SMA early warning signals: how to read movement on your book

What does reading SMA movement tell you?

Reading special mention account (SMA) movement means comparing the SMA bucket of every account at consecutive month-ends, then tracking how much balance enters, rolls forward or cures, because a stable SMA total can hide a book that is getting worse.

The outcome is a monthly view that answers three questions:

• How much fresh stress entered the book?
• How much existing stress got worse, and how much cured?
• Where is the movement concentrated, and who is acting on it?

The SMA categories and reporting duties below come from the Reserve Bank of India (Commercial Banks – Resolution of Stressed Assets) Directions, 2025, dated November 28, 2025 and updated as on October 1, 2026. Those Directions do not prescribe a method for reading movement. The transition table, roll rates and cure rates in this guide are general working practice.

What do you need before reading SMA movement?

Reading SMA movement needs clean inputs and clear internal rules:

• Account-level snapshots at consecutive month-ends, with the same cut-off each time.
• Facility type, so revolving and non-revolving facilities are tested separately.
• Days overdue for loans, and days continuously above the lower of limit or drawing power for cash credit and overdraft. Drawing power is the amount a borrower can currently draw against eligible security.
• Segment fields such as product, ticket size, branch, industry and sanction vintage.
• Aggregate exposure per borrower, to identify accounts within central reporting.
• Your bank’s Board-approved stressed asset policy. Paragraph 9 of the Stressed Assets Directions, 2025 requires a commercial bank to have one, covering resolution timelines and quantitative and qualitative signs of financial difficulty.

Escalation triggers and watchlist limits vary by institution.

How are SMA categories defined for term loans and cash credit?

SMA categories under paragraph 15 of RBI’s Stressed Assets Directions, 2025 differ by facility type, and revolving facilities have no SMA-0. A bank must recognise incipient stress by classifying the account immediately on default.

The scope of these SMA rules is set out as follows:

• Lenders covered: paragraph 3 covers commercial banks, meaning banking companies other than small finance banks, payments banks and local area banks, plus corresponding new banks and State Bank of India. Other lender types, including NBFCs, should check the direction for their own type.
• Accounts covered: paragraph 16 applies SMA classification to all loans, including retail loans, irrespective of exposure size. Agricultural advances governed by crop season-based asset classification norms are excluded.
• Default for revolving facilities: paragraph 6(4) treats an outstanding continuously above the lower of limit or drawing power for more than thirty days as default.

How do you read SMA movement step by step?

Reading SMA movement takes the same steps each month, in the same order.

1. Freeze the month-end snapshot. Take every account at one cut-off date. What good looks like: each account sits in one bucket, and the total ties to the book balance.
2. Test each facility against the right rule. Use overdue days for loans, and continuous days above the lower of limit or drawing power for cash credit and overdraft. What good looks like: no cash credit account appears in SMA-0, and drawing power is based on current stock and book debt statements.
3. Build the transition table. Put opening buckets in rows and closing buckets in columns, with a column for accounts past 90 days. What good looks like: rows add to the opening balance and columns add to the closing balance, with new sanctions, closures and write-offs shown as separate lines.
4. Calculate entry, roll and cure rates. Use a fixed formula every month, by value and by count. What good looks like: anyone can recompute the rates from the table.
   • Entry rate: balance moving from not SMA into any SMA bucket, divided by the opening not SMA balance.
   • Roll rate: balance moving to the next worse bucket, divided by the opening balance of the starting bucket.
   • Cure rate: balance returning to not SMA, divided by the opening balance of the SMA bucket being measured.
5. Cut the movement by segment. Split by product, ticket size, branch, industry and sanction vintage, and list the largest accounts by name. What good looks like: you can tell whether a rise comes from one account or from a pattern.
6. Check how each cure happened. Separate cures from borrower cash flow from cures funded by fresh limits. What good looks like: a cure backed by new bank money is flagged rather than counted as clean.
7. Add signs of stress that sit outside arrears. Review cash flow, internal rating movement and repeated late payment alongside the buckets. What good looks like: accounts with these signs are on the list even when they are not SMA.
8. Assign action and tie to reporting. Give each flagged account an owner, an action and a date. What good looks like: the review list agrees with what the bank reported to the central repository.

How do you interpret the direction of SMA movement?

The direction of SMA movement matters more than the size of the SMA total, and each pattern points to a different first action. Compare rates with your own earlier months, because no RBI benchmark for them appears in the Stressed Assets Directions, 2025.

What do RBI’s reporting and resolution rules require on SMA accounts?

RBI’s Stressed Assets Directions, 2025 set the following duties for commercial banks:

• CRILC reporting (paragraph 17): a bank reports credit information, including SMA classification, to the Central Repository of Information on Large Credits (CRILC) every month, for all borrowers with aggregate exposure of ₹5 crore and above.
• Weekly default report (paragraph 18): a bank submits a report of defaults by those borrowers by close of business every Friday, or the preceding working day if Friday is a holiday.
• Review period (paragraphs 6(13) and 23): once a specified lender other than an NBFC reports a borrower in default, the specified lenders jointly carry out a prima facie review within thirty days of the default.
• Action before default (paragraph 26): RBI calls default a lagging indicator and expects a bank to start implementing a resolution plan even before a default.
• MSME carve-out (paragraph 4): Chapters III and IV, which include the review period and paragraph 26, do not apply to revival and rehabilitation of MSMEs covered by Circular FIDD.MSME & NFS.BC.No.21/06.02.31/2015-16 dated March 17, 2016, as amended.

Worked example, made-up numbers

This made-up book of ₹1,000 crore holds balances flat, with no new sanctions, repayments or closures, so the table reconciles exactly. All figures are in ₹ crore.

The rates from this made-up table are:

The reading of this made-up book is:

• The SMA total looks better. SMA balance fell from ₹100 crore (10.0% of the book) to ₹95 crore (9.5%).
• The improvement is misleading. ₹8 crore crossed 90 days and left the ladder. Adding it back gives ₹103 crore, above the opening ₹100 crore.
• More balance worsened than improved. ₹67 crore moved to a worse bucket (30 + 10 + 10 + 9 + 8) and ₹43 crore moved to a better one (28 + 6 + 3 + 3 + 1 + 2). The remaining ₹890 crore stayed put.
• Direct entry to SMA-1 is cash credit. The ₹10 crore moving from not SMA to SMA-1 is all cash credit, which has no SMA-0 bucket. All SMA-0 balances here are term loans.
• One account drives the SMA-1 roll rate. ₹6 crore of the ₹9 crore moving to SMA-2 is one mid-corporate cash credit account. Without it, the roll rate is 3 ÷ 24, or 12.5%, against 30.0% with it.

This made-up result calls for two separate actions. The ₹6 crore account needs a name-level review, and at ₹6 crore of aggregate exposure it falls within CRILC reporting. The rate of 12.5% then needs comparing with earlier months, since one period does not show a trend.

What are the common errors in reading SMA movement?

Common errors in reading SMA movement come from relying on the total, testing the wrong balance or missing stress outside the buckets:

• Reading only the SMA total. The total can fall because accounts slipped past 90 days. Show flows next to the total.
• Expecting cash credit in SMA-0. Paragraph 15 of RBI’s Stressed Assets Directions, 2025 has no SMA-0 for revolving facilities, so an account can first appear in SMA-1. Track shorter overdrawn periods on an internal watchlist.
• Testing cash credit against the limit alone. Paragraph 15 uses the lower of sanctioned limit or drawing power. A stale drawing power also distorts the count of overdrawn days.
• Treating ₹5 crore as an SMA threshold. Paragraph 16 applies SMA classification irrespective of exposure size. The ₹5 crore level in paragraphs 17 and 18 applies to reporting.
• Treating no arrears as no stress. Paragraph 11 says financial difficulty may be identified even in the absence of arrears, and paragraph 10 lists signs such as a pattern of payment delinquency and cash flows assessed as insufficient to service debt.
• Counting every cure as clean. Paragraph 6(12) lists additional credit or released funds for an account in default, granted to aid curing the default, among the terms restructuring would generally include when a concession is granted for reasons relating to the borrower’s financial difficulty.
• Managing the bucket instead of the account. Paragraph 21 says action intended to conceal an account’s actual status or evergreen stressed accounts will face stringent supervisory or enforcement action, including higher provisioning and monetary penalties.
• Borrowing a benchmark from another book. Roll and cure rates depend on product mix and collection practice, so compare with your own history. Internal triggers vary by institution.