For most of the last decade, “credit risk in Indian banking” meant a crisis story. Ballooning bad loans. Undercapitalised public sector banks. A system searching for a floor. That story has changed. This analysis of Indian banking credit risk data draws straight from RBI’s own numbers. It shows a sector operating at some of the lowest stressed-asset levels in its history. At the same time, new risks are quietly building underneath: unsecured retail credit, gold loans, and a fast-growing NBFC sector. So for anyone building credit risk models on Indian data, the headline ratio matters less than the underlying trend. What matters is understanding where the improvement came from, and where the residual risk now sits.
This piece works through what RBI’s own publications actually show. Specifically, we draw on RBI’s Financial Stability Report (FSR) and its Report on Trend and Progress of Banking in India. We also use RBI’s supervisory data releases directly. First, we break the picture down by bank category. Then we connect it to how analysts actually model default risk in practice.
The single most-cited number here is the Gross Non-Performing Asset (GNPA) ratio of Scheduled Commercial Banks (SCBs). RBI’s Financial Stability Report was released in June 2026. It put the GNPA ratio of SCBs at 1.8% as of March 2026 — a multi-decadal low. RBI’s own baseline projection expects only a slight edge-up, to around 1.9% by March 2028. The central bank called this a modest, manageable increase, not a reversal of the underlying trend.
That figure looks even more striking against where the system stood less than a decade ago. In 2018, RBI’s Financial Stability Report flagged a system-wide GNPA ratio of 11.46%. Banks under Prompt Corrective Action (PCA) — RBI’s restrictive framework for weak lenders — ran even higher. The distance between 11.46% in 2018 and 1.8% in 2026 is the real story here. In fact, it represents one of the fastest deleveraging cycles any major banking system has gone through. And it wasn’t accidental. It followed a specific, traceable sequence of interventions. Any credit risk analyst working with this data should be able to name them.
Capital buffers moved in the same direction. RBI’s June 2026 FSR put the system-wide Capital to Risk-Weighted Assets Ratio (CRAR) at 17.7%. The Common Equity Tier-1 (CET1) ratio stood at 15.3%. Both are multi-decade highs, per RBI. Similarly, RBI’s Report on Trend and Progress of Banking in India 2024-25 shows the same trend over a longer horizon. System CRAR rose from 12.94% in March 2015 to roughly 17.4% by end-March 2025. Profitability tracked the same arc. Scheduled commercial banks’ aggregate net profit rose too — from ₹2.63 lakh crore in FY2022-23 to ₹4.01 lakh crore in FY2024-25, per RBI’s Trend and Progress data.
RBI’s stress testing adds another layer here, separate from the point-in-time ratios above. Under RBI’s macro stress test scenarios in the June 2026 FSR, the banking system absorbed hypothetical adverse shocks. Even so, aggregate capital ratios stayed comfortably above regulatory minimums. Compare that to RBI’s own 2018 FSR. That report projected a comparable stress scenario could push GNPA at PCA banks as high as 22.3%. It also warned that at least six banks faced a capital shortfall. The gap between those two outcomes, eight years apart, is telling. It may say more about systemic resilience than either year’s baseline GNPA ratio alone.
None of this means credit risk has disappeared, however. RBI’s June 2026 FSR was explicit on this point: new vulnerabilities are building even as legacy NPAs shrink. Specifically, RBI surveyed banks and NBFCs for the report. They flagged rising household debt, rapid growth in gold loans, and AI-enabled cyber threats as their top risks. In other words, the credit risk profile of Indian banking hasn’t shrunk so much as shifted. It has moved from the corporate and industrial exposures that dominated the 2015-2018 cycle, toward retail and unsecured lending instead.
Aggregate numbers hide where the improvement actually concentrated, though. RBI supervisory data as of 30 September 2025 breaks the 2.15% domestic GNPA ratio down by bank category. The Ministry of Finance cited this data in a Parliament reply. Public Sector Banks (PSBs) stood at 2.50%. Private Sector Banks (PVBs) stood at 1.73%. Foreign Banks stood at just 0.80%. Interestingly, the same data shows PSBs improving faster than private or foreign banks since March 2018. In other words, the segment that was historically the weakest link has also recovered the fastest. Even so, it still carries the highest absolute ratio of the three.
That PSB recovery shows up in profitability too, per RBI’s Trend and Progress report. PSBs’ aggregate net profit rose from ₹1.05 lakh crore in FY2022-23 to ₹1.78 lakh crore in FY2024-25. That’s a faster proportional gain than the system as a whole. It’s consistent with a segment recovering from a lower base.
Beyond the bank-vs-bank split, this Indian banking credit risk data also shows where credit itself concentrates. NBFC credit is now systemically significant. It reached 14.6% of GDP in FY25, up from 13.5% a year earlier. Industry and retail exposures made up a combined 81.1% share of that NBFC credit as of end-March 2025. Services accounted for the remaining 15.4%. On the bank side, credit to MSMEs grew faster than credit to large enterprises over the same period. Meanwhile, Small Finance Banks (SFBs) posted strong balance-sheet growth. Even so, their asset quality worsened year-on-year, even as their credit-deposit ratio improved to 86.4%. That divergence is telling: large banks and PSBs improving, SFBs under pressure. It’s exactly the kind of segment-level signal a system-wide GNPA number alone won’t surface.
Smaller institutions tell a related story, too. The number of Urban Co-operative Banks (UCBs) fell by 15 during FY25, to 1,457, as consolidation continued in that segment. Still, UCB credit growth accelerated to 6.7% — the fastest pace in six years. Fraud data adds a further wrinkle here. Private sector banks accounted for 59.3% of reported fraud cases by number in FY25. Public sector banks, meanwhile, accounted for 70.7% of the total value involved. Clearly, operational risk and credit risk don’t always concentrate in the same segments.
A GNPA ratio is a lagging, backward-looking measure. It tells you what has already defaulted, not what’s likely to default next. That’s the gap probability of default (PD) modelling closes. It’s also why this Indian banking credit risk data is a genuinely useful input for PD work. It’s not just a macro backdrop to quote in an introduction.
Two things from the data above feed directly into PD modelling in practice. First, the bank-category breakdown (PSB 2.50% vs PVB 1.73% vs Foreign 0.80%) is a natural segmentation variable. Pooling all SCBs into a single PD model would blur real structural differences in underwriting standards and portfolio composition. Second, the sectoral concentration data points the same way. NBFC credit is dominated by industry and retail, while MSME credit is outpacing large-enterprise credit. This shows where a PD model needs the most segment-specific calibration. After all, default behaviour in MSME and unsecured retail books doesn’t track large-corporate exposures.
Our guide to PD Estimation Methods for Credit Risk covers vintage analysis, logistic regression, and survival-based approaches for Indian data. If you’re working through the calculation mechanics specifically, our companion piece on How to Calculate Probability of Default can help. It walks through the formula with worked examples using international banking data.
The clearest illustration of how India’s credit risk profile actually shifted is the public sector bank recovery itself. Public data documents this recovery step by step. That makes it a real, traceable case study, rather than a hypothetical one.
It started with the Asset Quality Review (AQR), which RBI initiated in 2015. The AQR forced banks to recognise stressed assets they had previously kept off the books through restructuring. From there, the government and RBI ran a coordinated, multi-year intervention often called the “4R’s” strategy:
The Insolvency and Bankruptcy Code (IBC), enacted in 2016, played a central role too. It gave lenders a workable legal mechanism to recover value from defaulted large-corporate exposures. As a result, RBI and government data attribute roughly ₹3 lakh crore in recoveries to the IBC process alone. Separately, the government injected an estimated ₹3.2 lakh crore in direct recapitalisation into PSBs over this period. That rebuilt the capital base that a decade of rising NPAs had eroded.
The measurable result: PSB GNPA fell sharply. It dropped from a 2018 peak — system-wide GNPA of 11.46%, with PSBs running higher still — to just 2.50% by September 2025. That’s per RBI’s own supervisory data. The banks originally placed under PCA have all since exited it, too. RBI treats that as a key signal of restored balance-sheet health.
For a credit risk analyst, the PSB case is a useful reminder. A single stressed-asset shock in one segment can be modelled, tracked, and resolved over a defined multi-year horizon. That requires recognition, recovery infrastructure, and capital to all be addressed together. The resolution mechanism itself — IBC recoveries, in this case — is worth tracking too, in loss-given-default estimation, not just PD.
This data has practical implications too. For anyone building PD models, doing IFRS 9 / Ind AS ECL provisioning work, or advising lenders on portfolio risk, a few takeaways follow directly:
Sources: RBI Financial Stability Report, June 2026; RBI Report on Trend and Progress of Banking in India 2024-25; RBI Financial Stability Report, 2018; RBI supervisory data as of 30 September 2025 (cited via Ministry of Finance, Lok Sabha reply, published on pib.gov.in).
Understanding the theory behind PD, LGD, and EAD is the first step. Building bankable, interview-ready models — in Python or SAS, on real credit datasets, aligned to Basel and IFRS 9 — is what actually moves a career forward.
Explore Dexlab Analytics’ Credit Risk Modeling certification program to build PD, LGD, and EAD models from scratch, work through IFRS 9 ECL frameworks, and learn model validation techniques used by practicing risk teams.
.
Credit and Market Risk, Credit Risk, credit risk analysis, Credit Risk Analytics And Modeling, credit risk analytics training, Credit risk certification India, Credit risk modeling course India, Credit Risk Modelling, Credit Risk Modelling Using SAS, Credit scorecard modeling, Expected credit loss ECL, IFRS 9 modeling course India, Logistic regression scorecard, PD LGD EAD modeling, Python credit risk modeling
Comments are closed here.