Borrowing on a decentralised lending market almost always means locking up more value than you receive. This is not caution for its own sake; it is the only enforcement mechanism the contract has.
The contract has no recourse
Traditional lenders rely on identity, credit history and courts. A borrower who stops paying can be pursued, and that possibility is priced into an unsecured loan.
A lending contract knows an address and nothing else. It cannot sue, cannot garnish income and cannot verify who is behind a transaction.
Collateral held inside the contract replaces all of that. The lender's protection is not a promise but an asset the contract already controls.
The buffer absorbs price movement
If a loan were fully collateralised at exactly its value, any decline in the collateral would immediately leave the pool short.
The excess margin buys time. It allows the collateral to fall some distance before the position becomes unsafe, and it gives liquidators a window to act while there is still enough value to cover the debt.
How large that buffer is varies by asset. Collateral that trades in deep markets and moves modestly earns a higher borrowing allowance than a thin token that can gap downward.
Liquidation is a paid job, not an automatic event
Nothing happens on its own when a position crosses its threshold. Someone has to send a transaction that repays part of the debt and claims collateral in return.
The incentive is a discount on the collateral seized, sized so that liquidators find the work profitable even after paying transaction fees and accepting price risk.
That discount is a real cost to borrowers and a reason positions are usually kept well clear of their limits rather than run close to them.
Undercollateralised lending needs something extra
Lending less than full collateral is possible, but only by importing a constraint from outside the contract's own logic.
Some designs restrict the borrowed funds to a whitelisted set of contracts, so the money can be used but never withdrawn. Others rely on the borrower being a known institution bound by an off-chain agreement.
Both approaches reintroduce trust in a specific party, which is the trade the overcollateralised model was built to avoid. That trade-off, rather than any technical limit, is why the conservative design remains the norm.
Capital efficiency is the cost being paid
Requiring surplus collateral means a large amount of value sits locked to support a smaller amount of borrowing, which is inefficient by any conventional lending measure.
The design accepts that cost because it removes the need to know anything about the borrower, and it is the reason the same contract serves an individual and an institution identically.
Efforts to narrow the gap focus on better risk measurement rather than on trusting borrowers: isolated markets per collateral type, borrowing caps, and thresholds tuned to how each asset behaves under stress.