THE BUSINESS APPLICATION
Where and how to use it
Use it to distinguish applications suitable for normal review from those needing further evidence, and to compare the risk of different segments. Combine it with exposure, expected loss and contribution when assessing a proposed facility; a low default estimate alone does not make a loan profitable.
An MSME lender is considering more unsecured working-capital loans. The team compares applicants with similar repayment sources, investigates the reasons behind elevated estimates and refers incomplete cases for review rather than automatically rejecting them.
From evidence to a decision
- 01Application dateFreeze the available evidence
- 02Learn from historyCompare earlier loans and outcomes
- 03Estimate riskPD over a stated horizon
- 04Apply your policyReview, refer or decide
DATA REQUIREMENTS
What records does it need?
These are the records your team would bring together for this analysis. The exact fields and history needed depend on your lending products, the question you want to answer and the period you want to assess.
| Record category | What it contains | Why the detail matters |
|---|---|---|
| Application snapshot | Application and borrower IDs, decision date, requested amount, tenure, business characteristics and permitted financial features. | Only facts available at the original decision time. |
| Facility and performance history | Booked terms, dated payments, arrears and observed default events. | Past loans with a defined period of repayment history. |
| Outcome definition | Default policy, observation window and last observed date. | Unobserved or immature outcomes remain unknown. |
Past loan outcomes help assess how well an estimate reflects your borrowers. For a new decision, use only the information available at that time; later repayments help you review the result afterwards.
Understand data readiness →WHAT YOU RECEIVE
The output
A horizon-specific probability or score, contributing factors and diagnostics. A separate policy determines what action is permitted.
WHAT TO WATCH
The limitations
Outcomes on accepted loans do not automatically describe rejected applicants. A model fitted elsewhere is not evidence of suitability for your MSME portfolio.
FOR RISK & ANALYTICAL SPECIALISTSHow the analysis works+
The modelling approach
Approaches include regularised logistic regression and interpretable scorecards. They learn a relationship between information available at application and later observed outcomes. Predicted probabilities must be calibrated and evaluated on later periods that were not used for fitting.
What your risk team should review
Calibration, discrimination, temporal holdout performance, missing-data behaviour, segment stability and selection bias.
The right approach depends on your portfolio and available history. Review the fit to your borrowers, the reliability of the estimates and the effect of missing information before using the result in a lending decision.