[ Credit Risk ]

First EMI Default: An Origination Diagnostic, Not a Collections Problem

First EMI default and 90 day default trace back to origination, not deterioration. Three causes, a six question file review, and the loop back to credit policy.

Zeus Dhanbhoora

10 min read · 12 November 2026

First EMI Default: An Origination Diagnostic, Not a Collections Problem
Contents
  1. Three causes of early default, three different responses
  2. Why early default is the fastest feedback loop in lending
  3. The 90 day default review: six questions for every file
  4. Account dressing is an early default mechanism
  5. Building the early default loop
  6. What early default analysis does not tell you
  7. Where Fiscus fits
  8. Frequently asked questions

A first EMI default, also called a first payment default, is a failure of the first scheduled instalment on a new loan. Together with default inside the first 90 days, it is the fastest origination quality signal a lender has, because the borrower had no time to deteriorate: the capacity was never there, the intention was never there, or the loan was structured so it could not work. Each cause has a different fix, and none of them sits in collections.

A borrower who stops paying in month eighteen deteriorated. Something happened: a customer was lost, a facility was called, income fell. The credit decision may have been sound and the world moved.

A borrower who misses the first EMI, or stops paying by month three, did not deteriorate. There was no time to.

These are different events with different causes. Most lending operations put them in the same delinquency bucket, route them to the same collections queue, and lose the distinction, which wastes the most actionable signal a credit function receives.

Three causes of early default, three different responses

Early default has a short list of explanations, and separating them is most of the work.

The capacity never existed

Income was overstated, obligations were missed, or both. A borrower servicing ₹14,000 a month of pawnbroker gold loan interest and chit fund contributions against ₹30,000 of income has no room for a new EMI, and nothing in a bureau pull says so, because pawnbrokers and chit funds do not report. The fix is assessment capability: obligation discovery beyond the bureau and income derived from the account rather than accepted from the application.

The intention never existed

First party fraud. The application was built to pass origination, with no plan to repay beyond what was needed to avoid immediate scrutiny. The fix is origination fraud control.

The structure was wrong

The borrower can pay and the loan was built so that they could not. The presentation date falls before the salary or receipt date. The EMI was sized to a peak month in a seasonal business. The tenure assumed a payment capacity the borrower has only in some months. This is a design failure, and it is routinely misclassified as a credit failure.

The third category deserves more attention than it gets, because it is cheap to fix and the likeliest to be mislabelled. RBI already requires loan agreements to state exact repayment due dates, so the EMI date is a decision taken at origination. RBI's April 2024 circular on charging of interest requires interest to run from actual disbursal, with the broken period charged pro rata, and nothing in it stops a lender setting the first EMI to the borrower's inflow date.

When the date is wrong, the cost lands on the borrower. Under RBI's day end process, the account is overdue at the end of the due date and flagged SMA-0. A borrower whose EMI is presented on the 3rd and whose income lands on the 7th collects a bounce charge and a recovery call. If a bureau reporting date (the 9th, 16th, 23rd or month end) falls before the payment clears, they also get a DPD entry on their bureau record and the SMS alert lenders must send when they report one. That is a date alignment problem, not a credit event, and the borrower has been mishandled by the lender rather than the reverse.

Why early default is the fastest feedback loop in lending

Here is why early default deserves its own process rather than a line in a delinquency report.

Conventional origination quality measurement runs on vintage curves: delinquency plotted against months on book, one curve per cohort. TransUnion CIBIL's vintage delinquency measure, for example, counts accounts ever 30 or more days past due within six months of origination and ever 90 or more within twelve. Reads like that take two to four quarters to mature. By the time the April cohort is visibly bad, you have originated May through September under the same policy, scorecard cut off and channel mix.

First EMI and 90 day default come back in 30 to 90 days.

That is the only origination quality signal fast enough to change what you are still doing. A July policy change that weakens borrower quality shows up in September's early default rate, while the conventional vintage analysis is still working on January.

Many lenders have this data and do not use it this way. Early default appears in collections reporting and portfolio dashboards, but it is rarely routed back to the credit function as a diagnostic, segmented by the origination variables that make it actionable: sourcing channel, geography, product variant, scorecard band, policy exception status, and the credit officer or branch that approved the file.

It is also about to cost more. From 1 April 2027 most scheduled commercial banks move to expected credit loss provisioning, under which an account more than 30 days past due carries a rebuttable presumption of a significant increase in credit risk, and with it a lifetime loss provision. An early default stops being only a collections queue entry and becomes a provisioning charge in the quarter it happens.

The complementary signal, on a longer clock, is the banking derived cohort read: throughput deterioration and obligation accretion from month two. Early default tells you the file was wrong at sanction. Cohort drift tells you the borrower is going wrong now. A lender wants both, and many have neither.

The 90 day default review: six questions for every file

For every loan that defaults within 90 days, go back to the origination file and ask one question: what was visible at the time, and why did it not stop this?

This is the highest value review a credit team can run, and it is rarely done systematically. Six checks, each pointing to a different corrective action.

1. Was the account dressed?

Covered in detail below. When dressing is present, it explains the early default almost entirely.

2. Were obligations missed?

Recurring debits to lenders that did not appear in the bureau pull: chit fund contributions, pawnbroker and moneylender loans, family borrowing, or a loan disbursed in the last ten days or so that had not yet reached the bureau. If the file shows a comfortable FOIR and the account shows ₹11,000 a month of servicing nobody counted, the assessment described a borrower who does not exist. Obligation mapping from the debit side is the check.

3. Was income overstated?

The classic failure is a loan disbursal classified as a trade receipt or salary credit, which inflates income and hides the obligation behind it in one transaction. Our categorisation benchmark identified borrowing recorded as revenue as the single most expensive error class in business lending.

4. Was the EMI date aligned to the inflow date?

If the borrower's money arrives on the 7th and the presentation is on the 3rd, the first EMI was always going to fail.

5. Was the statement set complete?

Accounts referenced in transfers but never shared. A partial parse that dropped a continuation page and produced a complete looking analysis on incomplete data.

6. Was the file a policy exception?

Files approved with a deviation should be tracked as their own cohort. Deviation performance is often not measured at all, which means deviation practice runs with no feedback.

Run thirty early default files through those six questions and the pattern is usually not thirty different stories. It is two or three repeated failures, each with a specific fix.

Account dressing is an early default mechanism

This is the cleanest causal link in the subject, and it explains why early default clusters the way it does.

Account dressing is the deliberate operation of a genuine account to look creditworthy before an application: balances parked, funds circulated between related parties to inflate turnover, round figure credits manufactured, a quiet account made to look active.

Dressing costs money and effort to maintain. Nobody maintains it after disbursal, because the purpose has been served.

So the capacity visible in the statement was manufactured for the application window and stops existing once the loan is drawn. A borrower whose account was dressed fails early almost by construction, not because they deteriorated, but because the condition the lender assessed was never the borrower's actual condition.

That produces a usable test. If early default correlates with specific behavioural markers in the origination statement, such as nil credit windows at statement period boundaries, circular counterparty flows, round figure credits that do not look like invoice payments, or balances that spike before the period opens and collapse after it closes, then your origination fraud rules are either absent or not firing. That is a tunable control, and early default files are the labelled dataset for tuning it.

A dressed account passes every document forensic test there is. The statement is genuine, the metadata is clean, the balance trail reconciles. Only behaviour gives it away.

Building the early default loop

Five steps, none of which needs new data.

1. Define the cohort precisely

First EMI failures, and accounts reaching 30 or more DPD within 90 days of disbursal, as two separate series. They are related and not identical.

2. Strip out the mandate failures, but read the rest

Mandate not received, mandate expired, invalid or inactive mandate, and amount above or not matching the mandate are administrative failures with their own NACH return codes, and leaving them in the numerator makes the credit signal unreadable. Two cautions. A presentation date that misses the inflow comes back as code 04, balance insufficient, the same code as a genuine shortfall, so only the inflow date in the bank statement separates them. And account closed, payment stopped and mandate cancelled by the customer are signals of intent, not administration. Keep them in.

3. Segment by origination variables, not collections variables

Channel, geography, product, scorecard band, exception status, approver. The point is to locate the origination decision that produced the failure.

4. Run the six question review on a sample every month

Thirty files is enough to find a pattern. This is the step that converts a rate into a cause.

5. Route the finding to credit policy, not to collections

With a named owner and a defined change. An early default rate that goes into a dashboard and nowhere else has thrown away the value of the signal. RBI's 2026 Fraud Risk Management Directions already expect the early warning framework to provide for periodic review of credit sanction and monitoring processes; this loop is a practical way to do it.

The cadence matters more than the sophistication. Monthly, with a sample reviewed and a finding written down, beats a quarterly analytics exercise nobody acts on.

What early default analysis does not tell you

Volume limits it

A lender writing 200 loans a month has too few early defaults for stable segment level rates. The file review still works at any volume because it is qualitative, but rate comparisons need scale.

Not every early default is an origination failure

Genuine shocks happen in month two: job loss, illness, a business event. The file review is what separates these from systematic failures; a rate alone cannot.

Fraud rings distort the picture

A coordinated fraud attack produces a cluster of early defaults that looks like a channel or geography problem and is not. Check geographic and counterparty clustering within the cohort before concluding anything about policy.

It says nothing about loans that did not default

Early default analysis studies failures. A policy change that weakens quality without producing immediate failures is invisible here and needs the cohort drift read instead.

It cannot see what was never collected

If the statement set was incomplete at origination, the review will find that fact but not the contents of the missing accounts.

Where Fiscus fits

At origination, against the causes above: obligations discovered from recurring debits, including lenders absent from the bureau; income derived with non revenue credits stripped out instead of counted; account dressing and circular flow detection with rules tuned by borrower segment; the borrower's actual inflow date identified so the EMI can be set to it; and documents flagged when they cannot be read reliably instead of partially parsed into a confident wrong answer.

For the review loop, the value is that the origination analysis is reproducible and traceable. When a file defaults in month two, the question is what the statement showed, and an analysis where every figure carries the account, page and transaction behind it answers that in minutes instead of becoming an argument about what the tool had said.

What we do not do is run your early default review. That is a credit process, it needs an owner inside the institution, and no vendor can hold it for you.

Frequently asked questions

What is first EMI default? The failure of the first scheduled instalment on a newly disbursed loan, also called first payment default. It differs from later delinquency because the borrower had no time to deteriorate. The capacity was absent at origination, the intention to repay was absent, or the loan was structured so that the first EMI could not succeed.

Why is early default treated differently from normal delinquency? Because the cause is different. Later default usually reflects a change in the borrower's circumstances. Early default usually reflects an error in the original assessment, first party fraud, or an EMI date and amount that never matched the borrower's cash flow. The corrective action belongs in credit policy and origination controls, not in collections.

How quickly does early default reveal origination quality? First EMI and 90 day default return within 30 to 90 days of disbursal. Standard vintage delinquency measures, such as accounts ever 30 days past due within six months or ever 90 within twelve, need two to four quarters. Early default is the only quality measure fast enough to correct a policy while you are still lending under it.

What should a 90 day default review examine? Six things: whether the account showed signs of dressing, whether obligations visible in the banking were missed, whether income was overstated through misclassified credits such as loan disbursals, whether the EMI date matched the borrower's inflow date, whether the statement set was complete, and whether the file carried a policy exception.

Why does account dressing cause early default? Dressing is maintained only until disbursal, because it costs money and effort and its purpose ends there. The capacity the lender assessed was manufactured for the application window, so it disappears once the loan is drawn. A dressed file therefore fails early almost by construction, not through any later deterioration in the borrower's position.

Should first EMI bounces always be treated as credit events? No. Mandate failures such as an expired, invalid or under limit mandate have their own return codes and are administrative. A presentation date that misses the inflow returns as balance insufficient, the same code as a real shortfall, so check the inflow date in the statement before treating the bounce as a credit event.


Fiscus derives income, discovers obligations outside bureau coverage, and runs segment tuned behavioural fraud checks at origination, with every figure traceable to the transaction behind it. Book a parallel evaluation using loans from your own book that defaulted within 90 days.

Written by

Zeus Dhanbhoora

Zeus Dhanbhoora is the CEO of BridgeUp Tech, the company behind Fiscus. He previously co-founded Bacferim Technologies and was an associate at the law firm Bharucha & Partners. He writes the Fiscus credit desk blog on benchmarks, fraud detection and credit underwriting methods.

See it on your own cases.

Run a parallel evaluation. Bring statements your team has already analysed and compare output, depth and time.

Book a demo