Loan underwriting is fundamentally a data problem: the quality of a credit decision is a direct function of the quality, accuracy, and completeness of the data that informs it. For decades, Indian lenders have made that decision with incomplete data, self-declared income, fabrication-prone PDFs, and bureau scores that capture debt history but not cash flow behavior.
Account Aggregator data changes the inputs available for underwriting. This guide covers how to build an AA-powered underwriting model, what data elements to use, how to structure the analytical framework, and how to operationalize it within existing credit decision systems, enabling faster and more accurate credit decisions. To understand this foundation clearly, here’s what an account aggregator is and how it works in lending.
The Underwriting Data Problem Traditional Lenders Face
Traditional underwriting in India relies on three primary data inputs: the borrower’s stated income (on the application form), one to three months of PDF bank statements, and a bureau score from CIBIL, Experian, CRIF, or Equifax.
Each input has documented weaknesses. Stated income is unverified. PDF statements can be altered. These are common red flags in bank statement analysis. Bureau scores reflect the past, not the present, and miss cash flow nuance entirely.
The result is credit decisions that are less accurate than they need to be, leading to either excessive risk rejection (declining creditworthy borrowers with thin bureau files) or inadequate risk assessment (approving borrowers whose real financial position is worse than the declared data suggests).
AA data fixes all three gaps: it verifies income at source, prevents tampering, and provides a current behavioral view.
How AA Data Transforms the Underwriting Input Set
When AA data is integrated into the underwriting workflow, the available inputs expand substantially. A 12-month transaction feed from the borrower’s primary bank accounts provides the following:
Verified income: The actual salary or business receipt credits, not self-declared amounts.
Current obligation landscape: Real EMI debits, reflecting actual debt service burden.
Cash flow behavior: Average monthly surplus, balance volatility, frequency of low-balance periods.
Financial discipline signals: Bounce frequency, overdraft usage, and systematic savings behavior.
Income trend: Whether income is growing, stable, or declining over the assessment window, including income identification, EMI detection, and cash flow scoring. This is exactly how lenders analyze account aggregator transaction data in detail.
None of these inputs is available from a bureau report or a self-declared application form. Together, they enable a substantially more accurate credit decision.
Building a Credit Decision Model on AA Data
Income Verification Layer
The first step is income identification: distinguishing actual income credits from capital transfers, loan proceeds, or one-time receipts. For salaried borrowers, this is pattern-based: recurring credits from identifiable employer counterparties. For the self-employed, it requires ML-based classification trained on Indian bank transaction narrations.
After identifying income credits, calculate average monthly income, stability score, trend direction, and primary income source.
The verified income replaces or supplements self-declared income on the application form, providing the foundation for debt-service capacity calculation.
Cash Flow Volatility Scoring
Cash flow volatility is a predictor of repayment risk that bureau scores cannot capture. A borrower with stable, predictable cash flow is a lower repayment risk than one with identical average income but high month-to-month variability.
Calculate the monthly inflow-outflow ratio (average across the 12-month window), standard deviation of monthly surplus, number of months with closing balance below a threshold (e.g., one month’s EMI), and the ratio of peak balance to average balance.
A cash flow volatility score synthesises these metrics into a single risk indicator. Borrowers with high volatility scores may qualify for smaller loan amounts, shorter tenures, or higher pricing to compensate for the elevated repayment uncertainty.
Obligation-to-Income (OTI) Calculation
The obligation-to-income ratio, the proportion of monthly income committed to debt service, is the primary affordability metric. With AA data, this can be calculated with greater precision than bureau data allows.
AA data is shared in a structured, machine-readable format that eliminates the need for manual parsing. This standardization is defined within the Sahamati AA ecosystem documentation. This allows underwriting systems to process financial data instantly and consistently across institutions.
From transaction data: identify all recurring fixed-amount debits to financial institution counterparties (EMI payments), calculate total monthly debt service, and express it as a percentage of verified monthly income.
Most lending policies set OTI limits: a borrower with an OTI above 60–70% after the proposed loan EMI is typically declined. With AA data, this calculation reflects actual obligations, not the bureau’s 30–90 day lagged view.
Compliance Framework for Using AA Data in Credit Decisions
Using AA data in credit decisions requires compliance with several regulatory frameworks:
Use AA data only for the consented purpose. If used for underwriting, do not use it for marketing or other purposes.
Explainability: Borrowers have the right to understand why a loan was declined. If AA data contributed to the decision, the lender must be able to articulate what specific data points triggered the negative outcome in plain language.
Fair lending: AA data must not be used as a proxy for protected characteristics. Income volatility patterns must not systematically disadvantage specific demographic groups. Lenders should audit their AA-based models for disparate impact.
Data retention: Transaction data obtained through AA must be deleted within the retention period specified in the consent artefact. Implement automated data deletion workflows tied to consent artefact expiry, aligning with India’s evolving data protection framework. The RBI and DPDP compliance requirements for using account aggregator data are explained in detail here.
TAT and Default Rate Benchmarks: Early AA Adopter Data
Lenders that have integrated AA into their underwriting workflows report consistent improvements across key performance metrics:
Turnaround time: Assessment TAT drops from an average of 3–5 working days (with manual PDF processing) to under 4 hours when AA data replaces the document collection and review steps.
Early default rates: AA-underwritten loans show measurably lower 30-day and 90-day default rates compared to PDF-underwritten loans from the same cohort. The differential is attributed primarily to the elimination of fabricated statement fraud and more accurate income verification.
Approval rates on thin-file borrowers: Lenders using AA data report being able to approve a higher proportion of thin-file applicants, borrowers with limited bureau history but demonstrable cash flow strength, without increasing portfolio risk.
Key Takeaways
- AA data provides verified income, real-time obligation mapping, cash flow behavior scoring, and financial discipline signals, all unavailable from bureau reports or PDF statements.
- An AA-powered underwriting model should be structured in three layers: income verification, cash flow volatility scoring, and OTI calculation.
- Purpose limitation, explainability, fair lending audits, and data retention controls are mandatory compliance requirements for using AA data in credit decisions.
- Early AA adopters report TAT reduction from days to hours, lower early default rates, and higher approval rates on thin-file borrowers.
Frequently Asked Questions
No, they serve different analytical purposes. Bureau scores capture debt history; AA data captures cash flow behavior. The strongest models use both. AA data is particularly valuable for thin-file borrowers where bureau history is limited.
For self-employed borrowers, AA data is especially valuable because formal income documentation is often unavailable or unreliable. Cash flow-based income identification from 12 months of current account transactions provides a verified income estimate that is often more accurate than an IT return-based assessment.
Industry practice in India sets OTI limits between 50–65% for most retail loan products. With AA data’s higher accuracy in obligation identification, lenders may find that previously invisible obligations push actual OTI above policy thresholds, leading to better-quality decline decisions.
The AA data pull takes 60–90 seconds from consent to data delivery. Analysis of the structured feed takes 5–15 seconds, depending on the complexity of the model. The end-to-end time from consent initiation to underwriting completion is typically under 3 minutes for automated workflows.
AA can pull data from multiple FIPs simultaneously. The analysis engine consolidates data across accounts, calculating total income, total obligations, and net cash flow across all accounts, providing a complete financial picture even when the borrower’s financial activity spans multiple institutions.
Conclusion
AA-powered underwriting is not a marginal improvement on existing credit assessment. It is a structural upgrade to the quality of information available for credit decisions. Lenders that adopt AA don’t just process faster; they make better decisions using reliable, real-time data beyond bureau scores.
The implementation path is straightforward. The business case is clear. The regulatory alignment is strong, improving both accuracy and efficiency. A closer look at the measurable ROI of using an account aggregator in lending shows the full business impact. The only remaining question is sequencing: which use cases to integrate first, and how quickly to scale.
