Context
The timing of information can change a decision
In vehicle finance, a collateral requirement can add steps to completing the loan. Exempting an eligible customer can reduce that process burden, but requires an appropriate risk basis.
An application-process update brought information forward, including the customer’s intended loan amount. My role was to adapt the accompanying risk policy; I did not initiate the broader process redesign.
The opportunity was to use information available earlier to make a more appropriate preliminary eligibility decision and improve consistency with final review.
Key insight: customers did not always borrow up to the permitted amount
I examined the difference between the actual down payment at final review and the minimum down payment approved at the preliminary stage. In historical data, larger positive differences were associated with lower risk.
That relationship suggested a potentially useful signal: some customers chose to contribute more upfront cash than the minimum requirement. However, the actual final-review amount was only known later. To use the idea earlier, I needed a signal available at the time of the preliminary decision.
The newly available intended loan amount provided that bridge. It allowed an intended down payment to be inferred and compared with the policy minimum.
- Preliminary eligibility uses information already available
- Actual final-review down payment becomes observable later
- Historical analysis identifies a relationship after the fact
- Preliminary eligibility can additionally consider intended loan amount
- Intended down payment can be estimated earlier
- Policy can assess an earlier proxy for that relationship
The new signal was an intention available earlier, not the final realised behaviour. The policy therefore used a proxy rather than assuming the future amount was already known.
Approach
01Analyse the gap relative to the policy minimum
I compared customers relative to their own preliminary down-payment requirements, rather than looking only at the absolute amount of cash contributed. This distinguished “meeting the requirement” from “contributing more than required”.
The observed association supported exploring the signal. It did not prove that increasing a down payment caused lower risk, or that the signal could rule out fraud.
02Translate the idea into an earlier proxy
Using vehicle price and intended loan amount, I inferred the intended down-payment share, then compared it with the preliminary minimum. That comparison informed how selected applicants could be considered within the credit and repayment-capacity eligibility matrix.
Because intended amounts could change before final review, I treated this as application-stage information rather than a guaranteed future commitment.
| Symbol | Meaning |
|---|---|
| Vehicle price | |
| Intended loan amount | |
| Intended down payment as a share of vehicle price | |
| Minimum required down-payment share | |
| Difference above the minimum, in percentage points |
A fictional vehicle costs 100 units and the customer intends to borrow 65, so the implied down payment is 35% — 10 percentage points above a 25% minimum. Financed amount equals price minus down payment; fees and other adjustments are omitted. The calculation creates a relative signal; it does not show the actual eligibility threshold or mean that this applicant would receive an exemption.
Technical noteDistinguish the historical relationship from the operational signal+
The first expression describes the historical behaviour examined. The second describes the earlier proxy used to inform the policy. They are related but not identical: intention can change, and the later amount is not available at the earlier decision point.
03Adjust the matrix while preserving the boundary
I refined the rating-based eligibility matrix to extend exemptions to selected applicants. Mandatory collateral conditions remained in place. A favourable additional signal could influence eligibility within the permitted scope; it could not override those requirements.
- Applicant information is first checked against mandatory collateral conditions.
- If a mandatory condition applies, collateral is required.
- If not, the eligibility matrix is evaluated, using credit and repayment-capacity ratings and the intended down-payment gap.
- The matrix produces the eligibility decision.
Conceptual rule boundary only. Internal conditions, thresholds and execution order are not reproduced.
04Validate the rule and monitor the affected group
I wrote the requirements, independently reproduced the relevant rules in SQL and compared batch-test results to investigate discrepancies before implementation.
Monitoring then followed the applicants who became eligible under the revised policy, alongside overall conversion through the new flow — keeping the two scopes distinct.
OutcomeReported outcome
More pre-approved applicants went on to final review
Comparing the new application flow with the old flow over the same five-week window, 73.0% of pre-approved applicants submitted for final review within 10 days, against 64.8% — a difference of 8.2 percentage points. The new-flow figures include the applicants who became newly eligible for exemption under the revised policy.
New flow vs old flow, same observation window. The comparison reflects the process update and the revised eligibility policy together, so the 8.2 pp difference should not be read as the effect of the exemption rule alone.
Technical noteHow the conversion figure is defined+
Measured separately for the old and new flows over the same window: 64.8% and 73.0%. The difference is reported in percentage points (73.0 − 64.8 = 8.2 percentage points), not as a relative uplift.
At the time of review, no obvious adverse risk signals had emerged in the monitoring discussed. Early monitoring does not establish long-term risk performance; the result is presented as a scoped operational observation rather than a causal estimate.