How does India borrow?

Where non-food bank credit went

Start with the broad map. In April 2007, personal loans were Rs 4.7 lakh crore, services were Rs 4 lakh crore, and industry was Rs 6.7 lakh crore. By April 2026, personal loans were Rs 69.6 lakh crore, services were Rs 59.5 lakh crore, industry was Rs 45.4 lakh crore, and agriculture was Rs 26.3 lakh crore.

That is the central answer. India's bank-credit book has become more personal and more services-heavy. Industry did not disappear; it rose in absolute terms. But it no longer carries the whole story. The useful question is not only whether credit rose. It is where that credit now sits.

The latest credit mix

The latest current-classification month in the IDH/RBI sectoral series is April 2026. The order is now personal loans first, services second, industry third, agriculture fourth. That ordering is a balance-sheet fact, not a mood.

Personal loans are therefore not a side pocket in the banking system. Services are also not a small residual category. Together they explain why the old shorthand of bank credit as mainly industrial credit is no longer enough.

What is inside personal loans?

Personal loans are not one product. Housing or mortgage loans are the anchor at Rs 33.5 lakh crore in April 2026. Other personal loans are Rs 17.3 lakh crore. Vehicle loans are Rs 7.4 lakh crore. Gold loans are Rs 4.9 lakh crore. Credit-card outstanding is Rs 3 lakh crore. Loans against fixed deposits and education loans are each about Rs 1.6 lakh crore.

This is the most important clarification in the article. Housing is still the base, but the personal-credit story is not just homes. It includes unsecured or less clearly labelled other personal loans, vehicle finance, card balances, gold-backed borrowing, education loans and loans against existing assets.

Which personal loans grew fastest?

From April 2007 to April 2026, total personal loans rose about 14.9x. Other personal loans rose about 20.7x. Credit-card outstanding rose about 15.1x. Housing or mortgage loans rose about 14.4x. Vehicle loans rose about 12.2x.

That long-history comparison is only for lines where the old-classification monthly history exists. It says the retail book widened beyond housing. It does not include gold, because the IDH probe did not find a clean old-classification monthly gold-loan pair.

The personal-loan engine

The line chart shows the path behind the ranking. Housing builds steadily. Other personal loans become the second large retail bucket. Vehicles and cards become visible but smaller channels. Education remains much smaller.

Read this as outstanding stock, not monthly disbursement. A rising line means more credit remains on bank balance sheets at the reporting date. It can reflect new borrowing, older loans not yet repaid, longer loan tenors, formalisation, or classification changes.

Since 2019, gold lending grew fastest

Gold needs its own window because the comparable monthly data starts in the current classification. From January 2019 to April 2026, bank gold loans rose from about Rs 0.24 lakh crore to Rs 4.9 lakh crore, about 20x. Over the same window, credit-card outstanding rose about 3.6x, vehicle loans about 2.6x, other personal loans about 3.3x, and housing about 2.9x.

The 20x number should not be read like the 2007-to-2026 chart. It starts from a smaller base and uses a shorter window. But it is still a major signal: gold is the fastest-growing named retail product in the comparable current-classification period.

Gold became a separate retail-credit story

The rupee level now matters as much as the growth multiple. By April 2026, bank gold-loan outstanding was larger than credit-card outstanding and education loans. It was no longer a tiny line hidden at the edge of personal credit.

That does not mean every rupee is household distress. Gold-loan balances can rise because more households borrow, because banks formalise lending that used to sit outside banks, because gold prices raise collateral capacity, or because reporting improves. But gold clearly belongs in the story of how India borrows.

Bank data is not the whole household-debt story

The RBI/IDH charts above are the right source for Indian bank-credit composition. They tell us homes, other personal loans, vehicles, gold, cards and education. BIS Total Credit answers a different question: how much household credit exists across the whole credit system?

On the BIS measure, credit to Indian households and NPISHs from all sectors was 45.5% of GDP in 2025-Q3. In domestic-currency terms, that is about Rs 150.6 lakh crore. That is much larger than the Rs 69.6 lakh crore of bank personal loans in the April 2026 IDH/RBI table because the definitions, lender universe and dates differ.

This is why BIS is useful but not a replacement. It gives scale and a denominator. It does not break Indian household credit into housing, gold, cards, vehicles or education. For those product lines, RBI/IDH remains the source.

Is India highly household-indebted?

Against peers, India is not at the top of the household-debt table. IMF Global Debt Database puts India's household debt at 40.8% of GDP in 2024. Canada is about 100.1%, Korea 90.1%, Thailand 86.7% in its latest 2023 point, the UK 76.2%, the US 69.4%, China 61.4%, Germany 49.9%, Brazil 36.4%, South Africa 33.7%, Mexico 16.7%, and Indonesia 16.2%.

So the article should not say simply that India is uniquely over-borrowed. The sharper point is compositional: India's bank book has changed fast, especially toward personal credit, gold loans and finance-linked services. The macro household-debt ratio is still moderate compared with many richer and East Asian economies.

Banks are not the whole credit system

BIS also shows why bank-credit charts need a boundary line. In 2025-Q3, India's private non-financial credit from all lenders was 97.4% of GDP. The domestic-bank slice was 59.5% of GDP. Banks are central, but they are not the entire credit system.

This matters for services credit too. When banks lend to NBFCs, a loan can sit in the bank-credit table as services or finance-company credit while the end borrower may be a household or business elsewhere. The RBI table shows the bank's immediate borrower category. It does not always show the final household or firm using the money.

Is credit running hot?

The BIS credit-to-GDP gap adds a risk lens. It compares private non-financial credit-to-GDP with a backward-looking long-run trend. For India, the latest BIS gap is -3.1 percentage points in 2025-Q3: actual private non-financial credit is 97.4% of GDP against a BIS trend of 100.5%.

That is not a clean bill of health, and it is not a forecast. But it keeps the interpretation disciplined. The visible Indian story here is not just an economy-wide credit boom. It is a change in where credit sits: retail products, gold collateral, NBFCs, trade, priority-sector categories and infrastructure.

Services credit is finance-heavy

Services credit is not just restaurants, software and transport. In April 2026, NBFCs alone had Rs 20.6 lakh crore of outstanding bank credit. Trade had Rs 13.2 lakh crore. Other services had Rs 12.2 lakh crore. Commercial real estate had Rs 6.4 lakh crore. HFCs and PFIs add more finance-adjacent credit.

That makes services a layered credit bucket. Some of it is ordinary business credit. Some of it is credit to financial intermediaries that may lend onward. Some of it sits around trade and real estate. Services credit is not the same thing as household borrowing, but it is part of the same shift away from the old industry-only shorthand.

Where services credit went

The history of services credit is led by finance-company and trade credit, not by a simple catch-all service economy story. NBFC credit is especially important because banks lending to NBFCs can later become credit to households, small firms or other borrowers outside this table's immediate classification.

That is why services needs to be read carefully. It is a borrower-sector label on a bank balance sheet. It may point to onward lending, working capital, trade finance, commercial real estate, or direct services activity depending on the sub-line.

Priority-sector credit adds another lens

Priority-sector lending is a policy lens, not a clean borrower-sector lens. In April 2026, the priority-sector table shows micro and small enterprises at Rs 29.7 lakh crore, agriculture at Rs 26.2 lakh crore, weaker sections at Rs 19.8 lakh crore, housing at Rs 10.4 lakh crore, and medium enterprises at Rs 7.1 lakh crore.

Do not add those numbers to the non-food sectoral buckets above. The point is different: Indian bank credit is shaped not only by demand for loans, but also by directed-lending categories that push banks toward agriculture, small enterprises, housing and weaker-section lending.

Which industries borrow most?

Industry still matters, and the latest industry-detail map is led by infrastructure at Rs 14.9 lakh crore. Power is Rs 8.5 lakh crore. Basic metals are Rs 5.1 lakh crore, iron and steel Rs 3.5 lakh crore, roads Rs 3.4 lakh crore, engineering Rs 3.1 lakh crore, chemicals Rs 3.1 lakh crore, textiles Rs 3 lakh crore, and gems and jewellery Rs 1.2 lakh crore.

The point is balance. The industrial credit book is still large, capital-intensive and concentrated. But the broad industry bucket is now smaller than personal loans and services in the April 2026 sectoral split.

Infrastructure still anchors industrial credit

Inside industry, infrastructure is the line to watch. It is much larger than most other industrial categories because infrastructure projects are capital-intensive and debt-heavy. Power and roads sit inside that broader infrastructure story, so the industry charts should be read as a hierarchy, not as a list to blindly sum.

Read this as borrower-industry credit outstanding, not as a measure of output, project quality, profitability or whether the borrowing is healthy. The chart tells you where bank credit is parked, not whether every rupee was well used.

The industrial long tail

Once infrastructure is separated out, there is no single clean second industrial story. Engineering, chemicals, textiles, food processing, construction, transport equipment and gems and jewellery each form mid-sized credit pockets.

This long tail is useful because it keeps industry from becoming only an infrastructure chart. Industrial borrowing is still a cluster of many activities, some tied to domestic demand, some to exports, some to construction cycles, and some to commodity prices.

What changed in the data itself?

This article uses IndiaDataHub as the access layer because it exposes RBI Banking/Credit identifiers directly. The expanded pull fetches all grouped monthly lines found in the probe: 88 line items and 188 series variants, producing 22,531 observations. Many line items have old-classification monthly series from April 2007 and current monthly series from January 2019. The chart builder uses old-classification data through December 2018 and current-classification data from January 2019.

BIS and IMF are added only for macro context. BIS Total Credit and BIS Credit-to-GDP Gaps are quarterly cross-country credit-system datasets. IMF Global Debt Database HH_LS is annual household debt by loans and debt securities as a share of GDP. These sources are excellent for scale, peer comparison and risk context; they are not the source for India's loan-product taxonomy.

That splice is useful, but it is not magic. January 2019 is a taxonomy boundary, not just a month. RBI also notes later format and reporting changes, including a revised format from January 2021, merger effects from July 2023, a gold-loan classification note from May 2024, and a reporting-fortnight definition change from December 31, 2025.

What should a reader take away?

The short version is: India's banks are lending into a different economy than the one implied by the old industry-first shorthand. The credit book now has a large household-facing side, a large services-and-finance side, a policy-shaped priority-sector side, and a still-important industrial-infrastructure side.

That is why the composition matters. A rupee of bank credit can be a home loan, a gold loan, a vehicle loan, a card balance, an NBFC loan, trade finance, commercial real estate credit, an MSE loan, a road project, a power loan or gems-and-jewellery credit. The story changes depending on which line is growing.