Indian Banks Have Never Looked Healthier. So Where Is The Risk?

This is how it’ll impact general investors

Every major Indian bank reported this quarter and the numbers looked almost identical across all of them: bad loans at their lowest in two decades, credit growing fast, profits up. The RBI's June 2026 Financial Stability Report confirmed that system-wide Gross NPA (the total pile of loans where borrowers have stopped paying, expressed as a percentage of all loans) has fallen to 1.8%, the lowest in roughly 20 years, and every analyst note published since has been some version of the same headline: Indian banking is back.

Credit cycles almost never end with a visible crisis. They end because a handful of quiet pressures compound over two or three quarters while the headline numbers still look clean, and by the time GNPA starts moving it is already too late to position around it. Three of those pressures are building right now inside the Indian banking system, and none of them show up in the 1.8% figure.

The Number That Is Not the Whole Story

GNPA is a stock measure, meaning it tells you how much bad debt has accumulated on bank books up to today, not how fast new bad debt is forming right now. The ratio falls when banks write off old hopeless loans, sell them to asset reconstruction companies or simply grow their total loan book faster than new defaults appear, and India's banks have been doing all three simultaneously over the past four years, which is why the FSR's baseline projection has GNPA rising only slightly to 1.9% by March 2028 in a normal environment but jumping to 3.8 to 4.1% in a severe scenario, and the gap between those two numbers is the entire reason to run a stress-test rather than just read the headline.

The more useful leading indicators are the ones that move three to six months before GNPA does: fresh slippages (new loans turning bad each quarter), SMA-1 and SMA-2 accounts (loans that are 30 to 90 days overdue but not yet classified as fully non-performing), and credit costs (how much the bank is setting aside in provisions against future losses). Those indicators are still benign, which is exactly the window in which it pays to build a monitoring framework rather than assume the calm will persist.

Risk One: The Margin Squeeze That Has Already Started

Net Interest Margin, or NIM, is the simplest measure of how much a bank earns after paying for its own funding. If a bank lends at 9% and its deposits cost 5%, its NIM is roughly 4%. When the RBI cuts the repo rate (the overnight rate at which it lends to commercial banks, which serves as the floor for all other lending rates in the economy), that 9% lending rate falls quickly because roughly half of Indian bank loans are directly linked to the repo rate or the External Benchmark Lending Rate and reprice within a single quarter, while the 5% deposit rate barely moves for another two to three quarters because banks cannot force existing fixed-deposit holders to accept lower rates until their deposits roll over, so the bank finds itself earning less interest income before it has achieved lower interest expense, and that lag is the mechanical definition of NIM compression.

HDFC Bank's Q1 FY27 official presentation (July 18, 2026) is the clearest live illustration of this: NII (Net Interest Income, meaning the bank's core revenue from lending minus what it pays depositors) grew only 6.7% year-on-year to Rs335.3 billion while the loan book grew 15.4%, meaning the bank is writing significantly more loans but earning proportionally less from them. The reason is visible in the same filing: CASA ratio (the share of deposits sitting in low-cost savings and current accounts, which are cheap for the bank to hold) fell from 34% to 32% year-on-year while time deposits grew 17.4%, because customers are moving their money out of savings accounts and into fixed deposits to capture better rates, which directly widens the bank's funding cost before a single RBI rate cut has fully transmitted through.

For scale: a 10 basis-point (meaning 0.10 percentage point) NIM compression on a Rs30 lakh crore interest-earning asset base translates to roughly Rs3,000 crore in lost annualised pre-tax income before any management action, which is approximately HDFC Bank's full quarterly provisioning buffer.

What the Rate Cycle Looks Like Across Four Scenarios

Risk Two: Credit Is Growing Six Points Faster Than Deposits

RBI data for July 2026 shows bank credit growing at 19.3% year-on-year to Rs219.3 lakh crore while deposits grew at only 15.4% to Rs265.4 lakh crore. The critical distinction most commentary skips is between the stock credit-to-deposit ratio and the incremental one. The stock ratio (total loans divided by total deposits) looks comfortable because deposits still exceed loans in absolute terms, but the incremental ratio (for every new rupee of loans being written, how many new rupees of deposits are being raised?) can be dramatically higher and more stressful, meaning a bank that is funding nearly every new loan with freshly raised deposits has to compete aggressively for those deposits, bidding up the rates it offers and squeezing margins further.

When that funding gap persists the banking system follows a predictable sequence: deposit rates rise first, then CASA balances shrink as customers chase higher fixed-deposit rates, then banks start reaching for bulk deposits and certificates of deposit which carry higher costs and shorter durations (meaning the bank has to keep rolling them over and renegotiating rates each time they mature), then wholesale funding spreads widen, and then either loan growth gets rationed or margins fall further to compensate. IndusInd Bank's Q1 FY27 presentation (July 22, 2026) shows the extreme version of this: deposits grew just 4% year-on-year while advances actually fell 2%, which is what happens when a bank's funding franchise hits its limits before loan demand does, and that constraint is entirely invisible in IndusInd's GNPA reading of 3.25%.

Industry credit grew 17.5% year-on-year per RBI sectoral data, but that aggregate figure can easily conceal concentration risk: if that growth is sitting in infrastructure, real estate and NBFC lending rather than distributed across investment-grade corporate borrowers, the underwriting quality question becomes urgent well before the GNPA number reflects it.

Risk Three: A Regulatory Change Most Analysts Have Not Modelled

On January 16, 2026, SEBI published a consultation paper proposing to let Foreign Portfolio Investors (FPIs, meaning foreign institutions like hedge funds and asset managers that invest in Indian stocks and bonds) net their cash flows for eligible transactions. In plain terms: today when an FPI sells one stock and wants to buy another, the sale proceeds sit locked in the settlement system overnight before being available for the next purchase, and the proposal would allow those proceeds to fund the next purchase immediately, reducing the temporary cash that gets parked and freeing up liquidity.

That sounds like a pure operational improvement, and in normal markets it largely is, but the second-order risk is that if FPIs can recycle capital between positions more efficiently they can also rotate out of a sector more efficiently when they choose to de-risk simultaneously, and Indian bank stocks and bank bonds are consistently among the largest FPI holdings, so any change in the velocity or size of FPI flows hits bank securities with disproportionate force compared to the broader market. The actual scope of the final SEBI circular matters enormously here because the consultation paper only covered outright purchase and sale transactions while securities settlement between FPI and custodian would remain gross, and any analysis treating the consultation paper as the final rule is working from incomplete information.

Source: SEBI Consultation Paper on FPI Fund Netting, January 16, 2026

ICICI Bank enters this environment in the strongest position with GNPA around 0.87%, NII and PAT both growing well ahead of peers, and a deposit franchise that has shown more resilience than most, but even ICICI's forward scorecard needs to be watched for deposit pricing discipline, whether its unsecured retail and microfinance book shows any early delinquency pickup, and whether treasury income (gains from selling government bonds at a profit when bond prices rise as yields fall) remains a small fraction of operating profit rather than a growing crutch.

HDFC Bank is the most instructive case study in reading past the headline: GNPA of 1.17% and ex-agriculture GNPA of 0.91% look excellent, but the CASA erosion from 34% to 32% year-on-year with time deposits growing 17.4% is a direct preview of where cost-of-funds pressure is heading, and the 6.7% NII growth against 15.4% loan growth confirms the margin dynamics are already in motion before a single full rate cut has transmitted through.

IndusInd Bank is the most complex story: PAT of Rs1,037 crore, NII of Rs4,685 crore, GNPA falling from 3.43% to 3.25% quarter-on-quarter with NNPA (the net NPA figure, calculated after subtracting provisions already set aside against those bad loans) at 0.95% and an average Liquidity Coverage Ratio (the buffer of liquid assets a bank must hold to survive a sudden 30-day stress event) of 127% all point to a genuine recovery, but 4% deposit growth and advances falling 2% confirm the funding franchise is still rebuilding and the turnaround needs several more quarters of proof.

The Five-Category Dashboard to Watch Every Month

The most useful output of a stress-test framework is not a single number but a set of directional signals that give three to six months of lead time before GNPA reacts. Direction matters more than any single breach of a threshold, and what the table below tracks is the difference between a signal worth monitoring and one that means the risk is materialising.

The Bear Case and What Could Kill It

These risks are conditional, not inevitable. The bear case weakens substantially if deposit growth reaccelerates and closes the gap with credit in H2 FY27, or if RBI rate cuts prove shallower or more delayed than the market currently prices (giving banks more time for deposit repricing to catch up), or if the 19% credit growth broadens into investment-grade corporate borrowers with better collateral rather than concentrating in unsecured retail and NBFC lending. RBI's own capital adequacy data showing CET1 ratios (the core equity buffer each bank holds as a percentage of risk-weighted assets, which is the first line of defence against losses) well above regulatory minima provides a genuine cushion to absorb a margin cycle without triggering solvency stress. The sector is genuinely strong, and the point is simply that genuine strength and late-cycle earnings risk coexist, and the banks that hold up are the ones that can grow deposits without overpaying for them, preserve NIM through the easing cycle rather than replacing lost spread with one-off treasury gains, and recognise stress early through provisions rather than restructuring it away.