Decision check: before interacting with a lending protocol, ask whether you understand what counts as eligible collateral, how the protocol values it, and whether that valuation is based on a spot market, a time-weighted average, or some other oracle mechanism. Also ask whether the collateral itself is easy to sell in size, because illiquid collateral can make liquidation outcomes worse even if the headline collateral ratio looks comfortable.
Utilisation and why rates change
Most DeFi lending markets use a utilisation model. Utilisation is the share of supplied assets that are currently borrowed. When utilisation is low, lenders may receive lower rates because there is abundant liquidity left in the pool. When utilisation rises, borrowing becomes more expensive and lender rates usually rise as well, because available liquidity becomes scarcer.
The intuition is straightforward: if nearly all supplied funds are already borrowed, the pool has little room to meet new withdrawal demand. Higher borrowing costs are used to ration demand and encourage some borrowers to repay or reduce leverage. In other words, the rate is not fixed by a central desk; it is usually a function of the pool’s utilisation, with parameters chosen through governance or protocol design.
This does not mean higher utilisation is always “better” for lenders. Extremely high utilisation can make liquidity fragile. Even if a pool shows a high nominal lending rate, the lender may face withdrawal delays, slippage if liquidity must be exited through secondary markets, or rate instability if the model reacts sharply to small changes in demand. A sensible check is to look for the utilisation bands at which rates accelerate, because those bands often reveal where the protocol expects stress to begin.
Variable rates are not a promise
Variable lending rates are often misunderstood as if they were a dependable income stream. They are not. They are a formula linked to supply and demand within the protocol, and that formula can move quickly as utilisation changes. A borrower adding leverage, a large repayment, a liquidity withdrawal, or a governance parameter update can all change the rate profile.
A useful mental model is that the rate is a price for scarce liquidity, not a guaranteed coupon. If utilisation rises, the borrow rate may climb to deter further borrowing; if utilisation falls, the rate may drop. This is important for anyone comparing DeFi lending to fixed-income products, because the path of the rate matters as much as the level. Even if the average looks attractive over some period, the rate may have spent long intervals far below that average.
Worked example, using hypothetical arithmetic only: imagine a pool with 1,000 units supplied and 800 units borrowed. Utilisation is 80 percent. If the protocol’s model raises rates sharply above 80 percent utilisation, a small additional borrowing event of 50 units would push utilisation to 85 percent. If the rate curve is steep in that range, borrowers may face a much higher cost almost immediately. That higher cost may help attract more supply over time, but it does not eliminate the underlying fragility in the short term. The example is not a prediction; it only shows how sensitive variable-rate systems can be near congestion.
Oracle pricing and liquidation mechanics
Liquidation is the process that closes or reduces a risky position when collateral value falls below a required threshold. In DeFi, that trigger usually depends on an oracle: a data source that tells the smart contract what the collateral and debt are worth. Because the protocol cannot inspect the real world directly, it must trust an external pricing mechanism to decide when a position is healthy and when it is not.
This creates a critical dependency. If an oracle price is stale, delayed, manipulated, or simply unrepresentative of the broader market, liquidations can occur too early, too late, or at the wrong price. An early liquidation harms the borrower by closing positions that may have recovered. A late liquidation harms lenders by allowing losses to grow. In a fast market, even a correct oracle can arrive too slowly to prevent losses, because the issue is not only the price reference but also the speed of enforcement.
Practical check: ask how often oracle values update, whether the protocol uses one source or several, and whether the liquidation threshold includes a margin of safety for price delays. Also ask what happens during market outages, oracle failure, or chain congestion. If a position cannot be liquidated quickly enough, the protocol may have to absorb more risk than intended.
Where bad debt comes from
Bad debt is debt that remains after collateral has been liquidated and still is not fully covered. In an idealised model, liquidations happen early enough and at good enough prices that lenders are made whole. In practice, several things can prevent that result: sharp price gaps, low liquidity, delayed oracle updates, liquidation competition, gas or network congestion, or collateral assets falling faster than liquidators can sell them.
Bad debt does not require a spectacular failure. It can appear simply because the liquidation buffer was too thin for the speed of the move. If the protocol relies on discounts or incentives to attract liquidators, those incentives may not be enough when markets are already dislocated. If collateral is difficult to sell, the liquidation process may realise less than the oracle-implied value. If the system is highly leveraged across many users, one stress event can produce a cascade of forced sales that further depress market prices.
This matters because bad debt is one of the clearest examples of the difference between protocol design and market reality. A ruleset can look robust on paper and still fail if several assumptions break at once. A careful user should ask not only whether the system has a liquidation mechanism, but whether that mechanism has been tested against rapid price movement, thin liquidity, and operational delays.
Governance, parameter risk, and incentives
DeFi lending protocols are often governed by token holders, delegates, or other decision processes that can change risk parameters over time. Those parameters may include collateral types, borrow caps, reserve factors, liquidation thresholds, and interest-rate curves. Governance is therefore not an abstract feature; it is part of the risk engine.
Parameter risk arises when governance changes make a previously understandable system behave differently. For example, lowering collateral requirements may increase capital efficiency for borrowers, but it can also reduce the buffer protecting lenders. Increasing borrow caps can help market access, but it can also magnify the scale of any failure. Changing rate curves may attract liquidity in the short term while increasing volatility in the long term. The point is not that governance is bad, but that it is active risk management with imperfect incentives.
Decision check: examine whether governance changes require delay periods, whether emergency powers exist, and whether those powers are narrow or broad. Short delay periods can make a protocol more responsive, but they can also reduce the time users have to react to risk changes. Broad emergency powers can protect the system in some incidents, but they also concentrate control. The relevant question is not whether governance is decentralised in slogan form, but whether the change process is predictable, bounded, and transparent enough for users to assess.
Smart contract risk and composability
Smart contract risk is the possibility that the code itself behaves unexpectedly. This can happen through logic bugs, upgrade mistakes, bad assumptions about token behaviour, or interactions between contracts that were each reasonable in isolation. DeFi lending systems are especially exposed because they often compose with price oracles, collateral wrappers, liquidity pools, and liquidation bots.
Composability is powerful but fragile. A lending protocol may be technically correct within its own code and still fail because a connected component breaks. For example, if collateral tokens have unusual transfer rules, if an external oracle stalls, or if an integrated liquidation path is unavailable, the lending market can become harder to maintain. Upgradable contracts can reduce some operational limits, but they also introduce governance and implementation risk around future changes.
A useful practical check is to separate three questions: has the core lending logic been audited or formally reviewed, are the surrounding dependencies simple or highly layered, and can the system be paused or modified in an emergency? None of these questions guarantees safety. They only help identify where the protocol’s real trust boundaries sit.
Failure scenarios and limits to what DeFi lending can do
The main failure scenario is not that one thing goes wrong, but that several reasonable assumptions fail together. A volatile market move can reduce collateral values. Oracle updates may lag. Liquidators may hesitate or face congestion. Governance may be slow to react. The combination can produce losses even in a system that seemed well collateralised moments earlier.
Another limit is that overcollateralised lending is not designed for every kind of borrower. It works best for users who can post sufficient collateral and who can tolerate forced liquidation if prices move against them. It is not a substitute for unsecured credit underwriting, nor is it a solution to every funding need. If a borrower cannot maintain the required buffer, the system’s rules will usually override the borrower’s preferences.
The most important limit is conceptual: DeFi lending can reduce some forms of counterparty risk by replacing discretionary credit decisions with rules, but it cannot remove market risk, oracle risk, governance risk, or code risk. In stress conditions, those risks can reinforce one another. That is why the right question is not whether a protocol offers yield, but which failures it is designed to absorb and which failures it simply inherits from the market.
Jurisdiction and risk caveat
This material is educational and general in nature. Legal treatment of crypto lending, collateral use, liquidation rights, and governance participation can differ by jurisdiction, and those differences may affect reporting, taxes, consumer protections, transfer restrictions, or enforceability of terms. Anyone considering actual use should verify local rules and obtain appropriate professional advice where needed.
Risk is also personal as well as technical. If you cannot monitor positions, understand liquidation thresholds, or tolerate rapid changes in collateral value and borrowing costs, DeFi lending may be unsuitable for your circumstances. The presence of on-chain rules does not remove the need for careful judgment.