Technology

System-Based Asset Classification: NBFC Daily NPA Guide

|Posted by Hitul Mistry / 12 Nov 26

System-Based Asset Classification: How NBFCs Automate Daily NPA Tagging Without Manual Overrides

Last reviewed: November 2026

System-based asset classification means your loan system, not a person, tags each account as standard, SMA or NPA at every day-end using repayment data and RBI rules. Provisions and income recognition then follow automatically from that tag.

This guide is for CROs, CFOs and Heads of Credit Operations at NBFCs. It explains what the rules require, how a daily engine should work and where manual overrides tend to slip in.

Key Takeaways

  • NBFC classification rules sit in the NBFC IRACP Directions, 2025, updated as on 1 October 2026.
  • SMA and NPA status must be set in the day-end process for each calendar date, and classification is borrower-wise.
  • An NPA is upgraded only when all arrears of interest and principal are paid, across every facility.
  • Base Layer NBFCs reached the 90-day NPA norm on 31 March 2026, so one threshold now applies across layers.

Is System-Based Classification Mandatory for NBFCs?

Not through a single named mandate, but the rules leave little room for anything else. Banks have had an explicit automation requirement since September 2020, and their April 2026 directions carry an automation annex.

For NBFCs, the obligation is built from several pieces. Together they make manual classification very hard to defend.

  • IRACP Directions. Day-end tagging, borrower-wise contagion and strict upgrade rules apply to every NBFC.
  • Base Layer at ₹500 crore and above. The 2026 cybersecurity and technology directions list system-based flagging of SMA and NPA accounts among expected MIS outputs.
  • Middle Layer and above. The same directions require that data moving between critical applications is automated, with no manual intervention or modification.

What Can NBFCs Borrow From the Bank Standard?

The bank framework is a useful benchmark even where it does not bind you. Its core idea is coverage: every borrowal account, including temporary overdrafts, sits inside the automated system regardless of size or product.

Exceptions are allowed only under delegated authority, with a record of who approved them. An NBFC that adopts the same discipline will find supervisory reviews and statutory audits far easier.

Which Rules Must the Engine Apply?

The engine is a rulebook in code. This table maps each rule in the IRACP Directions to the logic it needs.

RuleSourceEngine Logic
Overdue on the due datePara 18Flag any unpaid amount at the day-end for the due date
Day-end classificationPara 19Run SMA and NPA tagging for the calendar date, even if the batch runs after midnight
SMA stagesPara 19 illustrationSMA-1 on completing 30 days overdue, SMA-2 on 60, NPA on 90
Borrower-wise statusPara 23Propagate NPA to all facilities of the same borrower
UpgradeParas 24–25Upgrade only when all arrears across all facilities reach zero
Sub-standard to doubtfulParas 45–46, 52–53Age NPAs for 12 months at ML and UL, 18 months at BL
ProvisioningParas 32, 48, 55, 58Apply rates by asset class and layer
Income recognitionIRACP DirectionsStop accrual on NPAs and recognise income only on realisation
Ind AS floorParas 34–36Compare ECL with prudential provisions and route any shortfall to an Impairment Reserve

Which Products Need Special Logic?

A single DPD counter will not cover every product. Several cases need their own rules and test data.

  • High-frequency repayments. Weekly or fortnightly microfinance and daily-collection loans must count days from the oldest unpaid instalment, not the latest.
  • Leases, hire purchase, demand loans and bills. The Directions set separate overdue tests for each, so map them explicitly.
  • Restructured accounts. Upgrades after restructuring also depend on the Resolution of Stressed Assets Directions, not just on arrears.
  • Moratoriums and holidays. Due dates in the schedule must reflect any approved relief, or the engine will tag accounts that are not overdue.

Build a test pack for each case and rerun it after every product change. It is the cheapest insurance against a misclassified book.

Loan agreements also matter. For loans sanctioned from 31 December 2021, agreements must state due dates and examples of SMA and NPA dates, so the engine and the contract must agree.

How Does a Daily NPA Engine Work?

A sound engine runs as a fixed sequence inside the day-end batch. Each step writes an audit record.

  1. Close the day's receipts. Post every repayment, reversal and bounce received for the calendar date before classification starts.
  2. Compute days past due. Measure each account from its oldest unpaid due date.
  3. Assign status. Apply SMA and NPA thresholds by product and layer.
  4. Apply contagion. Group accounts by borrower and move every facility to the worst status.
  5. Test upgrades. Upgrade only borrowers whose arrears are fully cleared across all facilities.
  6. Calculate provisions. Apply asset class, age, security value and layer rates.
  7. Post to the ledger. Push provisions and income reversals to the general ledger in the same run.
  8. Publish outputs. Feed SMA lists to collections and classification to MIS and regulatory returns.

What Should the Daily Output Show?

Leadership needs a short exception view, not a full extract. Four numbers tell you whether the engine can be trusted.

  • Status movements. Accounts moving into SMA-1, SMA-2 and NPA since the last run, by product.
  • Upgrades. Accounts upgraded, each with proof that arrears reached zero.
  • Exceptions. Manual changes, with approver, reason code and value affected.
  • Reconciliation. Classified outstanding against the general ledger, with any difference explained.

A rising exception count is an early sign that the rules and the products have drifted apart. Treat it as a risk indicator for the board risk committee.

Data must flow between these steps without manual edits. The principles in our guide to straight-through processing apply directly to this batch.

Can your team prove that no one edited an NPA tag by hand last quarter?

Talk to Our Lending Technology Specialists

Visit Digiqt to see how we build loan management systems with rule-driven classification.

How Does One Borrower Move Through the Engine?

Dates matter more than amounts in classification. This illustrative case follows a borrower at a Middle Layer NBFC with two loans.

Calendar DateEventStatus at Day-End
5 March 2026Personal loan EMI not paidOverdue (SMA-0)
4 April 202630 days continuously overdueSMA-1
4 May 202660 days continuously overdueSMA-2
3 June 202690 days continuously overdueNPA, and the current LAP account also becomes NPA
20 June 2026Borrower pays one of three overdue EMIsBoth accounts stay NPA
10 July 2026All arrears on both loans clearedBoth accounts upgraded to standard

Check the counts: 26 days remain in March after the 5th, so 30 days lands on 4 April, 60 on 4 May and 90 on 3 June. The 20 June payment does not upgrade anything because arrears remain.

Now the provision. If the personal loan has ₹2.8 lakh outstanding and the LAP ₹9.2 lakh, the sub-standard provision is 10% × ₹12 lakh = ₹1.2 lakh.

The lesson for leadership: contagion and upgrade logic drive most of the money. A manual tag on either loan would misstate both.

Collections should act well before day 90. An early-stage collections agent can work SMA-0 and SMA-1 lists the morning after each run.

Where Do Manual Overrides Creep In?

Overrides rarely look like overrides. They usually hide in data fixes and timing.

  • Back-dated receipts. Posting a payment to an earlier date after classification has run can silently reverse a downgrade.
  • Customer ID gaps. Duplicate customer records break borrower-wise contagion.
  • Spreadsheet upgrades. Upgrading on a partial payment because an officer believes the borrower will pay breaches the rules.
  • Batch timing. Running classification before late-day receipts are posted produces wrong SMA lists.

Control each one with a rule, not a reminder. Exceptions should need maker-checker approval, a reason code and a permanent log that internal audit can query.

Prioritisation turns clean data into recovery. A collections prioritisation agent can rank SMA accounts by risk and value once the tags are reliable.

What Is Changing Next?

Provisioning is moving towards expected credit loss across the system. Commercial banks shift to an ECL-based framework from 1 April 2027, under directions issued on 27 April 2026.

Ind AS NBFCs already use ECL, with prudential norms as a floor. Since February 2026 they may also reflect default loss guarantee cover in ECL, which needs portfolio-level data from the same engine.

An expected credit loss intelligence agent can sit on top of clean classification data.

Ready to replace manual NPA tagging with an engine your auditors can test?

Talk to Our Lending Technology Specialists

Visit Digiqt to see how we build lending platforms and AI agents for credit operations.

Sources

Read our latest blogs and research

Featured Resources

Technology

Scale Based Regulation for NBFCs: What Each Layer Changes

Scale based regulation for NBFCs in 2026: the ₹1 lakh crore upper layer test, middle layer duties and the system changes CROs and compliance heads must plan.

Read more
Technology

Collection Efficiency for Lenders: Formulas and Levers

Collection efficiency explained for lenders: formulas with worked arithmetic, bucket metrics, why public benchmarks mislead, and the levers that improve it.

Read more
Technology

Vintage Analysis of a Loan Portfolio: A Lender's Guide

Vintage analysis of a loan portfolio explained: how to build cohort curves, read months-on-book data, link it to Ind AS 109 ECL and spot weak cohorts early.

Read more

About Us

We are a technology services company focused on enabling businesses to scale through AI-driven transformation. At the intersection of innovation, automation, and design, we help our clients rethink how technology can create real business value.

From AI-powered product development to intelligent automation and custom GenAI solutions, we bring deep technical expertise and a problem-solving mindset to every project. Whether you're a startup or an enterprise, we act as your technology partner, building scalable, future-ready solutions tailored to your industry.

Driven by curiosity and built on trust, we believe in turning complexity into clarity and ideas into impact.

Our key clients

Companies we are associated with

Life99
Edelweiss
Aura
Kotak Securities
Coverfox
Phyllo
Quantify Capital
ArtistOnGo
Unimon Energy

Our Offices

Ahmedabad

B-714, K P Epitome, near Dav International School, Makarba, Ahmedabad, Gujarat 380051

+91 99747 29554

Mumbai

C-20, G Block, WeWork, Enam Sambhav, Bandra-Kurla Complex, Mumbai, Maharashtra 400051

+91 99747 29554

Stockholm

Bäverbäcksgränd 10 12462 Bandhagen, Stockholm, Sweden.

+46 72789 9039

Malaysia

Level 23-1, Premier Suite One Mont Kiara, No 1, Jalan Kiara, Mont Kiara, 50480 Kuala Lumpur

Lewes

16192 Coastal Highway, Lewes, Delaware 19958, USA

software developers ahmedabad
ISO 9001:2015 Certified

Call us

Career: +91 90165 81674

Sales: +91 99747 29554

Email us

Career: hr@digiqt.com

Sales: hitul@digiqt.com

© Digiqt 2026, All Rights Reserved