The NPL Is Too Late: Why Stressed Loans Need an Asset Record Before Default

A loan's difficulties can begin well before day 91 or a formal non-performing classification. The earlier history may include a missed payment, failed direct debit, late salary or a business customer's lost major contract. It may also contain a promise to cure arrears, a changed payment date, a restructuring discussion, stale collateral information, a disputed amount or escalation to a specialist servicing team. Those events help explain the current position. Treating the NPL label as the beginning of the asset file discards evidence that a later servicer, buyer or reviewer may need.

India’s stressed-credit market is beginning to make that distinction more visible. On 3 September 2026, Mint reported that Asset Reconstruction Company (India) Ltd — Arcil — is moving beyond buying stressed loans and launching collections-as-a-service for banks and other financial institutions. The new service is not limited to non-performing assets. Arcil’s chief executive, Phanindranath Kakarla, specifically identified SMA-1 and SMA-2 loans as a major early-stress segment where institutions need collection support.

He also said Arcil already manages nearly 3.5 million NPA customers, giving the company an established collection infrastructure, team and geographic footprint. Mint: Arcil moves beyond buying bad loans, targets early-stage stress The move is more interesting than a new servicing product. It suggests that the operational boundary around distressed credit is moving earlier. The important question is no longer only:

Which loans have become NPLs?

It is increasingly:

Which loans are changing state, what happened, what was done, and what evidence supports the current position?

That is an asset-information problem.

Capture stress before non-performance

India’s Reserve Bank requires lenders to recognise incipient stress through Special Mention Account classifications. Under the RBI framework for relevant loan accounts:

• SMA-0
• Up to 30 days overdue
• SMA-1
• More than 30 and up to 60 days overdue
• SMA-2
• More than 60 and up to 90 days overdue

The existing framework generally treats a loan as non-performing after the relevant amount has remained overdue beyond the applicable 90-day threshold. Reserve Bank of India: Master Circular — Prudential norms on Income Recognition, Asset Classification and Provisioning pertaining to Advances That creates a visible sequence:

Performing → Payment missed → SMA-0 → SMA-1 → SMA-2 → NPA.

But this is only the regulatory classification layer. The real asset can follow many other paths. For example:

SMA-1 → Borrower cures arrears → Performing.

or:

SMA-2 → Payment arrangement → Partial cure → Further monitoring.

or:

SMA-2 → Restructuring discussion → Agreed modification.

or:

SMA-2 → No cure → NPA → Recovery / sale / enforcement.

A good servicing record needs to preserve the actual path.

Private-credit and distressed-asset systems often begin their serious data work after default. That is understandable. Once an asset becomes non-performing:

• recovery becomes more important;
• restructuring may begin;
• specialised servicing may be appointed;
• collateral is reviewed;
• legal action may become possible;
• investors may consider buying the exposure; and
• the asset may enter an NPL portfolio.

But starting the asset history at NPL classification throws away context. A reviewer may later need to know:

• when the first payment was missed;
• whether missed payments were isolated or repeated;
• whether the borrower contacted the lender;
• what explanations were given;
• whether promises to pay were made;
• whether those promises were kept;
• whether temporary payment arrangements existed;
• whether a restructuring proposal was discussed;
• whether collateral information was refreshed;
• whether documents were missing;
• whether another servicer previously handled the account; and
• what caused the final transition into default.

The NPL is the current state. It is not the whole history.

Imagine a loan with this sequence:

• 1 Jan
• Performing
• 5 Feb
• Instalment missed
• 10 Feb
• Borrower contacted
• 14 Feb
• Promise to pay recorded
• 20 Feb
• Partial payment received
• 5 Mar
• Next instalment missed
• 18 Mar
• Hardship request received
• 25 Mar
• Temporary arrangement proposed
• 10 Apr
• Arrangement not completed
• 15 Apr
• Collateral information refreshed
• 5 May
• Account reaches later delinquency stage

A field that says Status: Stressed cannot represent this. Neither can: Days past due: 75. Both are useful. Neither explains the asset. A structured lifecycle should preserve:

Event + Date + Actor + Amount + Source + Result + Current effect.

That is what makes the asset reconstructable.

Days past due are one of the clearest servicing measures. They answer: How long has an amount remained unpaid? They do not necessarily answer: Why? or What has happened since? Two loans can both be 45 days overdue and still be completely different servicing cases.

Loan A

The borrower missed one payment because of an administrative error. The borrower has sufficient funds. The arrear is scheduled to clear tomorrow.

Loan B

The borrower has lost employment. Several recent payments were late. The borrower has stopped responding. The secured property has not been inspected recently. Both may currently fall into the same overdue band. Their information needs are different.

Arcil’s move is important because it treats pre-NPA stress as a serviceable portfolio. The institution that owns the loan may not want to sell it. It may still want external help with:

• customer contact;
• arrears management;
• follow-up;
• payment collection;
• field operations;
• account monitoring;
• cure attempts; and
• escalation.

Mint reports that Arcil intends to provide these services to institutions that may not want to sell their stressed assets but still have NPA or early-stage stressed accounts requiring collections support. That separates two roles: Asset owner and Servicer / collection provider. The loan can remain on one institution’s balance sheet while another organisation handles part of its operational lifecycle. That distinction matters for asset infrastructure.

Separate ownership, servicing authority and contact history

A bank can own the loan. A third party can service it. A specialist collector can perform certain recovery activities. A law firm can handle litigation. A collateral agent can hold or administer security. A later investor can buy the asset. Those relationships may change over time. A simplified history might look like:

Bank originates loan → Bank services loan → Early stress detected → Specialist servicer appointed → Loan becomes NPA → ARC acquires loan → New servicer continues recovery.

The borrower obligation may be the same underlying asset. The parties around it change. A proposed useful evidence record should preserve both: Who owns the loan? and Who is responsible for servicing it now?

If servicing moves from Institution A to Servicer B, the loan should not lose its history. The transfer package may need:

• account identifier;
• borrower identity;
• contract;
• current balance;
• arrears;
• payment history;
• contact history;
• existing arrangements;
• complaints;
• disputes;
• collateral;
• notices;
• legal status;
• correspondence;
• consent or authority;
• prior servicing notes; and
• outstanding actions.

Without continuity, the new servicer may have to reconstruct the account from fragments. That creates cost. It can also create inconsistent treatment.

Collections teams often record promises to pay. A basic record might say: Promise to pay: Yes. That is too weak. A better record may include:

• Date promised
• Amount promised
• Payment date promised
• Communication channel
• Person contacted
• Authorised representative
• Reason
• Conditions
• Outcome
• Actual amount received
• Follow-up required

A promise that was made and kept is different from one repeatedly broken. The pattern can matter operationally. DaDepo does not need to score that pattern. But the history can be structured and visible.

A servicing file may contain dozens of communications:

• automated reminders;
• calls;
• emails;
• letters;
• app notifications;
• field visits;
• borrower responses;
• hardship requests;
• payment proposals;
• complaints;
• settlement discussions; and
• legal notices.

Not every communication belongs in a public or investor-facing data room. Personal data and confidential communications require appropriate controls. But the relevant servicing history can still need a structured record. For example:

Contact attempt → Successful / unsuccessful → Borrower response → Commitment / dispute / no action → Next step.

The asset history should not depend only on free-form notes.

Many servicing operations rely heavily on narrative notes. For example:

• Spoke with customer. Says business improving.
• Will try to pay next week. Follow up Monday.

That may be useful to the person who wrote it. At portfolio scale, it is difficult to aggregate. A structured representation could preserve:

• Contact date
• Contact method
• Contact outcome
• Borrower-reported reason
• Payment commitment
• Commitment amount
• Commitment date
• Next action
• Source note

The original note can remain as evidence. The structured fields make the portfolio easier to understand.

Ownership

Identifies who holds the relevant loan rights or economic exposure under the applicable structure. A change of servicer does not by itself transfer the loan.

Servicing

Identifies who is authorised to administer or collect the loan, under what mandate and from what effective date. Authority, borrower treatment and data access need their own evidence.

Distinguish delinquency, cure, restructuring and impairment

The distinction between delinquency and final credit deterioration matters. A missed payment can result from:

• temporary cash-flow timing;
• administrative error;
• bank-account change;
• payment-system failure;
• disputed amount;
• billing error;
• borrower hardship;
• seasonal business cash flow;
• fraud;
• insolvency;
• strategic non-payment; or
• many other causes.

A system should not infer the cause from the overdue status. It should preserve:

• Observed event:
• Payment not received

separately from:

• Reported reason:
• Borrower says account details changed

and separately from:

• Reviewed conclusion:
• [where an authorised institution records one]

Provenance matters.

Distressed-credit systems often focus on deterioration. Improvement matters too. A loan can cure. For example:

SMA-1 → Arrears paid → Current.

or:

SMA-2 → Restructuring implemented → Performance monitored.

The historical stress should not disappear just because the current state improves. A reviewer may still want to know:

• how long the account was delinquent;
• why it became delinquent;
• how it cured;
• whether previous cures occurred;
• whether the cure relied on refinancing;
• whether terms were modified; and
• whether the borrower is now performing under original or amended terms.

A current-status field should not overwrite the history.

Suppose the borrower cannot meet the original schedule. The lender changes:

• maturity;
• interest rate;
• instalment amount;
• repayment dates;
• collateral;
• covenant terms; or
• another material obligation.

The loan may become current under the new terms. That is different from curing under the original contract. The record should preserve:

Original terms → Stress → Restructuring proposal → Approval → Amended terms → Performance under amended terms.

This makes later review much clearer.

India is also preparing for a broader shift in bank provisioning. In April 2026, the Reserve Bank of India finalised an Expected Credit Loss framework that is scheduled to take effect on 1 April 2027. The new framework is forward-looking. It requires banks to assess whether credit risk has increased significantly since initial recognition. Where there has not been a significant increase, the framework uses a 12-month expected-loss approach.

Where credit risk has increased significantly, lifetime expected losses apply. The framework introduces three stages while retaining the existing NPA classification rules. Business Standard / Reuters: RBI finalises expected credit loss norms; rollout set for April 2027 This should not be confused with the SMA framework. SMA-0 / SMA-1 / SMA-2 and ECL Stage 1 / Stage 2 / Stage 3 are not simply interchangeable labels.

They serve different regulatory and accounting purposes. But both make one principle more important:

Credit risk changes before a loan reaches the final default bucket.

That increases the value of reliable lifecycle data.

Expected-loss models use much more than servicing notes. They can involve:

• probability of default;
• loss given default;
• exposure at default;
• forward-looking economic information;
• portfolio segmentation;
• significant increase in credit risk criteria; and
• institution-specific modelling and governance.

DaDepo should not calculate those regulatory values unless a future, expressly defined and properly governed service were ever created. But the evidence layer underneath credit assessment still matters. For example:

• payment history;
• delinquency transitions;
• restructurings;
• cures;
• collateral changes;
• borrower events;
• servicing transfers;
• historical recoveries; and
• documentation quality.

Those events can support the institution’s own regulated systems and professional assessments. A proposed role for DaDepo would be to help reviewers assemble a traceable asset record. Regulatory models and assessments remain with the institution.

Make portfolio events consistent without losing loan identity

Arcil says it already manages nearly 3.5 million NPA customers. At that scale, individual file knowledge cannot be the operating model. Portfolio systems need consistent answers. For every account: What is the current state? When did it enter that state? What event caused the transition? Who performed the last action? What evidence supports it? What is the next required review? That is the foundation for portfolio operations.

Mint reports that India’s asset reconstruction market is increasingly acquiring larger volumes of smaller-ticket retail loans from non-bank financial companies as corporate stressed assets become scarcer and retail stress increases. It reported that security receipts issued against retail loans represented 35% of total receipts in FY26, compared with 25% in FY25. Arcil itself acquired approximately ₹5,959 crore of financial assets in FY26, with NBFC-originated assets representing a large share of those acquisitions.

Mint: Arcil moves beyond buying bad loans, targets early-stage stress Large corporate restructurings can justify extensive manual attention. Millions of retail or SME accounts cannot. The data model therefore matters even more.

Suppose a lender transfers a servicing portfolio containing 100,000 loans. The portfolio summary may say:

• 100,000 accounts
• ₹X outstanding
• Average DPD: Y

But every underlying account can have a different:

• borrower;
• contract;
• balance;
• delinquency state;
• last payment;
• communication history;
• collateral;
• guarantor;
• dispute;
• restructuring;
• complaint;
• legal status; and
• recovery path.

The portfolio is an aggregation. The loan remains the operational asset. A system needs both levels.

At management level, a servicer may need:

Record item	Value or state
SMA-0 accounts	20,000
SMA-1 accounts	8,000
SMA-2 accounts	3,000
NPA accounts	5,000
Cured this month	2,300
Restructured	400

At account level, the reviewer may need to open one figure and see: Why is this loan SMA-2? The answer should trace back to the actual payment events and applicable classification state. Aggregation should never destroy provenance.

Keep balances, collateral and borrower treatment traceable

A loan balance changes with:

• scheduled principal;
• interest;
• fees;
• payments;
• reversals;
• waivers;
• write-offs;
• restructuring;
• settlement;
• recoveries; and
• legal costs where applicable.

A record saying: Outstanding: ₹500,000 is incomplete without:

• As of date
• Calculation source
• Components
• Last payment
• Adjustments

This becomes especially important when an asset is transferred between servicers or offered for sale.

A new servicer should not receive only the current balance. It should be able to reconstruct how the balance developed. For example:

Original principal → Scheduled amortisation → Payments → Missed instalments → Fees / adjustments → Restructuring → Recoveries → Current balance.

Without that history, disputes become harder to resolve. Buyers also have less confidence in a later portfolio sale.

Secured loans add another lifecycle. Collateral can change because:

• valuation expires;
• insurance lapses;
• title changes;
• registration changes;
• another lien appears;
• collateral is sold;
• an asset is damaged;
• guarantor support changes;
• perfection evidence is missing;
• enforcement becomes restricted; or
• the borrower substitutes collateral.

A loan can still be pre-NPA while its recovery position is already changing. That is another reason not to wait until default to organise the asset.

For a secured mortgage-style exposure: Loan timeline and Property / security timeline may develop separately. Example:

• Loan:
• Performing -> SMA-1 -> SMA-2

while:

• Collateral:
• Valuation current -> insurance expired -> new lien detected

The risk picture comes from both. A single delinquency field cannot represent it.

A third-party servicer may receive information about the loan. That does not automatically give it unlimited authority. The servicing arrangement may define which actions the servicer can take. For example:

• contact borrower;
• accept ordinary payments;
• negotiate within limits;
• grant short payment arrangements;
• waive certain fees;
• refer to legal;
• approve restructuring;
• initiate enforcement; or
• settle below a threshold.

Different decisions may require different approvals. A structured servicing record should therefore distinguish: Current servicer from: Permitted action and Approval authority. A proposed record could preserve that evidence. It should not decide what authority the servicer legally has.

Collections activity can generate disputes. A borrower may say:

• balance is wrong;
• payment was already made;
• payment was applied incorrectly;
• borrower is not the liable party;
• loan terms were amended;
• hardship arrangement was agreed;
• communication was inappropriate;
• collateral belongs to someone else; or
• the debt is legally disputed.

Those statements need a different state from ordinary collection activity. For example: Collection active and Balance dispute open can coexist. The system should not allow a call outcome to overwrite a formal dispute.

Servicing data often comes from several sources:

• lender core system;
• servicer system;
• payment provider;
• borrower communication;
• credit bureau;
• collateral register;
• legal counsel;
• field agent;
• external valuation; and
• uploaded documents.

Conflicts are inevitable. Suppose:

• Servicer system:
• Last payment 4 May

while:

• Bank ledger:
• Last payment 5 May

A good system does not silently select one. It records:

• Source A
• Value
• Timestamp

and:

• Source B
• Value
• Timestamp

then requires reconciliation. That is how provenance becomes operational.

The strongest asset-state model does not rely only on someone manually changing: Status = SMA-2. It preserves the events supporting the state. For example:

Record item	Value or state
Instalment due	1 June
Amount due	₹25,000
Amount received by 1 August	₹0

The institution’s classification process can then determine the applicable regulatory status. DaDepo does not need to replace that official calculation. A proposed evidence model could preserve the material supporting and explaining the institution’s classification; it would not perform that official calculation.

Prepare records that survive transfers and later recovery

Today’s operational servicing data can become tomorrow’s due-diligence file. If the lender later sells the portfolio, a buyer may want to know:

• historical delinquency;
• cures;
• restructurings;
• payment arrangements;
• collection attempts;
• complaints;
• disputes;
• collateral changes;
• recoveries;
• legal actions; and
• current servicer.

If those events were structured from the beginning, sale preparation becomes easier. If not, somebody must reconstruct them later.

NPL portfolios are notoriously document-heavy. A buyer may receive:

• loan agreements;
• amendments;
• account statements;
• security documents;
• valuations;
• collection notes;
• borrower correspondence;
• legal notices;
• court filings;
• settlement offers;
• data tapes; and
• servicing extracts.

The documents can be accurate individually while the package remains difficult to understand. The ideal NPL data room should not begin on the day of sale. It should emerge from the living loan record.

As a proposed direction for DaDepo, Arcil’s move is a useful model because it shows that the proposed evidence record should begin before the asset is officially non-performing. A loan or debt-portfolio record could contain several layers.

Record layer	Information to retain
Loan identity	internal loan identifier; lender; borrower; co-borrower; guarantor; facility type; origination date; original principal; currency; maturity; repayment frequency; governing agreement; current legal owner; current servicer.
Contract and terms	original agreement; schedules; interest terms; repayment schedule; fees; covenants; security; guarantees; amendments; waivers; restructuring documents; current effective version.
Current balance	principal outstanding; accrued interest; fees; arrears; overdue amount; disputed amount; written-off amount; recoveries; current total; calculation source; as-of timestamp; reconciliation status.
Payment history	instalment due date; expected amount; amount paid; payment date; payment source; allocation; partial payment; reversal; failed payment; returned payment; missed payment; reconciliation evidence.
Delinquency state	days past due; first overdue date; SMA or other applicable servicing classification; NPA/NPL classification where applicable; classification date; prior state; current state; transition reason; source system; reviewer; as-of time.
Servicing	current servicer; prior servicer; servicing agreement; transfer date; authority; contact strategy; queue or team; assigned case owner; servicing status; outstanding action; escalation; next review; transfer history.
Borrower contact	contact date; communication channel; successful contact; identity or authority check where required; borrower response; stated reason for arrears; promise to pay; promised amount; promised date; actual outcome; hardship request; complaint; dispute; next action.
Cure and restructuring	cure attempt; cure amount; cure date; temporary arrangement; restructuring proposal; approval status; amended terms; effective date; performance under amended terms; breach; termination; return to performing; monitoring period.
Collateral	collateral identifier; collateral type; ownership evidence; security document; registration; priority evidence; valuation; valuation date; insurance; guarantor; encumbrance; inspection; change event; enforcement state; release.
Legal and recovery	demand; notice; acceleration; counsel; court or tribunal; case reference; enforcement action; insolvency; settlement; judgment; recovery; write-off; closure; current legal state.
Provenance	source document; lender system; servicer system; payment source; borrower statement; registry; valuer; lawyer; AI-extracted field; user-confirmed field; professional review; effective date; version; last updated timestamp.

At portfolio scale, the same fields can be aggregated without losing the underlying account history.

A monthly data tape might say:

Record item	Value or state
Loan 1001	SMA-1
Loan 1002	SMA-2
Loan 1003	NPA

Next month:

Record item	Value or state
Loan 1001	Performing
Loan 1002	NPA
Loan 1003	NPA

If the first file is simply replaced by the second, important information disappears. The system should retain:

• Loan 1001
• SMA-1 -> Performing
• Reason: arrears cured
• Date: ...

and:

• Loan 1002
• SMA-2 -> NPA
• Reason: continuing non-payment
• Date: ...

The change is often more useful than the snapshot.

If servicing becomes a more modular service, asset data needs to move safely between organisations. A lender may: Originate then: Outsource early collections then: Bring account back in-house or Transfer to specialist NPA servicing or Sell the asset. Each transition creates a handoff. A handoff should not mean rebuilding the loan from PDFs. That is where a proposed durable evidence record can add value.

AI assistance and bounded DaDepo preparation

The following AI uses describe proposed assistance within this evidence model. They do not establish that DaDepo currently automates these workflows. Extraction and inconsistency flags would remain suggestions for review, not legal, credit, compliance or valuation decisions.

Loan-servicing portfolios contain repetitive documents and events. AI can help extract or reconcile:

• borrower and lender names;
• facility identifiers;
• due dates;
• balances;
• payment schedules;
• payment events;
• arrears;
• promises to pay;
• restructuring terms;
• collateral references;
• valuation dates;
• legal notices;
• servicer names;
• transfer dates;
• complaint references; and
• recovery events.

Across a portfolio, AI can also flag:

• inconsistent balances;
• missed payments not reflected in current status;
• a loan classified differently across systems;
• broken promises to pay;
• stale collateral valuations;
• servicing notes referring to missing documents;
• borrower disputes without a formal status;
• restructuring terms not reflected in the payment schedule;
• accounts transferred without complete histories;
• duplicate loan records;
• missing guarantor information; or
• a current servicer different from the servicing agreement.

AI should not independently determine:

• whether a borrower is in legal default;
• which regulatory classification must apply;
• whether an account should be moved to SMA or NPA status;
• whether a borrower is unwilling or unable to pay;
• whether a collection action is lawful or appropriate;
• whether a hardship request should be accepted;
• whether restructuring should be offered;
• whether collateral should be enforced;
• whether a security interest is valid or perfected;
• whether a borrower should receive additional credit;
• the ECL stage or regulatory provision;
• the expected recovery;
• or the legal enforceability of the debt.

Those remain lender, servicer, legal, accounting, risk and regulatory decisions.

DaDepo's supported contribution is private intake of supported documents and preparation of a draft for user review. Its Parties and contractual roles evidence table can record the party, specific contract, source-described role, document and clause reference, relevant source-stated date and review status. Suggested rows require the user's review and confirmation. The wider relationship models, automated monitoring, reconciliation and decision workflows discussed here are proposed capabilities, not a description of existing DaDepo functionality.

DaDepo does not need to become a collection agency. It does not need to run a call centre. It does not need to decide which borrower should be contacted next. The useful role is to help preserve the asset record around the servicing process. A loan can be represented as:

Loan → Borrower → Contract → Payment schedule → Payment events → Delinquency state → Servicing actions → Cure / restructuring / default → Collateral → Recovery.

Each transition should retain a source. A proposed DaDepo extension could help make the record easier to:

• understand;
• review;
• update;
• transfer between authorised participants;
• prepare for portfolio diligence;
• connect to supporting documents;
• expose missing evidence;
• compare across accounts; and
• preserve through later NPL sale or restructuring.

The proposed evidence record should answer: What is this loan? What is its current state? How did it get here? Who owns it? Who services it? What has the borrower paid? What has been agreed? What remains unresolved? Which evidence supports each answer? That is a natural extension of structured asset information.

Creating or reviewing a servicing or debt-portfolio Asset Passport does not mean that DaDepo has:

• originated the loan;
• confirmed the borrower’s creditworthiness;
• calculated an ECL;
• determined an ECL stage;
• determined SMA or NPA classification;
• performed regulated reporting;
• acted as lender;
• acted as servicer;
• acted as collection agency;
• contacted a borrower;
• determined a collection strategy;
• approved a payment arrangement;
• approved restructuring;
• provided hardship advice;
• determined legal default;
• accelerated a loan;
• perfected security;
• determined lien priority;
• valued collateral;
• initiated enforcement;
• provided debt counselling;
• made a credit decision;
• guaranteed cure;
• guaranteed recovery;
• purchased a stressed loan;
• operated an ARC;
• determined investor suitability; or
• provided legal, accounting, regulatory, credit, servicing, collections, restructuring, investment or valuation advice.

Important: DaDepo provides technology and information tools. It does not provide lending, servicing, debt collection, debt counselling, credit decisioning, ECL calculation, accounting, legal, regulatory, restructuring, enforcement, investment or valuation advice or services unless a specific service is expressly identified and lawfully provided. Loan classification, borrower treatment, collections, restructuring, security enforcement, provisioning and regulatory reporting remain subject to the applicable law, regulatory framework, lender policies, professional judgment and transaction-specific facts.

Review questions for an early-stress loan

Before an early-stage stressed loan is transferred to another servicer, escalated internally, restructured or later prepared for sale, ask:

1. Identity: Which exact loan or facility is this?
2. Borrower: Who is legally responsible for repayment?
3. Owner: Who currently owns the loan?
4. Servicer: Who is responsible for servicing it today?
5. Terms: Which contract and amendments are currently effective?
6. Schedule: What payment should have been made and when?
7. Balance: What is currently outstanding?
8. Arrears: What amount is overdue?
9. As-of date: When was the balance last reconciled?
10. First missed payment: When did delinquency begin?
11. Current state: What servicing or regulatory state is currently recorded?
12. Evidence: Which payment events support that state?
13. Contact: Has the borrower been contacted?
14. Response: What did the borrower actually say?
15. Promise: Has a promise to pay been made, and was it kept?
16. Hardship: Has a hardship or temporary-payment request been raised?
17. Dispute: Is the balance, liability or treatment disputed?
18. Cure: Has the account previously cured?
19. Restructuring: Have original terms been changed or is a modification under discussion?
20. Collateral: Is security information current?
21. Valuation: Is the latest collateral valuation still relevant?
22. Guarantor: Are guarantee details and contact information current?
23. Servicing transfer: Has another servicer previously handled the account?
24. Authority: What actions is the current servicer authorised to take?
25. Legal state: Have notices, acceleration or enforcement steps begun?
26. Portfolio: Is the loan included in a broader servicing, financing or sale pool?
27. Privacy and access: Who is authorised to see borrower and servicing information?
28. Provenance: Can every material current-state field be traced to the source that supports it?
29. History: Can the state transitions be reconstructed without relying on one person’s memory?
30. Next review: What event or date requires the account to be reviewed again?

If the answer to “what happened to this loan between performing and default?” cannot be reconstructed, the servicing record is incomplete.

The loan's history starts before the NPL label

India provides a particularly clear example because the SMA framework makes early delinquency states visible and Arcil is explicitly moving collection capability upstream. But the principle applies globally. Mortgage servicers. Consumer lenders. SME lenders. Auto-finance companies. Private-credit managers. Debt purchasers. NPL investors. Special servicers. All face versions of the same problem. The economically important asset state often changes before the contractual or regulatory end-state is reached.

A borrower misses a payment before a default is declared. Collateral deteriorates before enforcement begins. A restructuring conversation begins before documents are amended. A payment promise exists before cash arrives. A dispute begins before litigation. A loan cures before a portfolio sale. The asset is always moving.

A non-performing loan is easy to describe as a category. It is more useful to understand it as the result of a lifecycle.

Origination → Performance → Early warning → Missed payment → Delinquency → Servicing intervention → Cure / restructuring / continuing stress → Default → NPL → Recovery / sale / enforcement / resolution.

Every arrow can matter later. That leads to a broader rule for private-asset infrastructure:

The asset record should begin when the loan is created—not when the problem becomes severe enough to receive a new label.

Arcil’s move into early-stage collections makes that principle unusually visible. The next generation of distressed-credit infrastructure will not only know which loans are bad. It will know how each loan changed, who acted, what evidence supports the state, and what happened before default. That is a much stronger foundation for servicing, portfolio review, restructuring and eventual asset transfer.

Sources

• Mint: Arcil moves beyond buying bad loans, targets early-stage stress
• SEBI: Asset Reconstruction Company (India) Limited — Red Herring Prospectus, filed 2 September 2026
• Reserve Bank of India: Master Circular — Prudential norms on Income Recognition, Asset Classification and Provisioning pertaining to Advances
• Business Standard / Reuters: RBI finalises expected credit loss norms; rollout set for April 2027