Bank Loan Approval with Regulatory Explainability
The Risk Committee of a commercial bank needs to automate the granting of personal loans while strictly complying with financial regulation, which requires explaining to the customer and the regulator the exact reason for any refusal.
Where the money comes from
What to do now
- 1Define with risk and compliance the permitted variables and the explanations given to the customer.
- 2Train the tree and validate it with loans from previous years.
- 3Pilot in one channel with decisions reviewed by analysts.
For your committee
One page with the decision, the impact, the risks, the steps and the KPI to track.
The print dialog opens: choose "Save as PDF".
Watch out: more volume is not always more profit
Approving more to grow the portfolio: if the price does not reflect the risk, marginal loans bring income today and losses tomorrow. The goal is risk-adjusted profitability, not volume granted.
KPI for the committee
12-month default rate by score band and approval rate.
Detail for your data team
Proposed technique: Decision Trees
Why: Decision Trees translate data into explicit business rules ('If income > €2,000 AND debt ratio < 30% -> Approve'). It is the ideal model when transparency and legal justification matter more than gaining an extra 1% of accuracy.
Cost of getting it wrong: Approving loans that default erodes the bank's capital; refusing without a clear explanation leads to regulatory penalties.
Data needed: Core Banking: Direct-deposit income, average account balance and tenure as a customer.; Credit Bureau (CIRBE / ASNEF): History of prior defaults and total debt with other institutions.; Customer Application: Amount requested, term and purpose of the loan.