How A Lender Sets Its Acceptance Cut-Off

Applications are ordered by risk and then divided by a threshold. Where that threshold sits is decided by the lender for its own reasons, and moving it changes who is accepted without anything changing on any file.

The score orders, the cut-off decides

A scoring model ranks applicants by the estimated likelihood of a defined bad outcome. It produces a position in a queue, not a verdict on the application.

The lender then chooses a point in that queue. Everyone above it is accepted, subject to policy rules, and everyone below is declined or referred.

Because the ranking and the threshold are separate, the same applicant with the same score is acceptable to one lender and not to another.

The threshold is set on expected economics

For each band in the queue, the lender can estimate the proportion likely to go bad and the revenue the rest will generate. The cut-off is placed where the combination stops paying.

Raising the price moves that point, because higher revenue supports higher expected losses. This is why an applicant below a prime cut-off may be offered a more expensive product instead.

Costs of acquisition, servicing and collections all sit in the same calculation, so a lender with a cheaper operation can accept applicants a costlier rival must refuse.

Policy rules sit outside the score

Above the cut-off, applications still pass through absolute rules: minimum age, residency requirements, recent insolvency, existing arrears with the same lender, or an undischarged formal arrangement.

These are treated as knock-outs rather than as points, because the lender has decided the case regardless of how favourably the rest of the file reads.

An applicant declined by a policy rule cannot compensate with a stronger score, which is why some declines appear disproportionate to the file behind them.

Cut-offs move with conditions

Thresholds are reviewed as arrears trends, funding costs and growth targets change. Tightening is achieved by moving the line up rather than by announcing a change in criteria.

Nothing about this is visible externally, so a borrower reapplying after a period of tightening can be declined on a file that has improved.

The same mechanism operates in reverse during expansion, which is why acceptance for marginal applicants appears to loosen and tighten in waves across the market.

Referral sits between the two answers

Applications near the line are often referred rather than decided automatically, which sends them to an underwriter with authority to look beyond the score.

Referral is expensive, so lenders limit how many cases enter it. A narrow referral band is a cost decision as much as a risk one.

Rules on automated decisions, on what must be disclosed to a declined applicant and on the right to request human involvement differ between jurisdictions and continue to develop.