Lending Infrastructure

Loan tape: meaning, fields and a worked example

See what a loan tape contains, with a synthetic three-loan example, a field dictionary and checks for cut-off dates, balances and history before you share it.

Loan tape: meaning, fields and a worked example: cover art
Contents 11 sections
Quick answer

A loan tape is a structured file of loan-level data used to review a portfolio. Each row in the example below is one loan. The columns describe its terms, balances, status and history at an agreed cut-off. The useful starting point is a field dictionary that the recipient understands and finance can reconcile.

If you prepare an NBFC’s funding or pool-review data, agree the cut-off and field definitions before you build the file. Two columns called “overdue” can carry different amounts if one includes charges or uses a different reporting date. A clean spreadsheet can still describe the wrong book.

The example below shows the format on three synthetic loans. It is a starting point for a discussion with a recipient, not an RBI-prescribed template or a demonstration of a Lokta export.

What is a loan tape?

A file that lets a buyer, investor, rating agency or funder examine loans individually and compare them as a pool. A typical tape includes identifiers, terms and current balances, with historical measures where the review needs them.

RBI’s directions refer to loan-level data and underlying exposures. The recipient’s specification determines how that information is organised in your file. Keep the field names and definitions together so a reader doesn’t have to infer what an unfamiliar column means.

What does a loan tape example look like?

This illustration uses a cut-off of 21 October 2026, after day-end. Every money value is in INR, without a multiplier. Overdue amount includes overdue principal and interest, with charges excluded. Peak DPD is the highest observed DPD from 22 October 2025 through 21 October 2026, inclusive. These are invented loans and identifiers, with no borrower personal data.

Loan IDPrincipal outstanding (INR)Overdue amount (INR)DPD at cut-offPeak DPD in window
SYN-0011,00,0001,00018
SYN-00250,0000012
SYN-00350,0003,0003560

The three rows total ₹2,00,000 principal outstanding and ₹4,000 overdue. Reconcile those measures separately: overdue principal is already part of principal outstanding, so adding the two totals would double-count that component.

SYN-002 is current at cut-off but reached 12 DPD within the observation window. Today’s zero cannot stand in for its payment history. SYN-003 is 35 DPD today and has previously reached 60. The SMA and NPA classification guide explains why a DPD field alone does not establish regulatory status.

Who asks an NBFC for a loan tape?

Buyers of loan exposures, investors, rating agencies and funders can ask for loan-level data. The purpose affects the fields: a pool-selection review, a funding line and an ongoing investor report need not use identical layouts.

For loans covered by the Transfer Directions, paragraphs 39-40 put due diligence with the buyer’s staff and apply it loan by loan. Paragraph 40 allows portfolio-level checks for the remainder under specified conditions, including individual checks on at least one-third by both value and loan count and retention of at least 10% economic interest by the seller. A tape supports that work; it doesn’t replace the underlying due diligence.

Agree what the recipient needs and when it needs it. Keep the specification with the version of the file you sent.

What fields does a loan tape carry?

The sample uses a small set. A fuller dictionary should include each field’s type or unit, definition, source and missing-value rule. The rows below are suggested definitions for an evaluation, not a universal industry schema.

FieldDefinition and unitSource or time basisIf missing
Loan IDStable identifier, stored as textLoan record; unique within the declared poolResolve before matching or aggregating rows
Borrower IDIdentifier linking facilities of the same borrowerBorrower record; share an appropriate pseudonymous keyFlag the gap in borrower-level checks
Cut-off dateReporting date, with day-end and time-zone conventionSame reference date for this file’s snapshot fieldsReject an undated snapshot
Terms and datesRelevant disbursal, maturity and repayment dates, rate and frequencyLoan terms effective at cut-off; retain dated modificationsRequest the source rather than assume current terms applied throughout
Principal outstandingUnpaid principal in declared currency and unitsLoan balance as of cut-offReconcile or label incomplete; never fill with zero
Overdue amountEligible unpaid dues; list included componentsDue schedule and allocations as of cut-offDistinguish unknown from confirmed no arrears
Current DPDDays past due under the stated counting conventionOldest eligible unpaid due at cut-offKeep unknown status visible
Peak DPDMaximum observed DPD within a declared windowDated historical observations, including cut-off if specifiedMark unknown or partial coverage; do not use zero
Instalments paidCount of instalments satisfied under the recipient’s definitionReceipt allocation history; state treatment of partial paymentsDo not substitute months since disbursal
Classification and changesRegulatory class and dated modifications, write-offs or transfersApplicable rules and event recordsFlag incomplete history and its effect on use

A field such as net seasoning needs the selected recipient’s formula. Loan age, scheduled instalments and instalments actually paid are different measures. Define how moratoria and unpaid instalments affect the chosen measure.

For a column received from an external dataset, obtain its dictionary. A truncated merchant-history field name is not enough to infer the observation period or calculation. Borrower data also needs an agreed sharing scope: provide only the fields the recipient is entitled to receive.

What do rating agencies look for in a pool?

ICRA’s June 2025 methodology compares a pool with the wider portfolio using measures including seasoning, overdue status at cut-off, peak DPD, loan-to-value, geography, tenure, instalment-to-income and bureau scores. It also considers past payment behaviour. Recipients often group those overdue fields by collection bucket.

That is why the dictionary needs time windows. A score at origination and a refreshed score are different observations. Peak DPD over the last year and lifetime peak DPD are different measures too. Ask the agency which version it needs rather than relabel a nearby field.

Monthly history also supports the roll-rate and cure-rate calculations and the collection efficiency measures used in portfolio review. Their formulas remain with those guides.

What do RBI’s rules say about loan-level data?

Keep three parts separate:

STC information before pricing

Paragraph 51 of the Securitisation Directions calls for sufficient loan-level data, or summary stratification data for granular pools, to be available to potential investors before pricing, in accordance with applicable laws.

STC information during the deal

Paragraph 52 calls for timely loan-level or granular-pool stratification data and standardised investor reports at least quarterly throughout the securitisation. Data cut-offs should align with investor reporting.

Wider disclosure and data controls

Paragraph 111 addresses access to materially relevant data on individual underlying exposures, cash flows and collateral. Paragraph 13 of the Transfer Directions requires the board-approved policy to cover systems for capturing, storing and managing data.

STC means simple, transparent and comparable. The first two requirements above sit within those criteria. They should not be presented as one mandatory spreadsheet layout for every transfer or funding arrangement. Check the directions’ applicability and the deal documents before deciding what to send.

Where third-party servicer data is used, paragraph 109 of the Securitisation Directions also addresses information-sharing mechanisms and certification by the third party’s auditors, preferably at intervals no longer than a year. Agree that evidence as part of the reporting arrangement.

Why is a loan tape hard to produce?

A snapshot may contain the balance you need today while omitting the observations needed for a historical field. If earlier DPD values were overwritten, a peak-DPD column needs reconstruction from reliable records. If terms changed, the file needs the effective dates so a reader can interpret earlier payments correctly.

A mapping problem is different. When the history exists but a recipient uses another field name or unit, document the conversion and reconcile it. When history is absent, mark the limitation and agree how to handle it. Renaming a current value doesn’t recover a missing past observation. The same gap limits portfolio analysis that still runs in Excel: the output is only as good as the history the loan system kept.

What should the loan system keep?

Retain the due schedule and its changes, receipt allocations and dated balances. Keep the source observations needed for every historical field you promise to supply. For a transferred or securitised book, retain the pool membership, transfer date and economic share alongside the servicing record.

The recipient’s requested granularity matters. Month-end DPD can support a month-end series, but it cannot prove the highest DPD reached between those dates. If daily observations are unavailable, describe the peak as the maximum observed on the dates you do have and obtain the recipient’s agreement.

How do you validate a loan tape?

These are suggested checks for the file and its mapping:

  1. Check that the agreed loan population appears once per loan ID and that borrower keys link the intended facilities.
  2. Confirm the cut-off, currency and units. Reconcile principal and overdue totals separately to the same pool’s source records.
  3. Test historical fields against their declared windows and coverage. Where the window includes cut-off, peak DPD should not be below current DPD under the same convention.
  4. Keep unknown values distinct from zero. Document gaps instead of silently filling them.
  5. Ask the recipient to confirm field definitions and permitted data sharing. Keep the accepted mapping, file version and reconciliation together.

How should an NBFC produce its loan tapes?

Begin with the recipient’s dictionary and one agreed cut-off. Build a small sample, reconcile it and resolve field questions before producing the full file. That sample can expose a mismatch in units or history without repeating it across the pool.

Lokta’s Loan Management keeps the post-approval loan record on a double-entry ledger. The data needed for a particular tape still depends on the recipient’s definitions and the history available. Bring a sample specification to review your loan-book data requirements.

Frequently asked questions

What is a loan tape?

A loan tape is a structured file of loan-level data used to review a portfolio. In the example here, each row is one loan and the columns describe its terms, balances, status and history at an agreed cut-off. A buyer, funder or rating agency specifies the fields it needs.

What fields should a loan tape include?

Agree the fields with the recipient. A starting dictionary includes loan and borrower identifiers, relevant dates, terms, outstanding principal, overdue amounts, DPD at cut-off and historical fields such as peak DPD. Define the units, observation windows, source records and treatment of missing values for each field.

What is peak DPD in a loan tape?

Peak days past due is the highest DPD observed within a stated period, which might be the loan's life or a shorter window. It differs from DPD on the cut-off date. A current loan can still have a history of arrears, so retain the historical observations and label the window. Missing history is not zero DPD.

Do RBI rules require loan-level data for securitisation?

RBI's STC criteria call for sufficient loan-level data, or summary stratification for granular pools, before pricing. They also call for timely loan-level or granular-pool data and standardised investor reports at least quarterly during the deal. The directions separately require access to materially relevant data about individual underlying exposures. These requirements do not prescribe the illustrative file shown here.

How do you validate a loan tape?

Check unique loan IDs, a consistent cut-off, declared currency and units, and totals reconciled to the same pool's source records. Test historical fields against their stated windows, document missing values and get the recipient to confirm the field mapping. Keep the file version and the reconciliation evidence together.


Sources:

Bring us your live book

See what agents can do after approval.

Talk to us
Founder-led adoption

Adopt the agentic loan servicing platform.

Lokta is built for enterprise deployment, VPC or single-tenant cloud, with an audit trail in every state change. We work with a select group of institutions through a founder-led model: deep adoption, deliberate scope, a delivery window the team commits to in writing.