The Borrower Has Not Missed an EMI: Why a 20% Sales Drop and a 35% Receivables Jump Still Change the Credit Risk
Banking · Credit Risk · India · 2026
The Borrower Has Not Missed an EMI: Why a 20% Sales Drop and a 35% Receivables Jump Still Change the Credit Risk
A worked reading of one stressed-but-current account, set against RBI’s 2024 fraud risk directions, the SMA framework and the expected credit loss staging rules that begin on 1 April 2027.
Repayment is a lagging indicator. It tells you what a borrower could still afford last month, not what they can afford next quarter. That is the uncomfortable premise behind every credit monitoring framework, and it is why an account with a perfect twelve-month repayment record can still be the one that costs a lender money. The question worth sitting with is not whether the EMI landed. It is what the rest of the account has been doing while it did.
Quick Summary
Take a borrower whose monthly sales fell from ₹50 lakh to ₹40 lakh, receivables rose from ₹20 lakh to ₹27 lakh and bank credits fell from ₹45 lakh to ₹34 lakh over six months, with every EMI paid on time. Sales are down 20%, receivables up 35% and credits down 24%. The most telling derived number is not in the list: debtor days have gone from about 12 to 20. Under RBI’s July 2024 directions, banks run board-approved early warning systems precisely to catch this pattern, and from 1 April 2027 the expected credit loss rules will make acting on it a provisioning obligation rather than a judgement call.
What we know from the regulatory record
The scenario is illustrative, but the framework around it is not. Everything below comes from RBI instructions or published supervisory data rather than from industry lore.
- RBI issued three near-identical Master Directions on Fraud Risk Management on 15 July 2024, covering commercial banks and All India Financial Institutions, cooperative banks, and NBFCs including housing finance companies.
- Those directions require a board-approved framework for Early Warning Signals and Red Flagging of Accounts, integrated with core systems and supported by data analytics, with the Risk Management Committee of the Board overseeing effectiveness.
- The 2024 directions are principle-based and deliberately dropped the prescriptive indicator annex of the 2016 regime, so each institution now owns its own indicator list.
- An account is a Special Mention Account when payment is overdue 1 to 30 days (SMA-0), 31 to 60 days (SMA-1) or 61 to 90 days (SMA-2). Beyond 90 days it is a non-performing asset.
- Lenders report credit information, including SMA status, to CRILC for borrowers with aggregate exposure of ₹5 crore and above, monthly, with weekly reporting of defaults.
- RBI’s June 2026 Financial Stability Report put the gross NPA ratio of scheduled commercial banks at 1.8% as of March 2026, a multi-decadal low, with net NPAs at 0.5%.
Hold those first two numbers together, because the tension between them is the whole argument. Headline asset quality is the best it has been in decades. Yet the same report noted that SMA-0, SMA-1 and SMA-2 accounts together made up 9.4% of retail advances at public sector banks. Stress does not arrive as a single event. It accumulates in accounts that are still, technically, performing.
Reading the account: three numbers, one direction
Lay the borrower’s figures side by side and the shape is immediately visible. Every line has moved, none of them in a helpful direction, and all of them within a single six-month window.
then
now
then
now
then
now
Percentages sharpen it further. Sales are down a fifth. Receivables are up by more than a third against a falling sales base, which is the combination that matters, because rising debtors on rising sales is growth while rising debtors on falling sales is collection failure. And bank credits, the money actually flowing through the operating account, have fallen faster than sales.
Worked example: the number nobody printed
Debtor days approximate how long the borrower waits to be paid. Six months ago: ₹20 lakh of receivables against ₹50 lakh of monthly sales, multiplied by 30, gives 12 days. Today: ₹27 lakh against ₹40 lakh, multiplied by 30, gives about 20 days. That is a 69% deterioration in the collection cycle, far larger than either headline percentage. In cash terms, ₹7 lakh that used to sit in the bank account is now sitting with customers, and the annualised sales run rate has fallen from ₹6.0 crore to ₹4.8 crore.
Why bank credits are the line a monitoring officer reads first
Sales figures come from the borrower. Receivables come from the borrower. Credits into the operating account come from the bank’s own core system, which makes them the one series that cannot be presented favourably. That is why turnover routed through the account, compared with declared turnover, is a standard monitoring ratio.
Run it here. Six months ago the borrower routed ₹45 lakh against ₹50 lakh of sales, or 90%. Today it is ₹34 lakh against ₹40 lakh, or 85%. The five-point slip is small on its own, but it means the shortfall between declared sales and banked cash has widened from ₹5 lakh to ₹6 lakh a month even as sales shrank. Either collections really have slowed, or some of the turnover has started moving through an account the lender cannot see. Both readings warrant a conversation; they simply warrant different conversations.
| Signal | What it often means | The question to ask | What would clear it |
|---|---|---|---|
| Declining sales | Demand loss, lost customer, or capacity constraint | Is the fall concentrated in one product or one buyer? | A matching seasonal pattern in prior years |
| Falling bank credits | Slower collections, or turnover diverted elsewhere | Which other bank accounts does the borrower operate? | GST returns and credits that reconcile |
| Rising receivables | Weakening buyers or extended credit to hold volumes | What is the ageing beyond 90 days, by customer? | A clean ageing profile with named, solvent buyers |
| Rising inventory | Unsold stock, or a build for a confirmed order | Is there a purchase order backing the build-up? | Documented orders and a delivery schedule |
| Cheque returns | Day-to-day liquidity is already tight | How many in the last two quarters, and to whom? | A single technical return with a bank explanation |
| New borrowing appearing | The working capital gap is being plugged with debt | What do the bureau report and CRILC show? | Sanctioned expansion capex, disclosed in advance |
| Falling profitability | Pricing pressure or cost pass-through failure | Is the margin fall at gross level or below it? | A one-off cost with a clear end date |
| Delayed statutory dues | Cash is being rationed between creditors | Are GST and provident fund filings current? | Filings up to date with challans on record |
| Loss of a major customer | Concentration risk has crystallised | What share of revenue did that buyer represent? | A replacement contract already signed |
One red flag is not a default, and the framework says so
This is where credit analysis separates from alarmism. A seasonal manufacturer will show falling sales every year in the same quarter. A borrower supplying a government department may have receivables stretch for reasons that have nothing to do with solvency. A single cheque return can be a technical error at the payer’s bank. Treating any one of these as a default signal produces false positives, and a monitoring system that cries wolf gets ignored, which is worse than not having one.
What converts a flag into an assessment is five tests applied together: the trend over several periods rather than one reading, the frequency with which the signal recurs, the severity of the movement, the reason offered and whether it is verifiable, and the overall profile of the borrower including promoter support and collateral. The grid below is one way to structure that, though the thresholds are illustrative rather than regulatory.
| Metric | Comfortable | Watch | Act | This borrower |
|---|---|---|---|---|
| Sales change, 6 months | Better than minus 5% | Minus 5% to minus 15% | Worse than minus 15% | Minus 20%, act |
| Receivables versus sales | Moving together | Gap of 5 to 20 points | Gap above 20 points | 55-point gap, act |
| Debtor days | Stable or falling | Up 10% to 30% | Up more than 30% | Up 69%, act |
| Credits to declared sales | Above 90% | 75% to 90% | Below 75% | 85%, watch |
| Signals moving together | One, explained | Two, related | Three or more | Three, act |
On four of five rows this account sits in the act column, and the fifth is a watch. That is the answer to the credit test posed at the outset. A single 20% sales decline with a documented seasonal history is a monitoring item. Three correlated declines, one of them in the bank’s own transaction data, with no explanation on file, is a serious warning that happens to be arriving through a borrower who is still paying.
The trap in a clean repayment record
Servicing an EMI is a small cash commitment relative to a working capital cycle. A borrower under strain will usually protect it, because the consequences of missing it are visible and immediate, while stretching a supplier or delaying a statutory payment is quiet. That ordering is exactly why repayment behaviour deteriorates last. By the time the first EMI bounces, the borrower has typically been rationing cash for months, and the options available to the lender have already narrowed.
The clock the borrower has not started yet
Formal classification is calendar-driven and unforgiving. Once a payment is missed, an account moves through the Special Mention buckets at a fixed pace and reaches non-performing status at 90 days overdue. None of that has begun for this borrower, which is precisely the point: everything useful a lender can do sits to the left of the first band.
EWS territory
SMA-0
SMA-1
SMA-2
NPA
Provisioning is what gives that timeline its cost. Under the prevailing income recognition and asset classification norms, a standard account carries a small general provision while a substandard one carries a far larger specific provision, and the requirement climbs with every year an account stays doubtful. The gap between the first two bars below is the financial reason monitoring exists.
From good practice to provisioning rule: the 2027 change
Everything described so far has been discretionary in its consequences. A bank that spotted stress early made a better decision; a bank that missed it made a worse one, and the accounting looked identical until the ninetieth day. That changes with the expected credit loss framework.
What a monitoring officer actually does next
Recognising the pattern is the easy part. The value of a framework is that it prescribes a sequence, so that the response is proportionate and documented rather than improvised. Work down this ladder and stop where the evidence stops.
The cheapest intervention is the earliest one
At this stage the lender still has the full menu: tighten drawing power, ask for a stock and receivables statement monthly instead of quarterly, seek additional collateral, reprice, or fund the working capital gap deliberately with covenants attached. After the first missed instalment, most of those options are either unavailable or expensive. Nothing about the borrower has changed in the intervening ninety days. What has changed is the lender’s negotiating position.
What is still unclear
Honest credit monitoring is candid about the limits of what the data can tell you.
- Whether the sales decline is seasonal. Six months of data cannot answer that. The same period in the two prior years can, and that comparison is the first thing to pull.
- Whether the receivables are stretched or bad. An aggregate figure gives no ageing, no customer concentration and no dispute history.
- Where the missing turnover went. A fall in banked credits is consistent with slower collection and with diversion to another lender’s account, and only reconciliation distinguishes the two.
- What each lender’s own indicator library contains. Because the 2024 directions dropped the prescriptive list, any published set of indicators, including the one in this article, is illustrative rather than authoritative.
- How SICR thresholds will be calibrated in practice. Banks are setting internal thresholds now, and the first full year of ECL reporting will not be available until well after April 2027.
Six habits that make monitoring more than a form
Frequently asked questions
What are early warning signals in banking?
They are indicators drawn from transactions, financial statements and account conduct that suggest a loan is heading towards stress or possible fraud before any payment is missed. Typical examples include falling bank credits, rising receivables, cheque returns, delayed statutory payments and new borrowing appearing on the bureau report. RBI requires banks and larger NBFCs to run board-approved frameworks that monitor them.
Can a loan be risky if the borrower has never missed an EMI?
Yes. Repayment is a lagging indicator, and a borrower under pressure usually protects the EMI ahead of suppliers and statutory dues because the consequences are immediate and visible. Deterioration in sales, collections and banked turnover typically appears well before the first bounce, which is exactly why credit monitoring exists alongside repayment tracking.
What is the difference between SMA-0, SMA-1 and SMA-2?
They mark how long a payment has been overdue. SMA-0 covers 1 to 30 days, SMA-1 covers 31 to 60 days and SMA-2 covers 61 to 90 days. Past 90 days the account is classified as a non-performing asset. For borrowers with aggregate exposure of ₹5 crore and above, SMA status is reported to RBI’s CRILC database.
What does RBI’s 2024 Master Direction say about early warning signals?
The three Master Directions on Fraud Risk Management issued on 15 July 2024 require a board-approved framework for early warning signals and red flagging of accounts, integrated with core systems and supported by data analytics, with the Risk Management Committee of the Board overseeing it. Unlike the 2016 regime, they prescribe no fixed indicator list, so each institution defines its own.
What is a red flagged account and how quickly must it be reported?
A red flagged account is one where the presence of one or more early warning signals raises suspicion of fraudulent activity, triggering deeper investigation. Where the account meets the CRILC threshold, the red flagging must be reported to RBI within seven days, and the decision on whether it is a fraud is ordinarily expected within 180 days.
How does the expected credit loss framework change credit monitoring?
From 1 April 2027, banks move from provisioning after a loss event to provisioning for expected losses. Assets with no significant increase in credit risk sit in Stage 1 with 12-month expected losses; those showing a significant increase move to Stage 2 with lifetime expected losses, even while fully current. Banks have until 31 March 2031 to absorb the impact on existing books.
Does one early warning signal mean the borrower will default?
No, and treating it that way produces false positives that make the whole system less useful. A seasonal business will show lower sales in predictable quarters, and a receivables delay may have a documented reason. Analysts weigh trend, frequency, severity, the explanation offered and the overall borrower profile before a flag changes the credit view.
Are Indian banks currently seeing stress in their loan books?
Headline asset quality is strong. RBI’s June 2026 Financial Stability Report put the gross NPA ratio of scheduled commercial banks at 1.8% as of March 2026, a multi-decadal low, with net NPAs at 0.5% and capital ratios at multi-decade highs. The same report noted that SMA accounts made up 9.4% of retail advances at public sector banks, so early-stage stress persists beneath the headline.
The short version
Underwriting asks whether to lend; monitoring asks whether the risk has changed since. For this borrower it has. Sales down 20%, receivables up 35% and banked credits down 24% are three correlated movements inside six months, and the derived collection cycle has stretched 69%, from about 12 days to 20. None of that is a default, and any one of the three could have an honest explanation. Together, with nothing on file to explain them, they belong in the serious warning column, and the cheapest time to act on them is now, while the borrower is still paying.
