Hello, and welcome to this week’s edition of Points & Figures, put together by Kulsum.

Last week, we looked at why fewer Indians are flying within the country. This week, we look at the other half of the story: flying abroad.

Until this year, it looked like a market that could only grow. In January–March 2015, about 1.21 crore passengers flew between India and the rest of the world. By October–December 2025, that had risen to 208 lakh a quarter, a rise of more than 70%, with Covid the only big interruption. Then 2026 happened. In April–June, the number fell to 172 lakh, 9% fewer than a year earlier.

The obvious question is whether it is bouncing back. For Indian airlines, which report to DGCA every month, the answer so far is no. They carried 25.1 lakh international passengers in September, 11% fewer than in September 2025. The worst of the fall came in March and April, and traffic has crept back up since, but it is still well short of last year.

There is a twist here. Within India, airlines cut flights about as fast as passengers disappeared, so planes stayed roughly as full as before. Abroad, demand fell faster than the airlines could cut. In August, Indian airlines offered about 7% fewer seats on international routes, measured by distance flown, but the distance flown by their passengers fell 12%. They filled 76% of their seats, against 81% a year earlier, and they have flown emptier than last year in every month since April.

So where did the missing travellers go? Mostly, nowhere.

Before the war, the six Gulf states accounted for about half of all flyers to and from India: 51% in April–June 2025. A year later, that share had fallen to 43%. If people had simply switched destinations, flying to Bangkok or London instead of Dubai, the rest of the world should have picked up the slack. It didn’t. Flyers to and from non-Gulf countries grew 4%, slower than the 8% growth of a year earlier.

One caution on these numbers. DGCA counts passengers by where their flight lands, so someone flying to London via Dubai shows up as a UAE passenger. Part of the Gulf drop is therefore travellers who were only changing planes there. But if those travellers had rerouted through other hubs, we would expect to see it elsewhere, and we don’t. Turkey, home to Istanbul, another big transfer hub, saw 36% fewer flyers. These trips appear to have been put off, not moved.

The one long-distance market that fell sharply was the US: 36% fewer direct flyers in April–June. Air India suspended or cut several long-haul routes for June to August, mostly to North America, citing fuel prices and airspace restrictions. Those cuts began only in June, so they explain part of the fall, not all of it.

The bigger shift is in who is doing the flying.

For most of the past decade, Indian airlines had been steadily winning a bigger share of India’s international travel. In early 2015, they carried 37% of it. Apart from a dip in 2019, when Jet Airways shut down, that share kept climbing, and reached about 46% in 2024 and 2025. In just two quarters, it has fallen back to 37%, where it started a decade ago.

The numbers behind that are stark. In April–June, Indian airlines carried 23 lakh fewer international passengers than a year earlier. Foreign airlines carried 6 lakh more. In other words, foreign airlines picked up only about a quarter of the trips Indian airlines lost. The rest simply didn’t happen.

Air India Express took the biggest hit. It carried half as many international passengers as a year earlier, 8.5 lakh fewer, which on its own accounts for more than a third of the Indian airlines’ loss. Air India fell 27% and IndiGo 15%. Meanwhile, most of the big foreign airlines flying to India held steady or grew.

Air India Express took the biggest hit. It carried half as many international passengers as a year earlier, 8.5 lakh fewer, which on its own accounts for more than a third of the Indian airlines’ loss. Air India fell 27% and IndiGo 15%. Meanwhile, most of the big foreign airlines flying to India held steady or grew. Emirates, the largest of them, carried 7% more passengers to and from India, even as total traffic between India and the UAE fell 22%.

That says something about how the two kinds of airlines are built. Big Gulf airlines like Emirates make much of their money carrying passengers onward from their home hubs to the rest of the world, so they could keep flying to India and fill seats with connecting travellers. Indian airlines, especially Air India Express, rely far more on direct trips between India and the Gulf, and those were exactly the trips that disappeared. That also answers the caution from earlier. The Gulf drop was mostly people travelling to the Gulf, not through it.

Some of this may reverse as Gulf routes recover, and Indian airlines have one thing working for them. Foreign airlines can only fly as many seats into India as treaties allow. Dubai’s airlines, for instance, have been capped at about 66,000 seats a week each way since 2014, a limit Emirates hit long ago. Unless India raises it, Emirates can fill its planes fuller, but it can’t add flights to take over the space Indian airlines vacated. Winning that space back, though, means putting planes back in the air with fuel still expensive and most Indian airlines losing money.

DGCA’s data for July–September, which will include foreign airlines, is due in the coming weeks. It will show whether Indian airlines have started to claw back share, or whether 37% is the new normal.

See you next edition :)

Subscribe now

Hello, and welcome to this week’s edition of Points & Figures, put together by Kulsum.

Indian airlines carried 118.2­ lakh passengers within the country in September 2026. That’s 6.5% fewer than a year ago, and the fewest in any month since November 2022.
This is not how Indian air travel usually behaves. Leave out the Covid years, and the number of domestic flyers has grown every single year for a decade, from about 10 crore in 2016 to 16.7 crore in 2025. Airlines were betting that this would continue. In 2023, IndiGo ordered 500 planes and Air India ordered 470, two of the biggest aircraft orders ever placed.

Instead, September was the fourth month in a row with fewer flyers than a year earlier. And the gap has kept widening: 4.8% in July, 6.3% in August, 6.5% in September.

(The jump in May is a quirk of comparison. In May 2025, 32 airports were shut to civilian flights for several days during Operation Sindoor, so May 2026 is being compared against an unusually weak month.)

So what changed? A war. The conflict in West Asia has hit Indian aviation twice. The first blow landed suddenly, on flights to the Gulf. The second came more slowly, through the price of jet fuel. It’s the second that is now shrinking travel within India. But the first is worth understanding too, because it shows just how exposed Indian flying is to that one corner of the world.

Before the war, roughly half of all flyers between India and the rest of the world were travelling to or from the six Gulf states. That isn’t surprising: about one crore Indians live there, 43 lakh in the UAE alone. Then, on 28 February, after US and Israeli strikes on Iran, Gulf countries shut their airspace. Within days, Indian airlines had cancelled more than 1,200 flights, and the number of international flyers they carried fell 35% in a month.

The traffic didn’t snap back. Even in April–June, India had 21 lakh fewer Gulf flyers than a year earlier. Kerala, with its deep ties to the Gulf, felt it the most. Its four international airports usually carry about a quarter of India’s Gulf traffic, but they accounted for a third of the drop. At Kannur, Gulf traffic fell by almost two-thirds.

Foreign airlines came through it better. In April–June, Indian airlines carried 27% fewer international passengers, even as foreign airlines carried 6% more. Indian carriers were also nursing an older handicap. Pakistan has kept its airspace closed to them since April 2025, forcing longer detours on flights to Europe and North America that Air India estimated would cost it about ₹5,000 crore a year.

The second blow, through fuel, was slower, but it reached every single flight. Fuel makes up 30–40% of what it costs an Indian airline to fly. And according to IATA, jet fuel prices rose 121% worldwide in the year to April, and 170% in India, in rupee terms.

Indian taxes make every rise bigger. Jet fuel is kept out of GST. Instead, it attracts about 11% central excise, plus state VAT that runs as high as 25–29% in some states, and airlines can’t claim any of it back as a credit. Since these taxes are charged as a percentage of the price, every rise in the price of fuel drags a bigger tax bill along with it. Bringing jet fuel under GST would fix much of this, but states have opposed the move, since it would cost them revenue.

The government did step in. State-run oil companies passed on only part of April’s price increase. Landing and parking charges were cut by 25% for three months. And Delhi and Maharashtra slashed their VAT on jet fuel to 7%. In June, a ₹10,000 crore scheme to hold fuel prices steady was approved, but airlines didn’t sign up, and it lapsed unused. Prices kept climbing. Jet fuel for domestic flights cost ₹121.28 a litre in September and about ₹137 in October, nearly 50% more than a year earlier.

All of this left airlines with very little room to absorb the shock. IndiGo, India’s largest airline and usually a reliably profitable one, made a loss of ₹238 crore in April–June, against a ₹2,176 crore profit a year earlier, as its fuel bill rose 86%. Air India and Air India Express together lost ₹22,238 crore last financial year, a hole dug mostly before the war began. And ICRA expects the industry as a whole to lose ₹36,000–38,000 crore this year.

That left airlines with two levers to pull: fly less, and charge more.

From 1 June, Air India and IndiGo trimmed their domestic schedules for three months. Between June and September, India saw about 22,000 fewer domestic flights than a year earlier, a 6% drop. Tickets got costlier too. IndiGo has been adding a fuel charge to tickets since March, and the government told Parliament that average fares on 72 domestic routes were 20.5% higher in June 2026 than in March 2025. Even so, IndiGo’s Rahul Bhatia said the airline could pass on only part of the fuel increase to passengers. There’s a limit to how far fares can go. Push them too far, and people simply stop flying.

And people did stop flying.

You might expect fewer flights to mean fuller planes. That isn’t what happened. Flyers fell roughly as fast as airlines cut seats, so planes stayed about as full as they were a year earlier. August shows what happens when airlines don’t cut. That month, they flew only 2.8% fewer flights than a year before, and just 79% of seats were filled. Leave out the Covid years, and that’s the lowest share for any month since at least 2016.

That’s the bind airlines are in. Cut flights, and they give up passengers along with the costs. Keep flying, and the seats go empty. Their losses suggest that, at today’s fuel prices, neither option works particularly well.

And a squeeze like this hurts most for the airline with the least room to manoeuvre.

That airline is SpiceJet. It carried 6.1 lakh domestic passengers in December 2025. In September 2026, it carried just 73,000, on about 24 flights a day. Only around 10 of its planes are flying, while 37 sit parked, and some of its pilots have refused to fly over salaries that have been delayed since April. In August, it cancelled 20.92% of its flights. To be fair, SpiceJet was a small player even before this, with IndiGo and the Air India group already carrying about nine in every ten domestic flyers. But its slide takes away one of the few alternatives passengers had left.

The airlines that are still flying are being choosy about where they send their planes.

On India’s three busiest routes, Delhi–Mumbai, Bengaluru–Delhi and Bengaluru–Mumbai, the number of flyers actually rose 7% in September from a year ago. On all other routes put together, it fell 8%. Of the next 22 busiest routes, 20 lost flyers. That’s no accident. These three routes connect the country’s biggest business cities, where planes tend to fill up even when tickets are expensive, because many of those on board have to make the trip anyway. When planes are scarce and fuel is costly, those are the seats an airline can count on selling.

Everyone else is paying for that. Of India’s ten busiest air hubs, only Delhi and Pune had more flyers in September than a year ago. Goa, a holiday destination where travel is more of a choice, lost 15%. Hyderabad, Kolkata and Chennai each lost between 12% and 16%. And smaller towns were losing flights even before all of this: only 348 of the 679 routes launched under UDAN, the government’s regional connectivity scheme, are still running.

Even Mumbai’s shiny new airport has been caught by the timing.

Navi Mumbai airport opened to passengers on 25 December 2025. In September, it handled 3.3 lakh domestic flyers, or 11% of the city’s total, down from 16% in May. Meanwhile, Mumbai’s two airports together carried almost exactly as many flyers as they did a year earlier. A second airport is supposed to increase the city’s air traffic. So far, this one has mostly split it, though holding steady is no small thing in a year when the country as a whole is flying less.

What happens next depends mostly on fuel, and that isn’t getting any easier. October’s price is the highest this year, and IndiGo raised its fuel charge again on 6 October. Air India is reported to be seeking to delay hundreds of the planes it ordered, a sharp turn from the confidence of 2023. And the government is considering reviving its fuel scheme.

October to December is usually a busy stretch for Indian airlines, with festivals, weddings and winter holidays. This year, those months will show how much of India’s flying habit was built on cheap tickets, and how many people come back to the airport once tickets stop being cheap.

See you next edition :)

Subscribe now

Hello, and welcome to this week’s edition of Points & Figures, put together by Kulsum.

Suppose someone owes you money and gives you a cheque. The cheque bounces. You take the legal route to recover the payment, and now you are waiting for a court case to move.

One reason it may be stuck is surprisingly basic: getting the summons delivered.

In a September 2025 judgment, the Supreme Court identified serving summons as a major source of delay in cheque-bounce cases. It directed courts to use electronic service under applicable rules and introduce payment facilities through QR codes or UPI links to help disputes settle early.

Someone is waiting for money. The system first needs to get a notice from the other person.

The scale makes this more than a procedural detail. The judgment recorded about 6.5 lakh pending cheque-bounce cases in Delhi’s district courts as of September 1, 2025. They accounted for 49.45% of its trial-court backlog.

Nearly half the pending cases, involving one kind of payment dispute.

In our earlier edition, Justice, pending, we looked at how long cases wait in India’s courts. This time, we wanted to look more closely at what courts are being asked to do. Because the answer changes what it would take to make that wait shorter.

A big shoutout to Data For India’s Ameya Bokil and Apoorv Anand, whose analysis makes much of this edition possible.

The dataset covers roughly 3.3 crore cases, with 4.36 crore records because a case can cite more than one legal provision. It excludes pending cases and covers only part of the court system. The newer cheque-bounce and virtual-court figures below come from separate sources.

The range of laws here is a useful place to begin. Courts are dealing with offences, payments, road rules, alcohol regulation and marriages. All of these become “cases”, even though the work involved can be very different.

Take driving.

Rash driving is the most frequently cited individual IPC provision. The Motor Vehicles Act brings another set of driving-related matters into the picture, including driving under the influence, licence violations and unregistered vehicles.

An accident can also lead to a different kind of proceeding. The Motor Vehicles Act provides for compensation claims, allowing an injured person to approach a claims tribunal.

The same incident can therefore raise separate questions. Did the driver commit an offence? Is someone entitled to compensation? Who should pay?

A note on reading these charts: they count provisions cited, not convictions or the actual frequency of crimes. The IPC labels belong to the period studied; the Bharatiya Nyaya Sanhita came into force in July 2024.

The law also shapes which activities end up in court in the first place. That becomes clear when we look at alcohol.

The excise-law chart might look like a straightforward state comparison. But the rules and the categories differ.

Bihar’s excise records mainly cover the period before its 2016 prohibition legislation. Gujarat’s prohibition law sits in a separate category too.

We cannot use these bars to rank drinking or the strictness of enforcement. They show why understanding a court’s workload requires looking at the laws that send matters its way.

Not everyone comes to court over an offence or an unpaid debt. Some come to end a marriage or ask for married life to resume.

The marriage records include contested divorce and divorce by mutual consent. Alongside them are about 2.85 lakh citations of Section 9 of the Hindu Marriage Act, called restitution of conjugal rights.

This lets a spouse ask for a court order to resume living together as a married couple. The court must be satisfied that the legal conditions are met, including that one spouse withdrew from the other’s company without reasonable excuse.

These records tell us what people are asking courts to do. They cannot tell us India’s divorce rate or how the relationships turned out.

That distinction between a legal proceeding and its outcome matters in business too. Even when parties choose arbitration, there can still be work for a court.

Under the Arbitration and Conciliation Act, courts can provide interim protection, enforce an award or consider an application to set it aside on specified grounds. Those are the three provisions shown here.

For example, money or property involved in a dispute may need protection before the arbitrator reaches a decision. Afterwards, the award may need enforcement. These are roles the law gives courts; their presence in the data does not establish that arbitration has failed.

This takes us back to the person waiting for a cheque payment. Starting a proceeding, reaching a decision and receiving money are different steps. Improving the system means understanding where a particular matter gets stuck.

The cheque-bounce judgment focused partly on notices and payment facilities. Traffic challans have their own route through virtual courts, where people can pay online or ask to contest the case in a physical court.

By January 31, 2026, virtual courts had received 10.13 crore challans and recorded 9.05 crore disposals. Separately, 97.72 lakh challans had been paid, according to the government’s March update.

We cannot read those disposals as payments. The release does not explain the whole gap, and the option to contest a challan does not, by itself, account for it either.

That leaves a question worth asking whenever we hear that cases have been cleared: what does “cleared” mean here?

For a cheque dispute, we would want to know whether the payment issue was resolved. For an arbitration award, whether it was enforced or set aside. For a challan, whether it was paid, contested or dealt with another way.

The number of cases tells us how much work there is. We also need to know what happened to the problems that brought people to court.

See you next edition :)

Subscribe now

Hello, and welcome to this week’s edition of Points & Figures, put together by Kulsum.

When someone in the family finds a job, the first response is usually relief. Then come the questions. How much does it pay? Is it permanent? Will they need to move? Is there room to grow?

Employment numbers answer the first question: are people working? The questions that follow need a closer look.

To find some answers, we bring together the government’s Periodic Labour Force Survey, CareEdge’s September employment report and Azim Premji University’s State of Working India 2026. Our colleagues explored the university’s research in The Daily Brief and a Subtext conversation with its authors, Rosa Abraham and Tamoghna Halder.

Together, these sources help us look beyond whether someone has work to what that work offers.

In 2025, India’s unemployment rate was 3.1% for people aged 15 and above, under the survey’s “usual status” measure. It was 2.4% in rural areas and 4.8% in urban areas.

There are two terms here worth breaking down before explaining what this statistic really means.

For one, the “unemployment rate” measures the unemployed as a share of the labour force: people working, or seeking or available for work. Someone outside that group, such as a student who is neither working nor looking for work, is not counted as unemployed.

Secondly, there’s the phrase “Usual status”, which assesses a person’s activity over the preceding 365 days. Its employment count includes very different arrangements, from salaried positions to unpaid help in a family enterprise.

A lower unemployment rate therefore does not, by itself, establish that people have better-paying or more secure work.

The distinction becomes clearer when we look at working women.

Among employed rural women, 32.9% are unpaid helpers in household enterprises, while 9.3% have regular wage or salaried employment. Among employed urban women, the regular wage or salaried share is 50.9%.

These figures describe women who are already working, but not what proportion of all women have work.

The categories themselves are broad. Self-employment includes both thriving businesses and small activities with uncertain earnings. A salary does not automatically guarantee good pay or job security, either. Still, grouping these experiences under “employed” leaves a great deal unsaid.

For young people, these differences start with the search for work.

In the Subtext conversation, Rosa discusses how family circumstances shape that search. Some people can afford to wait for a suitable opening, while others need to start earning immediately. Taking available work may meet an urgent need without ending the search for something better.

The State of Working India report examines young men’s transition into employment, using survey panels from 2017–19 and 2022–24. Among graduates in its sample, 48.8% found some employment within a year; 6.7% found permanent salaried employment.

The second group is included in the first. The study also includes direct moves from education into work and measures finding employment at least once, rather than keeping it.

These are results for a particular study sample, not the odds facing every graduate today.

To understand the opportunities people encounter, we also need to ask where employment is concentrated.

In 2025, agriculture, forestry and fishing accounted for 43% of employment. Manufacturing accounted for 12.1%, and construction for 12.0%. Office employment of knowledge workers is only one small part of the picture.

For someone seeking work, the scope of an industry is also only a starting point. Its jobs need to be accessible given their location, qualifications and experience. A large employer of workers nationally may offer few suitable openings nearby.

Services broaden the picture further. The term brings technology companies and banks to mind, but it also includes the shop down the road, the transport operator and the local restaurant.

Wholesale and retail trade, including motor-vehicle repairs, account for 10.8% of total employment. Transport and storage account for 4.5%, followed by education at 3.3% and accommodation and food services at 2.2%.

Information and communication, a category broader than IT, account for 1.3%, while finance and insurance account for 1.2%.

Large companies are another part of this employment picture. CareEdge’s analysis offers a narrower view of how their workforce has grown.

Across its sample of 236 Nifty 500 companies, combined headcount increased from about 37.6 lakh in FY18 to 51.3 lakh in FY25. It was almost unchanged in FY26, at the precision reported.

It’s worth nothing that this sample has a very strong sector tilt towards IT and banking, finance and insurance. It cannot stand in for employment across the economy.

Now, what does unchanged headcount mean? It doesn’t have to mean recruitment stopped. A company may hire people to replace those leaving without expanding its total workforce. But this sample’s combined workforce barely expanded in the latest year.

Alongside these different employment opportunities sits a much larger population of degree holders than two decades ago.

State of Working India estimates that India’s graduate population, across all ages, rose from 4.98 crore in 2004 to 14.94 crore in 2023. Employed graduates increased from 3.15 crore to 8.63 crore, and salaried graduates from 1.77 crore to 5.05 crore.

Graduate employment grew substantially, but more slowly than the graduate population.

These are totals in two survey years, not one graduating class followed through its career. Salaried graduates are included among employed graduates. And the difference between the graduate population and employed graduates is not an unemployment count: it also includes people outside the labour force.

The comparison leaves us with a broader question. As access to education expands, what opportunities are available to the people gaining those qualifications?

Among salaried workers aged 20–29 in the report’s 2023 observations, graduate men earned an average of about ₹19,573 a month, compared with ₹10,507 for non-graduate men. For women, the corresponding averages were ₹18,829 and ₹7,199.

These figures are expressed in 2020 rupees: prices have been adjusted to a common year so inflation does not distort comparisons. They are neither today’s salaries nor starting salaries.

The comparison also does not establish that a degree alone causes the entire earnings difference. It describes young people who are already in salaried employment.

Within that group, however, graduates earn more on average. The difficulty of finding work and the earnings advantage associated with education can both be true.

The unemployment conversation is too vast to be covered with a single metric. An unpaid family helper, a casual worker and a salaried employee may all count as working, while facing very different financial circumstances.

The challenge has two parts: expanding opportunities for people entering the workforce and improving the work people already do. Education matters, but so do the opportunities available after it.

For a family, finding work can bring immediate relief. The next questions are whether the income covers expenses, whether there is anything left to save, and whether it will still be there next month.

See you in the next edition :)

Subscribe now

Hello, and welcome to this week’s edition of Points & Figures, put together by Kulsum, with a special shoutout to our in-house automobile expert, Vignesh, for sharing his insights.

Indians bought more than 24 lakh vehicles in August 2026, the highest ever for that month. That works out to roughly 80,000 new registrations a day. But the total brings together several vehicle markets, each moving at its own pace.

A vehicle is one of the most relatable big purchases an Indian household makes. It can shorten a commute, make a distant job practical, carry children to school or help a driver earn a living. It can also be the household’s first large loan.

The same registration number can therefore mean convenience, aspiration, work or debt, depending on what someone bought and why.

The mix gives us a better sense of what is happening. Of the 24.23 lakh vehicles registered in August, 17.15 lakh were two-wheelers. Passenger vehicles accounted for 4.02 lakh, followed by three-wheelers, commercial vehicles and tractors. A little over seven in every ten registrations had two wheels. We may talk about the Indian vehicle market through car launches, but the motorcycle and scooter remain at its centre.

There is a fairly simple reason for that. An entry-level two-wheeler costs far less upfront than an entry-level car. Both bring recurring expenses such as insurance, fuel or charging, servicing and, if bought on a loan, monthly installments. A car generally costs much more and is harder to park. So a rise in two-wheeler registrations and a rise in passenger-vehicle registrations tell us about different buyers.

August is part of a longer rise in vehicle sales. FADA, the Federation of Automobile Dealers Associations, tracks retail registrations across dealerships. Its figures show registrations climbing from 2.39 crore in 2023 to 2.61 crore in 2024 and 2.82 crore in 2025.

Passenger vehicles are growing within that larger market. Their retail registrations rose 22.2% year on year during April–August 2026, faster than the 18.1% increase in two-wheelers. But that alone does not tell us that cars have become broadly affordable, or that household incomes have grown at a similar pace. The comparison period made growth look stronger, and a large tax cut had recently helped sales.

In September 2025, the GST Council cut the tax rate on small cars and motorcycles with engines of up to 350cc from 28% to 18%. Larger sport utility vehicles also saw a lower effective tax burden. The effect showed up in year-end sales: Maruti Suzuki reported an almost 50% jump in small-car sales to dealers in December.

The conclusion is narrower than “India can now afford cars”. Tax cuts made several models cheaper and helped bring forward purchases that people had put off. Even after that relief, a new car remains beyond many household budgets.

Where those vehicles are being registered is changing too. In August, registrations at rural RTOs grew faster than those at urban RTOs across the major vehicle categories. Rural two-wheeler registrations rose 20.25% year on year, against 19.07% at urban RTOs. The gap was wider for passenger vehicles: 24.99% against 10.93%. Commercial vehicles grew 16.33% at rural RTOs and 12.79% at urban ones.

Those figures need some care. FADA calls a registration rural or urban based on the location of the RTO, not the buyer’s income or occupation. And rural vehicle demand is about more than farming. Two-wheelers, cars and commercial vehicles are all part of it. Tractor registrations were almost flat in August, up just 0.84% year on year, even as the other major categories grew faster at rural RTOs. The figures point to broader demand for mobility, though they do not tell us that every part of the rural economy is booming.

There is another shift alongside where people buy vehicles: what powers them. India’s electric vehicle market has grown large enough that percentage growth from a tiny base no longer tells the whole story. VAHAN data published by the Ministry of Heavy Industries shows annual EV registrations rising from 1.74 lakh in 2019–20 to 4.59 lakh in 2021–22, 11.83 lakh in 2022–23, 16.81 lakh in 2023–24 and 19.68 lakh in 2024–25. EVs went from less than 1% of registrations before the pandemic to 7.5% in 2024–25.

That growth looks very different depending on the vehicle. In August 2026, electric vehicles accounted for 65.3% of three-wheeler registrations, 10.68% of two-wheelers, 7.63% of passenger vehicles and 5.18% of commercial vehicles. A single national EV share hides four quite different markets.

Three-wheelers moved first partly because electric rickshaws already made up a large share of the category. Many three-wheelers are also used to earn a living. When a vehicle travels long distances every day, fuel savings can add up faster than they would for a lightly used private vehicle. But this category is shifting too. Passenger e-rickshaw registrations fell 22% year on year in August, while other passenger three-wheelers rose 23% and goods three-wheelers rose nearly 43%. Electric three-wheelers are no longer just an e-rickshaw story.

Two-wheelers crossed 10% electric penetration in August, while electric passenger vehicles remained below 8%. That gap is not simply a matter of whether buyers like EVs. Purchase price, daily distance, access to home or depot charging, available models and expected resale value all differ across categories. FADA’s figures show us where adoption is happening; they cannot tell us which factor drove each purchase.

The choices are varied even within passenger vehicles. In August, CNG accounted for 25.28% of passenger-vehicle registrations, while EVs had a 7.63% share and hybrids remained a much smaller part of the market. Buyers are spreading across several fuel options, rather than moving in a straight line from conventional cars to electric ones.

Strong registrations do not mean dealers have nothing to worry about, either. Passenger-vehicle inventory at dealerships rose from roughly 28–30 days at the end of April to 38–40 days by the end of August. FADA recommends 21 days. That does not automatically mean people have stopped buying cars. Dealers tend to stock up before the festive season because they expect sales to rise. But more unsold cars tie up more working capital, leaving dealers more dependent on festive demand meeting those expectations.

India’s vehicle market is growing in several directions at once. Two-wheelers still dominate everyday mobility. Lower taxes have helped passenger-vehicle sales. Registrations at rural RTOs are rising quickly, though nearly flat tractor sales show that rural demand is uneven. EV adoption is growing too, led by three-wheelers, while private cars lag behind.

Behind each trend is someone making a different calculation. One household may be wondering whether a scooter can replace two bus changes. Another may be weighing a car installment against rent and school fees. A three-wheeler driver may be thinking about how much of the day’s earnings will go towards energy. Those decisions are what give the registration figures meaning: a vehicle changes where someone can go, what work they can take and how much the journey costs them.

Subscribe now

Hello, and welcome to this week’s edition of Points & Figures, which arrived early, put together by Bhuvan. It’s a long one so stay with us :)

Executive Synthesis: The Anatomy of a Multi-Front Supply Squeeze

The 18 charts in this dossier do not describe a standard late-cycle cyclical slowdown. They map the architecture of an aggressive, synchronised supply shock rippling through global and domestic transmission plumbing.

Unlike the acute demand collapse of 2020 or the single-point energy embargo of 2022, the macro landscape in September 2026 is defined by a simultaneous tightening across three physical choke points:

1. Upstream Energy & Finished Distillates: Brent crude has vaulted past $111/bbl, while the physical refining complex has decoupled entirely. New York Harbor diesel futures are up 118% year-over-year at $4.99/gallon, blowing out the crack spread over crude to an extraordinary $95.15/barrel. This is compounded by acute buffer exhaustion: U.S. distillate inventories sit 12.8% below their five-year seasonal average, while global commercial oil inventories are drawing down at an aggressive 1.45 to 1.71 million barrels per day.

2. Industrial Feedstocks & Maritime Freight: Physical logistics and electrification metals are compounding landed input costs. The Baltic Dry Index has surged 51% over the past year to 3,327 on Capesize vessel shortages, COMEX copper is hovering near record highs at $6.53/lb (+41% YoY), and European TTF natural gas has climbed 56% to $17.93/MMBtu ahead of winter with depleted seasonal storage.

3. Global Agricultural Pressures: FAO’s broad food price index reached 133.3 in August, led by an 11.9% monthly explosion in sugar and broad-based gains across all five constituent food groups. In India, wholesale food inflation has violently swung from deep disinflation of -8.5% in October 2025 to +5.4% today, eliminating the food base-effect cushion.

The Central Domestic Tension: The 510-bps WPI–CPI Margin Dam

The defining macroeconomic vulnerability for India is the divergence between wholesale and consumer price formation. Wholesale inflation (WPI) is running at 9.92%, while headline retail inflation (CPI) printed at 4.82% in August.

A 510-basis-point wedge indicates that corporate balance sheets and state oil marketing companies (OMCs) are currently absorbing severe input cost shocks. However, corporate gross margins cannot absorb $111.71/bbl crude basket costs and double-digit raw material increases indefinitely. Crucially, retail services are already cracking: restaurant and accommodation inflation has surged from 2.87% in January to 8.38% in August.

The Sovereign Debt Reality: The End of Rate Relief

In past commodity spikes, central banks possessed the latitude to “look through” transient supply shocks by providing liquidity cushions. In September 2026, that luxury is dead. The 10-year U.S. Treasury yield is locked at 5.00%, with the 20- and 30-year long end trading above 5.36%–5.40%, re-anchoring global term premia at 16-year highs.

Simultaneously, India’s 10-year benchmark bond yield has climbed 59 basis points in a year to 7.074%. Trapped between imported energy inflation, widening trade deficits, and sticky domestic core services, the Reserve Bank of India faces a stark trilemma: defending growth through liquidity operations risks exacerbating rupee depreciation and importing further inflation.

Section I: Upstream Energy, Refining Chokepoints & Physical Depletion

1. India Crude Basket: The Return to Triple Digits & The OMC Trilemma

India’s crude oil import basket averaged $111.71 per barrel through the first 15 days of September 2026, a steep $21.52/bbl escalation from August’s completed average of $90.19/bbl. On 15 September alone, the daily spot basket closed at $131.19/bbl, approaching the historic stress levels seen in June 2022 ($116.01/bbl monthly average).

With India importing over 85% of its crude requirements, a sustained triple-digit crude price forces an acute policy trilemma across three transmission buffers:

• The Corporate Buffer (OMC Under-Recoveries): If state-owned oil marketing companies (IOCL, BPCL, HPCL) freeze retail petrol and diesel pump prices to protect consumer purchasing power, marketing under-recoveries blow out rapidly, impairing balance sheets and slashing dividend yields to the exchequer.

• The Household Buffer (Pump Price Pass-Through): Passing the $21/bbl shock directly to retail retail fuel pumps immediately ignites transport inflation, which has already jumped from 0.09% to 4.60% YoY in CPI.

• The Fiscal Buffer (Excise Tax Cuts): Relieving retail consumers by slashing central excise duties erodes fiscal consolidation targets, forcing higher government market borrowing in an already saturated bond market.

Every dollar increase in the crude basket widens India’s current account deficit by an estimated $1.5 billion annually. At $111+/bbl, energy ceases to be a background macro variable and becomes an active constraint on fiscal and monetary policy.

2. NY Harbor ULSD Futures: The 118% Explosion in Industrial Fuel

While crude oil has dominated political headlines, the real systemic crisis is concentrated in middle distillates. New York Harbor Ultra-Low Sulfur Diesel (ULSD) continuous futures closed the second week of September at $4.99 per gallon, marking a 118% surge over the past 52 weeks from the sub-$2.30 levels prevailing in late 2025.

Diesel is not a discretionary consumer fuel; it is the universal circulatory fluid of the global economy:

• It powers long-haul heavy road transport, freight rail networks, inland barges, and ocean feeder vessels.

• It operates off-road agricultural machinery during the critical autumn harvest season and runs construction earthmovers and mining haulers.

• It fuels industrial backup generation across emerging markets facing power grid volatility.

Because freight and logistics operators operate on thin margins with automated fuel surcharge mechanisms, a 118% increase in diesel futures transmits with near-zero lag into landed wholesale goods prices, compounding distribution costs across every tier of manufacturing.

3. Diesel–Brent Crack Spread: The Broken Refining Transmission Chokepoint

The weekly diesel crack spread—the benchmark margin between New York Harbor diesel spot prices ($4.928/gal × 42 = $206.98/bbl) and physical Brent crude ($111.83/bbl)—vaulted to $95.15 per barrel for the week ending 11 September 2026. This is just shy of the all-time historic peak of $111.50/bbl recorded during the initial outbreak of the European war in May 2022, and dwarfs the 10-year historical floor of $4.70/bbl observed in June 2020.

This $95/bbl crack spread dismantles the conventional assumption that fuel prices track upstream crude linearly. It proves that the bottleneck is downstream in secondary conversion refining:

• Global refining capacity remains severely constrained following structural closures of older hydroskimming refineries across North America and Europe between 2020 and 2023.

• Specialized desulfurization, hydrocracking, and catalytic reforming capacity cannot keep pace with sulfur-spec regulations and high heavy-crude slates.

• Middle distillate yields cannot be expanded arbitrarily without producing unmarketable surplus naphtha or low-value fuel oil.

As long as the refining crack spread remains above $90/bbl, global fuel users will face emergency-level energy costs even if crude oil production were to expand tomorrow.

4. U.S. Distillate Inventories: Seasonal Exhaustion Entering Autumn

Physical distillate stocks in the United States stood at 107.86 million barrels for the week ending 11 September 2026. This is 12.8% below the five-year seasonal average (123.7 million barrels) and sits below every single comparable calendar week in the 2021–2025 reference envelope.

Inventories are the physical shock absorber of a commodity market. When stocks are robust, sudden supply interruptions—such as Gulf Coast refinery outages or regional pipeline bottlenecks—can be met by drawing down storage without causing wild price dislocations.

At 107.9 million barrels, that operational cushion is exhausted. The United States is entering the high-demand autumn agricultural harvest and winter residential heating oil season with zero inventory elasticity. In an inelastic supply regime, any marginal demand spike or refinery glitch must be resolved through severe price rationing.

5. OECD Commercial Oil Inventories: Cushion Drained to 58.8 Days

According to the EIA September 2026 Short-Term Energy Outlook, OECD commercial oil inventories provided forward demand coverage of just 58.8 days of supply in August 2026, breaching the lower bound of the five-year same-month historical range (58.9 to 61.6 days).

Forward days of supply is the single most comprehensive metric of global supply resilience, capturing commercial crude and refined product holdings across all developed economies relative to projected daily consumption (excluding government-held Strategic Petroleum Reserves):

• A reading below 59 days signals that developed market supply chains are operating on just-in-time logistics.

• Compounding the thin commercial cushion, the U.S. and European strategic petroleum reserves have not been comprehensively replenished following their 2022 emergency drawdowns, leaving public authorities with limited dry powder to suppress future price spikes.

The narrative of global energy abundance is directly contradicted by physical balance sheets.

6. Global Oil Inventory Trajectory: The Deepening 1.5 mb/d Structural Deficit

Global oil markets are running a sustained, structural physical deficit. Official EIA data confirms that global oil inventories drew down by 1.45 million barrels per day in Q3 2026, with the projected liquidation rate accelerating to 1.71 mb/d in Q4 2026. This marks the third consecutive quarter of heavy inventory liquidation following a 1.25 mb/d draw in Q2.

This structural deficit reflects three intersecting dynamics:

• Strict OPEC+ production discipline and quota adherence, reinforced by voluntary Saudi and Russian export curbs.

• Persistent upstream capital expenditure discipline among Western publicly listed oil majors, who continue to prioritize dividend payouts and share buybacks over volume expansion.

• Non-OECD demand resilience, particularly across developing Asian transport and petrochemical infrastructure, which has offset softening Western consumer demand.

A physical deficit of 1.5 to 1.7 million barrels every day cannot be solved by financial hedging. It requires physical inventory liquidation, driving days of supply ever lower and cementing a tight floor under global crude.

7. S&P GSCI Index: Upstream Commodity Inflation Vaults 39%

The S&P GSCI Spot Index closed at 756.42 index points for the week ending 14 September 2026, marking a 38.6% year-over-year advance. While still below its historic post-invasion spike of 808 in mid-2022, the index’s trajectory over 2026 reflects a powerful upstream cost push.

Because the S&P GSCI is production-weighted—allocating approximately 60% of its weight to energy commodities alongside industrial metals, precious metals, and agriculture—it functions as the primary early-warning barometer of upstream producer price inflation. The synchronized ascent of the index confirms that the inflation impulse is no longer isolated to episodic consumer categories; it represents a comprehensive re-rating of physical extractive inputs across the real economy.

Section II: Industrial Feedstocks, Freight Chokepoints & Power Grids

8. COMEX Copper: Doctor Copper at $6.53/lb and the Arbitrage Fracture

COMEX continuous copper futures traded at $6.53 per pound for the week of 14 September 2026, representing a 41.1% gain over the past 52 weeks and resting just shy of August’s all-time ten-year nominal peak of $6.71/lb.

Traditionally revered as “Doctor Copper” for its ability to diagnose the health of global industrial production, the current price surge carries a more nuanced diagnostic signal:

• Structural Electrification Capex: Structural secular demand for grid expansions, data center power delivery, renewable installations, and electric vehicles continues to generate an inelastic baseline demand for refined copper cathode.

• Upstream Supply Strains: Chronic mine supply downgrades across key Latin American producers (water constraints in Chile, social opposition in Peru) have drastically reduced smelter treatment and refining charges (TC/RCs) to near-zero levels, squeezing refined output.

• Regional Dislocation and Tariff Frictions: The premium of COMEX prices over London Metal Exchange (LME) and Shanghai Futures Exchange (SHFE) contracts reflects severe tariff distortions and freight chokepoints, resulting in regional inventory hoarding in North America while Asian cathode stocks remain uneven.

Copper is not signaling a roaring cyclical factory boom; it is signaling acute extractive and processing scarcity colliding with non-negotiable electrification mandates.

9. Commodity Futures Relative Performance: The Post-2019 Metals Re-Rating

When rebased to 100 on 16 September 2019—the common unbroken starting point across all seven continuous metal contracts—the past seven years reveal a dramatic secular hierarchy across metals, led by a massive tripling of precious and core industrial contracts:

• Silver (Index 359 / +259%): Up 42% over the past 12 months, propelled by dual tailwinds: monetary reserve hedging and aggressive industrial off-take in photovoltaic solar cells and automotive electronics.

• Gold (Index 285 / +185%): Up 37% over the past year, reflecting structural, non-price-sensitive accumulation by non-Western central banks diversifying foreign exchange reserves away from G7 fiat assets.

• Copper (Index 250 / +150%): Up 41% YoY, more than doubling over the seven-year period as the quintessential electrification backbone.

• Aluminum (Index 190 / +90%): Up 29% YoY, nearly doubling since its contract revival as power-intensive smelter curtailments across Europe and Asia elevate baseline production costs.

• Platinum (+89%) and Zinc (+68%): Form a tighter secondary cluster, with zinc rebounding strongly over the past 12 months on smelter power cost pressures.

• Palladium (-20%): The lone decliner over the 7-year horizon, weighed down by internal combustion catalytic converter demand headwinds and substitution into platinum.

The critical takeaway is the shift from dispersion to convergence: between 2019 and 2024, precious metals, battery metals, and bulk industrials took wildly divergent paths. Over the past 12 months, all seven continuous metals series have pushed upward together, creating an unavoidable procurement drag on global capital expenditure budgets.

10. Baltic Dry Index: Dry-Bulk Maritime Freight Rates Vault 51% to 3,327

The Baltic Dry Index (BDI)—the benchmark gauge of international dry-bulk maritime shipping costs—surged to 3,327 on 14 September 2026, marking a 51.0% annual increase and standing near its five-year cyclical highs.

The BDI captures the cost of moving unrefined, raw physical commodities across oceans: iron ore for blast furnaces, thermal and coking coal for utilities and steel mills, bauxite for aluminum smelters, and bulk grain shipments. The current surge is heavily concentrated in Capesize vessels (ships of 150,000+ DWT):

• Geopolitical detours around African transit capes (avoiding the Red Sea / Suez route) continue to soak up vessel-ton-miles, effectively reducing global fleet efficiency.

• Environmental fleet speed regulations (CII/EEXI compliance) have forced older vessels to reduce operational speeds, artificially restricting effective spot carrying capacity.

• Port congestions in East Asia and South America have tied up Capesize hulls, forcing dry-bulk charterers to bid aggressively for prompt tonnage.

For resource-importing economies like India, rising freight rates act as an unhedged tariff on raw inputs, lifting the landed rupee cost of imported coal, metallurgical inputs, and fertilizer feedstocks before processing even begins.

11. European Natural Gas: TTF Rebounds 56% to $17.93/MMBtu Ahead of Winter

The IMF European natural gas benchmark (Dutch Title Transfer Facility / TTF) reached $17.93 per MMBtu in July 2026, rising 55.8% year-over-year from $11.51/MMBtu in July 2025.

While still well below the catastrophic panic levels of August 2022 ($70+/MMBtu), Europe’s structural vulnerability has returned in a new form: the exhaustion of the gas cushion:

• European underground gas storage injection rates slowed markedly through the summer of 2026, leaving storage trajectories at risk of entering winter below the critical 75% target.

• Global liquefied natural gas (LNG) markets have become fiercely competitive, as unseasonable heatwaves across Northeast Asia forced Japanese and Chinese utilities to outbid European terminals for flexible Atlantic Basin LNG cargoes.

• Key transit agreements governing remaining pipeline flows through Eastern Europe expire at year-end, threatening a definitive cutoff of residual volumes.

Natural gas is the marginal pricing fuel for European electricity grids and the essential feedstock for synthetic nitrogen fertilizers. A sustained price above $17/MMBtu guarantees elevated industrial power tariffs across Europe and threatens another round of global chemical and fertilizer price inflation.

Section III: Global Food Pressures & Domestic Agricultural Inversion

12. FAO Food Price Index: All Five Commodity Pillars Inflating in Unison

The Food and Agriculture Organization (FAO) Food Price Index rose to 133.3 in August 2026, registering its highest reading since late 2022. The headline move was electrified by an 11.9% month-on-month surge in the Sugar Price Index, driven by severe harvest downgrades in Brazil and Southeast Asia.

However, the more profound structural threat is breadth. Every single one of the FAO’s five major sub-indices recorded positive price momentum in August:

• Sugar: +11.9% MoM on dry weather and milling delays.

• Vegetable Oils: Propelled higher by palm oil export curbs and soaring crude oil prices, which redirect soybean and rapeseed oils into biofuel refining mandates.

• Cereals: Rebounding on weather disruptions in North American wheat belts and Black Sea export risk.

• Dairy and Meat: Lifting steadily on elevated feed grain and logistics costs.

When one agricultural commodity spikes in isolation, food processors and household consumers substitute into cheaper alternatives. When all five pillars of the food basket inflate simultaneously, substitution becomes mathematically impossible, transmitting directly into food insecurity across import-dependent emerging nations.

13. India Wholesale Food Inflation: The Vicious Turn from -8.5% to +5.4%

India’s Wholesale Food Index has undergone a violent 1,390-basis-point regime shift: after bottoming out in deep disinflation at -8.49% YoY in October 2025, wholesale food inflation rebounded to +5.44% in August 2026.

This reversal permanently breaks the primary disinflationary anchor that shielded Indian consumer indices over late 2025:

• The Inversion of Base Effects: During 2025, elevated prior-year food price bases produced favorable statistical year-on-year declines. As of mid-2026, those statistical base effects have completely expired, turning wholesale food into an active inflationary engine.

• The Transmission Pipeline: Wholesale food price moves precede retail food CPI prints by four to eight weeks. With wholesale food running at +5.4% and international edible oil and sugar prices climbing, the primary food component of CPI (carrying a dominant ~46% basket weight in the legacy series and over 39% in the 2024 series) is poised for further upward revisions.

Section IV: Domestic Margin Compression & Retail Transmission Lag

14. India Wholesale Inflation (WPI): The 9.92% Upstream Dam

India’s headline Wholesale Price Index (WPI) inflation accelerated to 9.92% YoY in August 2026, up from 3.98% in March. It is now knocking on the door of double digits.

The critical insight of this chart is the widening chasm between wholesale prices (+9.92%) and consumer retail prices (+4.82%). This 510-basis-point gap represents the pressure behind the dam:

• Wholesale prices measure factory-gate and mandis prices across raw materials, primary articles, fuel, and bulk manufactured goods.

• Consumer prices measure the prices paid by households at the final retail checkout counter.

• A 510-basis-point wedge indicates that Indian manufacturing enterprises, FMCG corporations, and logistics providers have so far absorbed a massive portion of input cost increases within their operating profit margins, choosing market share preservation over retail price hikes.

However, corporate margin absorption has hard empirical limits. As working capital debt costs rise alongside bond yields and inventory carrying costs escalate, corporations will be forced to pass these accumulated wholesale cost pressures through to the end consumer.

15. India CPI Components: The Entrenched Advance of Dining and Services

While headline CPI printed at an ostensibly manageable 4.82% YoY in August 2026, the internal component breakdown exposes an alarming acceleration in non-food core services.

Inflation in Restaurants and Accommodation (Dining & Hotels) skyrocketed from 2.87% in January to 8.38% in August 2026—an increase of 5.51 percentage points in just eight months. Simultaneously, Food and Beverages climbed to 5.95%, and Personal Care & Miscellaneous goods rose to 15.17%.

This dynamic dismantles the argument that Indian inflation remains a temporary “perishable vegetable vegetable shock”:

• Dining, hospitality, and personal services have zero dependence on monsoon rainfall variability. Their pricing reflects commercial real estate rents, commercial LPG tariffs, retail wage pressures, and the compounding cost of processed food ingredients.

• Service price increases are structurally sticky: while tomato or onion prices can collapse 50% in three weeks following a fresh mandi arrival, restaurant menus, hotel room tariffs, and salon services virtually never cut prices once raised.

16. CPI Fastest Risers: The Sequential Broadening of the Inflationary Impulse

Tracking the trajectory of individual CPI divisions between January and August 2026 reveals a dangerous sequential broadening:

• Dining and Accommodation: Surged from 2.87% to 8.38% (+5.51 percentage points), pulling away completely from the headline trend.

• Transport: Surged from 0.09% in January to 4.60% in August (+4.51 percentage points), capturing the sudden transmission of diesel and commercial fleet freight hikes.

• Food & Beverages: Advanced from 3.77% to 5.95% (+2.18 percentage points).

• Paan, Tobacco & Intoxicants: Rose 1.64 percentage points.

• Information & Communication: Gained 1.32 percentage points on telecom tariff hikes.

• Benchmark Headline CPI: Advanced 2.08 percentage points (from 2.74% to 4.82%).

The trajectory proves that inflation is migrating across sectors in waves. The early risers (dining and core services) established a high baseline; the intermediate movers (transport and telecom) joined in the second quarter; and food is accelerating into the third. Inflation that broadens across multiple unrelated divisions is far harder for a central bank to control than an isolated spike in a single commodity.

Section V: Sovereign Yield Surge & The Death of the Global Pivot

17. India 10-Year G-Sec Yield: Breaching 7.07% and the RBI’s Straightjacket

India’s benchmark 10-year government bond yield closed at 7.074% on 15 September 2026, marking a 58.5-basis-point increase over the past 52 weeks and firmly breaching the critical 7.00% psychological and institutional ceiling.

This yield surge represents the domestic bond market pricing in a severe collision of macro headwinds:

• Imported Inflation and Rupee Vulnerability: With Brent crude over $111/bbl and the U.S. dollar supported by 5% Treasury yields, the Reserve Bank of India cannot tolerate a widening interest rate differential without exposing the rupee to speculative depreciation, which would further inflate imported energy bills.

• The Elimination of Repo Rate Cuts: Market consensus expectations for an autumn 2026 RBI easing cycle have evaporated. With headline WPI at 9.92%, CPI tracking toward 5.5%+, and food disinflation dead, any monetary easing would risk unanchoring inflation expectations.

• Supply Indigestion: Record sovereign and state development loan (SDL) borrowing schedules are meeting reduced appetite from commercial banks and institutional provident funds, requiring higher clearing yields to clear primary auctions.

A 7.07%+ G-Sec yield resets corporate borrowing costs across the economy, pushing commercial paper, AAA corporate debentures, and bank lending base rates higher, directly raising the cost of capital for Indian capex.

18. U.S. Treasury Curve: The 5% Anchor and the Resurgence of Term Premia

The U.S. Treasury curve has settled into an aggressive, multi-year high across all maturities as of 15 September 2026:

• 5-Year: 4.83%

• 10-Year: 5.00%

• 20-Year: 5.40%

• 30-Year: 5.36%

The defining characteristic of this curve is that the long end (20Y and 30Y) is trading 36 to 40 basis points above the 10-year note and over 50 basis points above the 5-year note. This is the structural return of the term premium:

• Investors are no longer merely anticipating near-term Federal Reserve policy rates; they are demanding substantial, permanent compensation for the risk of holding long-duration sovereign paper in an era of persistent supply-side commodity volatility and structural U.S. fiscal deficits exceeding 6% of GDP.

• The “pivot trade”—the speculative bet that central banks would rapidly slash rates back toward neutral at the first hint of economic deceleration—has been decisively broken.

As long as the world’s risk-free benchmark yields 5.00% at 10 years and 5.36% at 30 years, global dollar liquidity will remain deeply restrictive, tightening financial conditions across all emerging markets, penalizing external sovereign borrowers, and capping equity valuation multiples globally.

Conclusion: The Strategic Imperative for Allocators

The confluence of these 18 indicators signals that financial markets are moving from an era of cyclical rate volatility to an era of structural supply friction.

Investors, corporate treasurers, and portfolio allocators must recognize three operational realities:

1. The Cost-of-Capital Floor Has Shifted: With U.S. 10-year yields at 5.00% and India 10-year yields at 7.07%, the hurdle rate for capital investment has reset permanently higher. Capital-intensive businesses relying on cheap floating-rate debt will face ongoing margin erosion.

2. Gross Margin Contraction is Accelerating: The 510-basis-point gap between Indian wholesale (9.92%) and consumer (4.82%) inflation cannot persist indefinitely. Either corporate margins will suffer severe compression, or consumer inflation will re-accelerate toward 6%+. Companies lacking dominant pricing power will see earnings multiple compression.

3. Physical Supply Chains Supercede Financial Narrative: Whether in middle distillates, Capesize bulk freight, refined copper cathode, or natural gas storage, physical inventories are depleted. In a world of depleted buffers, financial interventions cannot manufacture physical supply. Allocators should maintain structural exposure to physical resource assets, upstream infrastructure, and pricing-power leaders while avoiding balance sheets exposed to unhedged energy and logistics liabilities.
   

   Subscribe now

Hello, and welcome to this week’s edition of Points & Figures, put together by Kulsum.

The idea is to tell stories about the Indian economy, financial markets and investing through data. Instead of throwing random charts at you, we dig into overlooked datasets to explain shifts and trends you may otherwise miss.

This edition draws heavily on Data For India’s research on judicial pendency. It shows how India’s court backlog has grown, where it sits and how long cases have remained unresolved. We use that work to follow a case from the moment an offence occurs to the years it may spend in court - and to ask what even a backlog of 5.4 crore cases leaves out.

Imagine someone steals your phone.

You report it to the police. If they register an FIR, investigate the theft, find a suspect and file a chargesheet, the case may reach court. Evidence is produced, witnesses are called and a judge decides whether the accused is guilty.

That is the neat version. A case can stall at every step.

The theft may never be reported. The police may not register it. The investigation may go nowhere. Later, a forensic report may be delayed, a witness may not appear or a hearing may be adjourned again and again.

That is why India’s crime data needs care. The National Crime Records Bureau, or NCRB, counts cases recorded by the police. It tells us how many crimes were registered, not how many actually occurred.

A government-appointed committee on crime statistics called the unrecorded part the “dark figure” of crime. People may stay silent because they fear the police, expect nothing to happen or prefer a private settlement. India still does not run a regular nationwide victimisation survey that could estimate what police records miss.

A state with more recorded crime may therefore be more dangerous, better at registering complaints, or both. A low number can mean safety. It can also mean that people have stopped expecting the system to help.

How the queue keeps growing

India’s judiciary has three broad levels. District and subordinate courts are where most civil disputes and almost all criminal trials begin. High Courts hear appeals, challenges to government action and some cases filed directly before them. The Supreme Court sits at the top.

The backlog of cases is spread across all three, but not evenly. India ended 2025 with roughly 5.4 crore pending cases. Nearly 4.8 crore of them were in district and subordinate courts. For most people, this is where the justice system begins and where it remains stuck.

The lower-court backlog has risen from around 2.6 crore cases in 2014 to nearly 4.8 crore in 2025. Pendency in the High Courts and the Supreme Court also increased by close to 50% over the decade, although from much smaller bases.

So why does the queue keep growing? The arithmetic is simple. Every year in which courts receive more cases than they dispose of adds another layer to the backlog.

Between 2018 and 2025, district and subordinate courts received more cases than they disposed of in most years. The pandemic made this imbalance much worse. Courts could move some work online, but trials that depended on witnesses, physical evidence and in-person appearances were much harder to conduct. Disposals recovered later, but they did not consistently overtake new filings.

This chart covers only district and subordinate courts. The India Justice Report 2025 separately says that, at the national level, a few High Courts have recorded annual case-clearance rates above 100% since 2017. That means they disposed of more cases than were filed during those years.

The two findings are not contradictory. They concern different levels of the judiciary. The High Court figure also does not mean that every High Court crossed 100%, or that decades-old cases were the ones being cleared. A court can keep pace with today’s filings and still spend years working through yesterday’s backlog.

Where the backlog sits

The national number hides huge differences between states. Pendency in Delhi and Haryana’s subordinate courts roughly tripled over the decade. Uttar Pradesh’s backlog doubled from about 55 lakh to 1.13 crore, while Maharashtra’s doubled to around 60 lakh. Gujarat was the only state to reduce its pending caseload.

Let’s not make the mistake of confusing this as a league table of good and bad courts. Large and growing states naturally generate more property disputes, traffic violations, commercial cases and criminal proceedings. More litigation can also mean that people are more willing and able to use courts.

The reverse matters too. An older Vidhi Centre for Legal Policy study notes that long delays can eventually discourage people from filing cases. Low litigation, like low recorded crime, is not automatically good news.

High Courts are just as uneven. Rajasthan’s pendency rose by around 200% over the decade, while Himachal Pradesh and Uttarakhand recorded increases of more than 150%. Tripura, Calcutta, and Jammu & Kashmir and Ladakh reduced their backlogs. Yet Allahabad still carried far more pending cases than the other High Courts shown below.

The percentages tell us where the backlog grew fastest. The absolute numbers tell us where the weight already was.

A backlog measured in decades

A growing queue is one problem. An ageing queue is worse.

As of March 2026, nearly one in three district and subordinate court cases had been pending for more than five years, and around one in ten for more than ten. The High Courts were worse where almost every second case was older than five years, and roughly one in four had crossed a decade.

By 16 July 2026, 11.73 lakh cases across the Supreme Court, High Courts and lower courts had been pending for more than 20 years. District and subordinate courts alone had 81,275 cases older than 30 years.

A case rarely loses 20 years at once. It loses them through adjournments, unserved notices, missing witnesses, transferred officers, vacant posts and evidence that has not arrived.

The age profile also changes sharply by state. More than half of Bihar’s pending lower-court cases were older than five years. In Tripura, the share was around 6%.

That comparison tells us where delay is concentrated but it does not tell us that judges in one state work harder than judges in another. States differ in caseload, staffing, infrastructure and the stage at which cases get stuck.

High Courts inherit part of their queue from lower courts and tribunals. A case can be tried, decided, appealed, sent back and heard again. They also hear writ petitions challenging government action on pensions, recruitment, land, taxes, detention and fundamental rights. One judgment may simply start the next proceeding.

By July 2026, the Allahabad High Court alone had more than 1.61 lakh cases pending for over 20 years. Madras had more than 76,000 and Bombay more than 55,000.

When waiting becomes punishment

What fills the queue also changes as a case moves up the system. Lower-court pendency is overwhelmingly criminal. Higher courts lean civil.

Around 77% of pending district and subordinate court cases were criminal as of March 2026. In the High Courts, 70% were civil; in the Supreme Court, the civil share was 79%. This follows the structure of the system. Criminal trials begin almost entirely in lower courts, while higher courts hear more appeals, civil disputes and constitutional cases.

The criminal backlog also has a consequence the chart cannot show, which is people waiting in jail.

At the end of 2023, India’s prisons had room for 4,39,119 people but held 5,30,333. Nearly 3.9 lakh were undertrials - people whose guilt had not yet been decided. They made up about 74% of the prison population, according to the Ministry of Home Affairs.

In one case decided in February 2025, an accused had spent five years in custody while prosecutors worked through a list of 100 witnesses. Only 42 had been examined, and the state could not say when the trial would finish. The Supreme Court granted bail, citing the accused’s fundamental right to a speedy trial.

That case is the danger hidden inside the 77%. For a civil litigant, delay can mean frozen property or unpaid money. For an accused who cannot secure bail, it can mean years in prison before the state proves its case.

The punishment may begin before the judgment.

How the queue can shrink

The backlog is produced at both ends. Some matters enter court even though they could have been resolved another way. The cases that do need a judge then move too slowly.

It is tempting to conclude that India simply has too many cases. But the number of filings is not a clean measure of excess litigation. More cases can mean that more people are able to report crimes and enforce their rights. Fewer cases can mean that disputes were settled early; they can also mean that people gave up. The aim cannot be a smaller queue at any cost.

There is still avoidable traffic. For example, the government issues tax demands, acquires land, recruits employees and runs benefit schemes, then frequently appeals when it loses. One study of admitted Supreme Court cases found that governments were the appellants in 73% of the sample. Their success rate was below 50% in some categories, including tax matters. Better decisions inside government, and stricter checks before an appeal is filed, could prevent some disputes from climbing the judicial ladder.

Minor offences need not always become full criminal cases either. In April 2026, Parliament passed the Jan Vishwas (Amendment of Provisions) Bill, 2026, which decriminalised 717 provisions across 79 central laws. The idea is to replace criminal consequences for many minor breaches with civil or administrative penalties. Virtual courts offer another route: 21 of them had handled more than 2.78 crore cases by March 2023. These measures can reduce future inflow, but they do little for the murder trials, bail applications, property suits and constitutional challenges already waiting.

Suitable civil and commercial disputes can also be resolved through mediation or Lok Adalats, sometimes before a suit is filed. But moving a dispute outside court is useful only when the process is fair. Serious crimes cannot be bargained away, and a worker, tenant or victim may not negotiate with a powerful counterparty on equal terms. Access to justice must not be reduced in the name of reducing pendency.

Then comes the second problem: giving necessary cases enough capacity to move. In early 2025, around 33% of High Court judge posts and 21% of subordinate-court judge posts were vacant. Vacancies were also substantial among stenographers, law clerks and other non-judicial staff. Nearly half of roughly 10,000 sanctioned posts in state forensic facilities were vacant in the data used by the India Justice Report. A case may sit on a court’s list even when the evidence is stuck in a laboratory.

This is why the backlog cannot be explained as simple laziness. Adjournments, judges’ absences and slow judgments do contribute. So do unserved notices, witnesses who do not appear, missing police records, delayed forensic reports and vacancies among clerks and prosecutors. Filling judge posts matters, but research cited by Data For India found only a weak relationship between judge strength and case-clearance rates. A judge can decide only the case that is ready to be heard.

There is no single reform that clears every blockage. India can keep minor regulatory offences out of criminal courts, settle suitable disputes earlier, stop departments from appealing weak cases, fill support and forensic vacancies, and finish the courtrooms and housing already being built. Each measure shortens a different part of the queue.

The visible queue

India does not have one queue of 5.4 crore cases. It has millions of queues running through police stations, government offices, forensic laboratories, courtrooms and prisons. A delay anywhere along that chain eventually appears under the same harmless-sounding label: “pending”.

Even the 5.4 crore figure counts only the cases that made it far enough to enter the system. The complaint never registered, the crime never reported and the person who gave up before approaching court remain outside the charts. A fall in pendency would mean little if it came from making that invisible group larger.

Real progress would look different: more people able to seek justice, fewer avoidable disputes entering a courtroom, and necessary cases reaching a decision sooner. India already has many of the tools. The work is to use them together without closing the courthouse door.

See you next edition :)

Hello, and welcome to this week’s edition of Points & Figures, put together by Kulsum & Mridula.

The idea is simple: to tell stories about the Indian economy, financial markets and investing through data. Instead of throwing random charts at you, we dig into overlooked datasets to explain shifts and trends you may otherwise miss.

This edition draws on research by AMFI and CRISIL Intelligence on India’s mutual fund industry. Their work looks beyond the headline growth in assets to show how the investor base is changing, why SIPs have become so important, where the money is being invested, and how far the mutual fund boom has spread beyond its traditional base.

Something fairly unusual has happened to mutual funds in India. They have gone from being one among many financial products to becoming part of the monthly routine for millions of investors.

Several things came together to make this possible. Digital onboarding made it easier to open and operate an account, the industry spent years advertising mutual funds and SIPs, and the strong equity market after the pandemic gave people a powerful reason to pay attention. SIPs then made the process almost automatic: instead of repeatedly deciding when to invest, people could set a fixed amount to leave their bank account every month.

The result is an industry that looks very different from five years ago. But the headline growth hides several stories underneath it.

The rich mutual fund machine

India’s mutual funds managed ₹31.43 lakh crore in March 2021. By March 2026, that had grown to ₹73.73 lakh crore, an annual growth rate of 18.6%.

Some of this came from investors adding money, while some came from the assets inside the funds becoming more valuable. Equity enjoyed a strong run during much of this period, and equity schemes’ share of industry AUM rose from 31.2% to 43.4%.

Mutual funds are taking up more space in the economy

In FY21, mutual fund AUM was equal to 15.9% of India’s GDP. By FY26, it had reached 21.3%.

Another number helps explain this change. Mutual funds accounted for 3% of the gross flow of annual household savings in 2020. By 2025, their share had risen to 13%.

This does not mean Indian households abandoned deposits, insurance, property or gold. It means that, within the money households saved during the year, mutual funds claimed a larger share than before.

India remains far below countries such as the US on the AUM-to-GDP measure. But countries have very different pension systems, income levels and financial markets, so the comparison is useful as context, not as a target India should necessarily reach.

Five years, nearly four crore new investors

A rising market can make an industry larger without bringing in anyone new. But that is not the full story here.

India’s unique mutual fund investor base grew from 2.3 crore in March 2021 to 6.1 crore in March 2026. The increase suggests that industry growth was supported by new participants as well as larger portfolios and market appreciation.

Awareness played a role. SEBI’s 2025 survey found that 53% of Indian households were aware of mutual funds and ETFs, making them the most widely recognised securities-market product.

Still, the 6.1 crore figure includes retail investors, HNIs and HUFs. It counts investors, not households, and cannot be directly compared with the survey’s household-awareness percentage. Awareness is clearly broader than participation, but these numbers cannot tell us the exact size of that gap.

One in every five rupees is now SIP money

SIP AUM rose from ₹4.25 lakh crore in March 2021 to ₹14.83 lakh crore in March 2026. Its share of industry AUM increased from 13.5% to 20.1%.

The longer history is even more striking. Monthly SIP contributions rose from ₹4,335 crore in March 2017 to ₹32,087 crore in March 2026. Around 76% of the cumulative gross SIP inflows recorded during these nine years arrived in the most recent five years.

But SIP AUM is not the amount contributed every month. It is the accumulated value of investments made through SIPs, including the effect of market gains and losses.

A growing share of SIP money is staying beyond five years

The SIP pool is not just larger; its age profile has changed.

The share of SIP AUM held for more than five years rose from 12.3% in March 2021 to 31% in March 2026. The share less than one year old fell from 37% to 21.1%.

This is important because investor enthusiasm is easier to sustain when markets are rising. By FY26, equity returns had become less exciting than during the earlier part of the period, but monthly SIP contributions remained high and a larger share of SIP assets sat in older holdings.

However, this is not an SIP-retention rate. It does not tell us whether every instalment continued uninterrupted. It only shows how SIP AUM was distributed across holding-period buckets.

For Indian investors, SIP mostly means equity

An SIP is a way of investing, not an asset class. But roughly 87% of SIP AUM was held in equity-oriented schemes in March 2026, ₹12.85 lakh crore out of ₹14.83 lakh crore.

This steady flow of domestic money has also changed the market’s ownership. Domestic mutual funds’ share of the total market capitalisation of NSE-listed companies rose from 7.2% in March 2021 to 11.4% in March 2026. In FY26, mutual funds invested ₹5.35 lakh crore in equities while foreign institutional investors were net sellers.

This does not mean SIPs alone supported the market: the ₹5.35 lakh crore includes equity purchases by mutual funds from different sources, not only SIPs. But the expanding SIP pool contributes to a larger, more regular domestic source of equity capital.

SIP AUM represented 55% of small-cap fund AUM in March 2026. It represented 47% of mid-cap AUM and around 40% of large-cap and large-and-mid-cap AUM.

Small-caps also went from 8.3% of total SIP AUM in March 2021 to 12.1% in March 2026. Mid-caps became the largest SIP category, accounting for 13% of SIP AUM.

This fits the strong interest in smaller companies during much of the period. But it is important to read the chart correctly: 55% of small-cap AUM being linked to SIPs does not mean 55% of all SIP money went into small-caps.

SIPs spread purchases over time. They do not remove the higher volatility associated with smaller companies.

The passive boom has several stories inside it

Passive fund AUM reached ₹13.73 lakh crore in March 2026, around four times its March 2021 level. Its share of industry AUM rose from 9.8% to 18.6%.

Index funds became more important, particularly for retail investors. Their share of retail passive AUM increased from 11.5% to 55%.

But the overall passive industry is not purely a retail story. Corporates held 69.6% of passive AUM, largely through ETFs. FY26 growth was also helped by inflows and price gains in gold- and silver-linked products. Therefore, the rise in passive AUM combines corporate ETF holdings, retail index investing and greater demand for commodities.

The regulatory environment is also becoming friendlier to passive-only fund houses. SEBI introduced its lighter MF Lite framework for passively managed schemes in December 2024. The framework is intended to reduce entry barriers and compliance requirements for passive products.

Direct investing is rising, but regular plans still lead

The direct share of industry AUM increased only slightly, from 43.4% to 45.1%. But among retail investors, it rose from 21.4% to 36.7%.

Direct and regular plans hold the same underlying portfolio, but direct plans exclude distribution commissions and therefore have lower expenses. Even with that cost advantage, nearly two-thirds of retail AUM remained in regular plans.

Access is only one part of investing. Some investors still rely on distributors to select funds, complete transactions or navigate market volatility. But the data does not show what service each investor received, so regular-plan AUM should not automatically be described as “advised money”.

The boom is spreading, but unevenly

AMFI divides locations into T30 and B30. T30 comprises the 30 locations with the highest mutual fund AUM. B30 covers all locations beyond those 30, based on AMFI’s location and PIN-code mapping.

B30 should therefore not be treated as a synonym for rural India. It includes every location outside the top 30, including sizable cities.

B30 AUM grew from ₹5.2 lakh crore in March 2021 to ₹13.9 lakh crore in March 2026. T30 AUM rose from ₹26.2 lakh crore to ₹59.8 lakh crore. B30 grew faster, but its share of industry AUM moved only from approximately 16.6% to 18.9%. More than four-fifths of assets remained associated with T30 locations.

Women’s AUM also rose from ₹5.84 lakh crore to ₹15.88 lakh crore. Yet women represented only 26.4% of individual investors in March 2026. The report separately shows that the average folio sizes of women and men were similar by then, suggesting the larger gap was in participation rather than average folio size.

Mutual funds have widened their reach, but income, awareness, access and distribution remain uneven. The industry has travelled further into India. It simply has not travelled everywhere equally.

Subscribe now

Hello, and welcome to this week’s edition of Points & Figures, put together by Kulsum & Mridula.

The idea is simple: to tell stories about the Indian economy, financial markets and investing through data. Instead of throwing random charts at you, we dig into overlooked datasets to explain shifts and trends you may otherwise miss.

This edition draws on research by CareEdge on India’s microfinance industry. Their work looks at how the sector has emerged from a difficult period, what drove the recovery, and which parts of the industry came out stronger.

P.S: Big thanks to our colleague Kashish for helping us put together this edition :)

Microfinance loans are, by definition, small, collateral-free loans given to low-income households. Under the RBI’s current rules, any collateral-free loan to a household earning up to ₹3 lakh a year counts as microfinance. The loans may be micro. The problem the industry created was anything but.

After Covid, demand for credit came roaring back. Households needed money, and lenders were eager to supply it. Then, in 2022, the RBI rewrote the rules of the industry. Earlier, NBFC-MFIs operated under fairly prescriptive limits on who they could lend to, how large those loans could be and even how they could price them. The new framework widened eligibility to all households earning up to ₹3 lakh a year, removed the old loan-size and pricing caps, and instead asked lenders to set their own interest rates under board-approved policies. The idea was to let lenders serve more low-income borrowers while keeping total household repayments within 50% of monthly income.

That was broadly a good thing. Microfinance has never lacked demand; the harder problem has been getting enough formal credit to the people who need it. The new rules made that supply easier to expand. And expand it did.

Perhaps a little too well. In the rush to grow, underwriting started slipping. Borrowers became increasingly stretched, repayments weakened, credit costs shot up and lenders began writing off loans at an uncomfortable pace. By FY25, the industry was firmly in trouble.

Two years later, microfinance finally appears to be recovering. But this is not a simple bounce back to normal. The industry had to shrink, clean up old loans and rethink how it lends. And the recovery has not reached every lender equally.

The charts that follow tell that story: what went wrong, what changed, and who actually made it out of the woods.

(Sidenote: We also spoke to a Tamal Bandyopadhyay veteran banking journalist about the microfinance recovery. If this story interests you and you want to go deeper, you can watch that interview here.

Subtext by Zerodha

Tamal Bandyopadhyay on microfinance industry cycle

Hi, I’m Kashish and a while ago, I did something slightly unhinged. I went through 144 concalls, across 18 microfinance companies, spanning eight quarters, just to watch a single credit cycle play out in slow motion. The idea was to see, in the actual words of the people running these businesses, how an entire industry lends its way into a crisis and th…

Read more

4 months ago · 22 likes · Zerodha

Closing one of microfinance’s biggest loopholes

To understand the recovery, it helps to start with one of the problems that made the downturn worse in the first place: borrowers taking loans from too many lenders.

One of the oldest problems in microfinance is evergreening. A borrower takes a loan from one lender to repay another, then keeps the cycle going. On paper, individual loans may still look current. Underneath, the borrower may already be under stress.

That is why headline NPA numbers do not always tell you enough. A useful early warning sign is how many lenders a borrower is dealing with, and how much total debt she is carrying.

Until recently, there were few hard guardrails around this. There was no industry-wide cap on total microfinance indebtedness and no limit on how many microfinance lenders could lend to the same borrower.

The industry is now trying to close that gap. Microfinance indebtedness is capped at ₹2 lakh, unsecured retail loans are included while assessing leverage, and a borrower can no longer take loans from more than three microfinance lenders.

The effect is visible. In March 2024, 13% of borrowers had loans from four or more lenders. By June 2026, that number had fallen to 4%. Meanwhile, the share dealing with two or fewer lenders rose to 81%.

The data in microfinance is still far from perfect. But this is a useful way to see whether borrowers are genuinely becoming less stretched, rather than simply looking healthy in reported NPA numbers.

Fewer borrowers, but larger loans

Those guardrails did more than reduce multiple borrowing. They also made lenders much more selective about who got a loan in the first place.

The tightening was sharp enough that the number of borrowers in the industry actually fell. Unique microfinance borrowers declined from 7.8 crore in March 2024 to 6.4 crore in March 2026. At the same time, the average disbursement increased from ₹48,300 to ₹61,500.

Some of this is healthy. Borrowers with several existing loans, overdue payments or weak repayment capacity are being filtered out. But it also means AUM alone can give an incomplete picture. A lender can rebuild its loan book by giving larger loans to a smaller group of safer customers.

So yes, the microfinance balance sheet may be getting healthier. But that does not automatically mean credit is reaching more people. For now, the opposite is true.

A lot of the old problem had to be thrown away

But tightening new lending only solves half the problem. Lenders still had quarters of stressed loans sitting on their books.

And a large part of that stress did not disappear because borrowers suddenly began repaying.

Across 16 CareEdge-rated MFIs, write-offs and sales to asset reconstruction companies during FY25 and FY26 were equivalent to more than 17% of their March 2024 loan book. Write-offs alone accounted for 13.1%.

This was necessary balance-sheet surgery. Recognise the loss, remove the worst assets and give the surviving portfolio room to recover.

But it also puts the recovery in perspective. Improving delinquency numbers cannot be read as a complete return to normal if a large chunk of the old problem has simply been taken off the books. The stronger evidence, then, has to come from what happens to loans made after the cleanup.

The loan book has finally stopped shrinking

Put tighter new lending and the cleanup of old loans together, and the result was inevitable: the industry became much smaller.

Lenders were making far fewer new loans, closing out older ones and removing stressed accounts. Microfinance AUM fell from ₹4.34 lakh crore in March 2024 to ₹3.15 lakh crore by December 2025.

That decline was one of the clearest signs that something had gone wrong. But a falling loan book by itself does not tell you the industry is getting healthier. It may simply mean lenders have stopped growing because they are still dealing with the damage.

The more important signal came when that trend finally reversed.

In March 2026, AUM grew 3.3% over the previous quarter—the first sequential increase after seven quarters of decline. It edged up again in June.

That is why several MFI management teams started saying the worst was behind them. By then, much of the older stress had been recognised or cleaned up. So incremental lending should, at least in theory, be more deliberate than what came before.

The recovery still looks modest compared with the growth before the crisis. But after nearly two years of shrinking, even some growth is a relief. More importantly, the industry does not need another sprint. It needs to prove that it can grow without recreating the same problem.

The lending tap is open again, just not fully

And once the loan book stopped shrinking, the next question was obvious: are lenders actually willing to lend again? When a lending cycle turns bad, the easiest way to avoid new defaults is to stop making new loans. That is exactly what happened.

Quarterly microfinance disbursements fell from approximately ₹1.08 lakh crore in Q4 FY24 to ₹57,000 crore in Q1 FY26. By Q4 FY26, they had recovered to nearly ₹78,000 crore—the highest level in eight quarters—before falling back to around ₹61,000 crore in Q1 FY27.

So the lending tap is open again. Just not all the way.

This is not a clean V-shaped recovery. Lenders are willing to grow again, but concerns around weather, rural incomes and fresh borrower quality are keeping them cautious.

That is probably sensible. The previous crisis was created partly because the industry confused rapid disbursement growth with progress. This time, every new rupee lent has to be weighed against the risk of returning to the same old habits.

The new loans are behaving better

More lending, though, only counts as a recovery if the new loans do not start going bad again. Write-offs can clean up old defaults. They cannot hide borrowers who have only just started missing payments. That makes the early overdue buckets particularly useful.

PAR 1–30 fell from 1.4% in March 2025 to 0.5% in June 2026. PAR 31–90 declined from 2.8% to 0.7%, while loans overdue by 91–180 days fell from 3.8% to 0.8%.

The improvement across all three buckets suggests that fresh loans are genuinely performing better. Fewer borrowers are slipping into default, and fewer overdue accounts are progressing into more serious stress.

This is probably the strongest evidence that the recovery is not only cosmetic. The industry has cleared a lot of the old damage and, at least for now, appears to be creating less new damage.

But there is one caveat. The industry is still lending far less aggressively than it did before the crisis. Credit discipline is easier to maintain while growth is subdued. The real test comes when lenders start chasing growth again.

Profitability is returning because bad loans cost less

That improvement in loan quality is now showing up exactly where you would expect it to: profits. The improvement in MFI profitability is not being driven by wider margins or dramatically lower operating expenses. It comes largely from one place: bad loans are becoming less expensive.

Credit costs rose from 2.5% of average total assets in FY24 to 8.7% in FY25. That was enough to push return on assets to –1.2%. By FY26, credit costs had fallen to 5.4% and return on assets recovered to 0.5%.

CareEdge expects Return on Average Total Assets (RoTA) to reach 2.4% in FY27 as credit costs fall further.

But this also shows how fragile the earnings recovery is. If lenders loosen standards, delinquencies rise and credit costs shoot up again, profits can disappear remarkably quickly.

For microfinance lenders, asset quality is not one part of the earnings story. Right now, it is the earnings story.

So, at an industry level, the recovery is becoming fairly convincing. But industry averages hide another part of the story: not every lender went through the downturn in the same way.

Banks stepped back faster. MFIs stayed in the game

When things got messy, everyone pulled back. Banks reduced microfinance lending, and NBFC-MFIs did too. The difference was that banks could afford to do it much more aggressively.

Microfinance is only one part of a bank’s loan book. If risk rises, it can simply send that money somewhere else. An NBFC-MFI has far less room to manoeuvre. Its branches, field officers and collection networks are built around this business.

So MFIs stepped back, just not as sharply. The result is that their market share actually increased.

Banks accounted for 33% of microfinance AUM in March 2024. By June 2026, that had fallen to 25%. NBFC-MFIs moved in the opposite direction, from 39% to 44%.

That does not necessarily mean MFIs have permanently won this market. Banks can ramp up again just as quickly if they decide the risk-reward has improved.

The more interesting question is whether they will.

If they do not, this shift in market share may end up looking less like a temporary crisis response and more like the industry’s new normal.

Large MFIs have more escape routes

And the divide is not just between banks and MFIs. Even within microfinance, the largest lenders have much more room to escape when the cycle turns bad.

The RBI has lowered the share of assets that NBFC-MFIs must hold in qualifying microfinance loans from 75% to 60%. In theory, that gives every MFI more room to diversify. In practice, the advantage belongs overwhelmingly to the largest institutions.

Non-microfinance products already account for 19% of large MFIs’ AUM, compared with 13% for medium MFIs and just 2% for small ones.

Diversification requires more than regulatory permission. A lender needs capital, technology, distribution and the ability to assess unfamiliar borrowers. Large MFIs can use their existing networks to move into adjacent retail loans. Smaller institutions remain almost completely dependent on traditional microfinance.

So when the microfinance cycle turns bad, large lenders have somewhere else to go. Small lenders are stuck inside the cycle.

And even if a smaller MFI wants to grow again now, there is another problem: it first needs someone willing to fund it.

And then there is the 98% problem

That is where the recovery starts looking decidedly two-speed.

On paper, funding conditions are improving. NBFC-MFIs raised nearly ₹77,600 crore in FY26, up from about ₹58,200 crore in FY25.

But almost all of that money went to one corner of the industry. Large MFIs accounted for 98% of incremental borrowing.

Large institutions raised fresh debt equal to 60% of their opening AUM. Medium MFIs managed 26%. Small MFIs raised only 14%.

This matters because MFIs depend on borrowed money to issue fresh loans. A small lender may have willing borrowers, good collections and deep local knowledge, but none of that produces growth if banks are unwilling to fund it.

The winners of the next microfinance cycle may therefore be decided before the loans are even made—by who can raise the money in the first place.

So, is microfinance really out of the woods?

Mostly, yes. Borrowers are dealing with fewer lenders, fresh delinquencies are falling, disbursements have restarted and profits are returning. The industry looks healthier than it did a year ago.

But it has become healthier partly by lending to fewer people, writing off a large part of the old loan book and concentrating fresh funding among the biggest institutions.

That creates a slightly uncomfortable conclusion. The microfinance sector may be recovering, while some smaller lenders—and some borrowers who depended on them are still being left behind.

Subscribe now

Hello, and welcome to this week’s edition of Points & Figures, put together by Kulsum.

The idea is simple: to tell stories about the Indian economy, financial markets and investing through data. Instead of sharing random charts without context, we delve into overlooked datasets to explain shifts and trends you may not otherwise encounter.

To most people, the capital markets are simply stocks. But underneath that familiar surface, the market is changing. Small towns are rapidly taking to mutual funds. SIPs carry more household savings each month. The wealthy are putting more money into PMS and AIFs. Office buildings and highways are becoming tradeable through REITs and InvITs.

There isn’t a single, clean wave of investors rushing into every asset class. In some places, the market is genuinely reaching new participants. In others, the same investor base is committing more money. And there are corners where the capital markets plumbing is being built in the hope that participants will arrive.

These shifts have left their footprint in the data, much of which you can find in SEBI’s latest annual report.

The investor map is changing

Going by mutual fund data, more of India is investing than ever before. India added around 72 lakh unique mutual-fund investors in FY26, taking the total from 5.42 crore to 6.14 crore.

Most of this shift came from outside India’s largest cities. The number of investors in Tier-II cities jumped 37.6%, from 1.03 crore to 1.42 crore over the year. That is now slightly more than the 1.34 crore investors in Tier-I cities. Tier-III locations still account for the largest number overall, at 3.39 crore.

Why? We think digital onboarding and payments have made it possible for one to start investing without going through a bank branch, broker or financial adviser. This ease of access has allowed mutual funds to break out of metro cities.

SIPs got bigger without adding many accounts

The story of SIPs is a little different.

Contributing SIP accounts increased from 10.05 crore to 10.45 crore — growth of only around 4%. Net SIP contributions, however, jumped 25.8%, from ₹1.57 lakh crore to ₹1.97 lakh crore. A net of ₹16,413 crore was added to SIPs every month, on average.

That is, even if the number of people making SIPs hasn’t picked up, among those that do, the habit has grown deeper. This is good for the markets. SIPs are a reliable bridge between household savings and the markets, bringing in money every month, even when things are volatile.

PMS is larger than it first appears

Portfolio Management Services grew in FY26, increasing their clientele by 8.2%, from 1.99 lakh to 2.16 lakh.

Their total assets under management, meanwhile, rose 9.6%, from ₹37.8 lakh crore to ₹41.4 lakh crore. That headline number, however, comes from two very different groups: large retirement pools and a much smaller base of private clients. Around ₹30.1 lakh crore was managed for EPFO and other provident funds. Outside that pool, discretionary PMS assets grew at a dramatic 16.5%, from ₹4.29 lakh crore to ₹5 lakh crore. That is, the money in individual managed accounts has gone up rapidly.

PMS is a high-ticket product. It’s nowhere close to mutual funds in reach. But it’s becoming a large part of the market in its own right.

Private markets are no longer a small corner

If you want to understand India’s private markets, look at its Alternative Investment Funds, or AIFs. These are privately pooled investment vehicles, which come in various “categories”.

Category I includes venture-capital, SME, infrastructure and social-impact funds. Category II includes private-equity and debt funds that generally do not use leverage except for day-to-day needs. Category III can use more complex or leveraged strategies. Category II was the largest part of the ecosystem in FY26.

Across 1,829 registered AIFs, commitments have risen from ₹6.4 lakh crore in FY22 to ₹16.9 lakh crore in FY26. This isn’t the cash these funds have already collected — it’s money investors have committed to give when called on to do so. Of this, ₹7 lakh crore has actually been raised.

Even so, the market is now large enough to play a serious role in financing businesses outside public exchanges.

Buildings and highways are becoming financial products

For most Indian households, investing in real estate usually means something like buying a house or a plot. The markets, however, have opened up a new option — REITs and InvITs break much larger income-generating assets into tradeable units, making it possible to invest without owning or managing a property or project directly.

REITs largely hold commercial properties such as offices. InvITs hold infrastructure such as roads, power networks and renewable-energy assets. REIT AUM increased from ₹1.63 lakh crore to ₹2.38 lakh crore in FY26. SM REIT (Small and Medium Real Estate Investment Trusts) AUM, more than doubled from ₹359 crore to ₹853 crore, although the segment remains tiny.

REIT fundraising nearly doubled to ₹9,300 crore. In part, this follows a widening of these products. REITs were reclassified as equity for mutual-fund investment from January 2026. Meanwhile, the public-issue process for SM REITs was simplified. That could bring more institutional money into the space while giving smaller real-estate projects a route to market.

InvIT fundraising, however, lagged slightly, falling 21.3% to ₹21,025 crore.

Smaller companies are finding their way to the exchange

It isn’t just investors who are entering new parts of the market. The range of companies raising money is widening too.

India’s SME platforms hosted 257 IPOs in FY26, up from 70 in FY22. Collectively, these companies raised ₹11,587 crore, 18.1% more than the previous year. Their average issue size also increased from ₹40.7 crore to ₹45.1 crore.

SME IPOs give smaller businesses access to public equity without requiring the size of a mainboard company. They also tend to be more focused on raising money for the business. SEBI’s analysis found that the offer-for-sale share in SME IPOs generally remained between 0% and 27%, while mainboard IPOs regularly had OFS shares above 60% from 2016 onwards. In other words, SME IPOs are more likely to bring fresh money into the company instead of mainly allowing existing owners to sell.

Cities are discovering bonds

For years, municipal bonds seemed like a failed product with no takers.

But FY26 changed the picture. Municipal corporations raised ₹1,756 crore through 14 issues. One year ago, the entire market consisted of one ₹100 crore issue.

Cities must pay for water, transport, sanitation and other infrastructure. For this, they usually depend on their own revenues, state and central government grants, budgetary support and loans. Municipal bonds add another source of long-term funding. They do not replace government budgets, but they can help finance specific projects, in exchange for better accounts, credit ratings and clearer disclosure of how the money will be used.

The surge wasn’t automatic. SEBI conducted outreach programmes with urban local bodies and expanded the rating framework for project-backed municipal bonds. The next test is repetition: will cities return to the market regularly, or will FY26 remain one unusually strong year.

Even donations are being given market infrastructure

The Social Stock Exchange is the smallest market in this edition, but perhaps the most unusual. These aren’t an investment product in the usual sense. It is closer to philanthropy placed inside a regulated market structure. Here, non-profit organisations use instruments known as zero-coupon, zero-principal securities. Investors provide money for a social programme to these entities, without receiving interest or even their original capital.

In FY26, five organisations raised ₹21.9 crore, compared with ₹14.8 crore raised through six issuances in FY25. For now, it remains an experiment rather than a major funding market. But it shows how far the idea of a securities market can stretch.

Corporate bonds are moving onto more organised rails

Corporate bonds don’t all trade like shares, on an exchange screen. A large part of the market is negotiated between institutions over the counter and then settled through clearing corporations. Some bonds are also transferred off-market through the depositories. When clearing corporation settlements rise, that could reflect both more trading and more activity moving onto formal settlement rails.

In FY26, the value of corporate bonds settled through clearing corporations increased 27.9% to ₹21.2 lakh crore. The number of listed corporate-bond trades settled through them rose much faster, from 11 lakh to 26.8 lakh.

The direction of policy, recently, has been towards opening the market — from a reduced face value for privately placed debt, to exchange-based trading.

Has this worked? Because the number of trades grew much faster than their value, the average transaction became smaller. There clearly have been more frequent, smaller trades. Does this prove retail participation, however? It’s unclear. The report does not identify who made them, and private placements still accounted for 98.8% of corporate-bond fundraising.

The ‘Request for Quote’ platform is part of the same shift. Instead of completing the entire trade through a private conversation, institutions can request and compare quotes electronically across corporate bonds, commercial paper, government securities and other debt instruments. Such a common electronic platform can improve transparency and price discovery in a market that has long been dominated by bilateral deals.

The use of this platform has grown sharply since its launch in 2020, but this was not entirely organic. Mutual funds, AIFs, portfolio managers, FPIs and brokers are required to execute a specified share of their bond trades through RFQ. This growth, therefore, reflects a regulatory push to put rails in place.

A wider market, but not in one single way

Two different changes emerge from all these charts.

Some pipes are carrying more money, from SIP contributions, to PMS assets. AIFs show a separate shift: private-market capital is becoming larger and more organised. At the same time, new pipes are being opened. Smaller companies are listing, municipalities are issuing bonds, non-profits are raising money on exchanges, and physical assets are being converted into tradeable units.

Not every product has become mainstream. But this is an early sign that India’s capital markets can now accommodate more kinds of money, assets and issuers than they could earlier.

Subscribe now

Hello, welcome to this week’s edition of Points & Figures. This edition is done by Kulsum & Mridula.

The idea behind this newsletter is simple: to tell stories about the Indian economy, financial markets, and investing through data visualisations. The goal isn’t to send you a bunch of random, out-of-context charts, because pretty much everybody does that already.

Instead, we take large datasets that are often overlooked or under-appreciated and use them to tell stories about trends, shifts, and developments in the Indian economy and markets that you might not otherwise come across. And like everything else we do under Markets by Zerodha, the quality bar remains extremely high.

In the previous edition, we looked at India’s registered factories as one national system. That gave us the broad picture. Factory employment has grown, contract labour has become more common, and the sector is becoming more capital heavy. We also saw that different industries create very different amounts of jobs and value.

But a factory does not exist in isolation. A car plant needs component makers, tool rooms, testing facilities, warehouses and roads. A refinery needs ports, pipelines, storage and chemical plants that can use what it produces. Once these networks form, new factories often prefer to locate near them because the suppliers, skills and infrastructure are already available.

Over time, this creates clusters. Chennai, Sriperumbudur and Oragadam became an automobile and electronics belt. Pune and Chakan built an automobile and engineering ecosystem. Gurugram, Manesar and Bawal developed around vehicles and components. Gujarat’s ports, refineries and chemical estates created an entirely different industrial geography.

The result is that India’s factory economy is not spread evenly across the country. A handful of states account for a large share of its factories, workers, output and value creation. Even among these states, manufacturing looks very different.

So what happens when we stop looking at India as one factory economy and look at the states instead?

India’s manufacturing map changes depending on what we count

Tamil Nadu is the clearest example of how one state can dominate through breadth. It accounts for about 15% of India’s registered factories and 15% of factory employment, the highest share on both measures.

Gujarat follows with about 13% of factories and employment, while Maharashtra has around 10% of factories but 13% of jobs. Together, these three states account for nearly four in ten registered factories and factory jobs among the states shown here.

But the ranking changes when we move from factories and jobs to money.

Gujarat produces the largest share of India’s factory output, at 17%. Maharashtra follows at 14%, while Tamil Nadu accounts for 10%.

Once the cost of raw materials, fuel and other inputs is removed, Maharashtra moves into first place. It generates 16% of India’s factory GVA, compared with Gujarat’s 14%. In simple terms, Gujarat’s factories process more gross value, while Maharashtra’s factories collectively create more value after input costs are deducted.

This gives us three different kinds of manufacturing leaders. Tamil Nadu leads through factories and jobs. Gujarat leads through output. Maharashtra leads through value creation.

It also shows how concentrated the factory economy is. The top five states account for more than half of India’s factory employment and GVA. Indian manufacturing is spread across the country, but much of its economic weight is concentrated in a relatively small group of states.

This concentration is not simply the result of large populations. Factories tend to reinforce the advantages of places where industrial networks already exist. An anchor manufacturer attracts suppliers. Suppliers make the location more attractive to the next manufacturer. Workers acquire industry-specific skills, logistics improve and specialised services emerge. Research on Indian manufacturing has found evidence of these agglomeration benefits, the economic advantages firms gain from locating near other firms.

That is why the same states remain prominent across several measures, even though their exact rankings change. They have not just accumulated factories; they have accumulated industrial ecosystems.

Why do the same states look so different depending on the measure? A large part of the answer lies in what their factories make.

States do not just make different amounts, they make different things

Food products appear among the three largest industries in almost every state shown here. That makes sense. Food processing can develop close to farms and consumer markets, and activities such as rice milling, dairy, edible oils and packaged foods are spread widely across India.

Beyond food, however, the state factory systems separate into distinct industrial belts.

Tamil Nadu is not only an automobile state. Vehicle factories around Chennai, Sriperumbudur and Oragadam sit alongside component makers, electronics plants and machinery businesses. Farther west, Coimbatore has long supported engineering and textile machinery, while Tiruppur is a major knitwear cluster. This mix helps explain how Tamil Nadu can have both a large factory count and the country’s largest factory workforce.

Maharashtra has another layered system. Pune–Chakan is a major automobile and engineering belt, but the state also has chemicals, pharmaceuticals, food processing and older industrial centres around Mumbai, Thane, Nashik and Aurangabad. Its strength is not one single industry but several large value chains operating together.

Gujarat’s geography is more closely tied to heavy, continuous-process industries. Its ports support the movement of crude oil, chemicals and bulk materials. Refineries and petrochemical complexes feed chemical, plastics and other downstream factories. The Dahej petroleum and chemicals investment region, industrial estates such as Ankleshwar, and the state’s gas and port infrastructure reinforce this model. Coke and refined petroleum alone account for 24% of its factory output.

Haryana’s manufacturing is shaped heavily by the Gurugram–Manesar–Bawal automobile corridor. Large vehicle makers encouraged component suppliers, logistics firms, testing facilities and skilled workers to locate nearby. Motor vehicles consequently account for about a quarter of the state’s factory output.

Telangana’s profile is led by pharmaceuticals, while food products dominate the output mix of Uttar Pradesh, Andhra Pradesh and Madhya Pradesh. Rajasthan has a stronger presence of basic metals, food and non-metallic minerals.

This chart therefore explains more than what each state produces. It shows the industrial ecosystems behind the remaining numbers.

A refinery can process enormous values using costly equipment and relatively few people. An automobile cluster supports large assembly plants as well as hundreds of component and engineering businesses. Food processing can spread across many smaller factories. Pharmaceuticals may create high value through formulations and technical knowledge without requiring the heaviest physical assets.

The industry mix decides what kind of jobs are created, how much capital factories need, how large plants become and how much of their output remains as value added. The next charts are consequences of these different industrial structures.

Tamil Nadu creates jobs; Gujarat creates output

The contrast becomes clearer when we compare each state’s share of factory employment with its share of output.

Tamil Nadu accounts for 15% of factory employment but only 10% of output. Its factories therefore absorb a much larger share of India’s manufacturing workforce than their share of gross production value.

Gujarat sits at the other end. It accounts for 13% of employment but 17% of output. Its factories process much more output with a relatively smaller share of workers.

Maharashtra is more balanced, with 13% of employment and 14% of output. Karnataka also has matching shares of about 6%, while Uttar Pradesh has a slightly larger employment share than output share.

This gap is not just a statistical curiosity. It tells us what manufacturing is doing for the state economy.

Tamil Nadu’s automobile, component, electronics, textile and machinery ecosystem creates work at several levels, from assembly plants to supplier factories and more labour-intensive production. Its factory system has become a major employment engine.

Gujarat’s petroleum, chemicals and metals industries work differently. They use huge plants to process very high-value inputs. This lifts output, but the same rupee value does not require a proportionate number of workers.

For a country trying to create jobs, Tamil Nadu’s model is especially valuable. But an employment-heavy system also faces a harder second task: workers’ wages and skills need to rise as factories become more productive. Otherwise, a state can employ many people without moving into higher-value production.

Gujarat faces the reverse challenge. Its capital-heavy industries are critical for energy, materials, exports and supplies used by other factories, but they cannot absorb labour on the same scale. Its growth therefore needs to create jobs through downstream chemicals, plastics, engineering and other businesses around the large plants.

The more useful question is not which model is better. It is whether each state is building the wider value chain around its existing strengths.

The next chart shows what is physically sitting behind these workers.

Some state factory systems run far more heavily on machines

Across the states shown, the average amount of fixed capital behind each factory employee varies almost fourfold.

Madhya Pradesh leads with roughly ₹50 lakh of fixed capital per person engaged. Gujarat follows at ₹35 lakh, and Andhra Pradesh at ₹32 lakh. At the other end, Telangana and Tamil Nadu have about ₹13 lakh of fixed capital per person.

This puts the jobs-versus-output chart into perspective. Tamil Nadu’s factory economy employs a large number of people with comparatively less capital behind each employee. Gujarat combines a similar employment share with nearly three times as much fixed capital per person, helping it generate a much larger share of output.

Madhya Pradesh stands out even more. Food products are its largest industry, but coke and petroleum are also among its top three. Large plants in capital-heavy industries can push up the amount of machinery and productive assets behind the average worker.

This is the deeper state-level divide: some manufacturing systems expand by bringing more workers into factories, while others expand through larger plants and more capital-intensive production.

Capital-heavy anchor plants can still create employment outside their own gates. A refinery supports transport, maintenance, storage and downstream chemical businesses. An automobile plant supports component suppliers and tool makers. The ASI employment number counts jobs inside registered factories; it does not capture every indirect job created across the surrounding cluster.

This is why the employment-versus-output chart and the capital chart need to be read as two layers of the same system. Direct factory employment tells us how labour-intensive production is. The surrounding supplier network tells us whether a capital-heavy cluster spreads its benefits more widely.

But the real question is not simply which state has more machinery. It is whether that machinery helps create more value.

More capital often comes with more value, but not in a neat line

Madhya Pradesh again ranks first when we look at GVA per person engaged. Its factories create around ₹17 lakh of value per employee. Maharashtra and Karnataka follow at about ₹15 lakh, while Gujarat and Rajasthan generate roughly ₹14 lakh.

At the broadest level, capital and value creation do appear connected. Madhya Pradesh has the most fixed capital per person and the highest GVA per person. Gujarat is also high on both.

But the relationship is not mechanical. Maharashtra and Karnataka create more GVA per person than Gujarat despite having much less fixed capital behind each employee. Andhra Pradesh has the third-highest capital per person, but its GVA per person is closer to Haryana’s.

This tells us that buying or attracting machinery is only the first step. What the state produces, how fully its plants are used, the value of its products, the depth of local suppliers and the cost of inputs all affect how much value eventually remains.

Maharashtra is the most interesting example. It has less fixed capital per employee than Gujarat, yet creates more GVA per employee and leads the country’s total factory GVA. Its more diversified automobile, engineering, chemical, pharmaceutical and food base may help it retain value across several stages of production instead of concentrating gross output in one input-heavy industry.

Karnataka shows a similar pattern on a smaller scale: its capital per worker is close to the middle of the group, but GVA per worker is among the highest. This suggests that a state need not have the heaviest machinery to create high value. The type of product, technology, design, intellectual property and position within the value chain matter too.

Gujarat illustrates why output alone can be misleading. Petroleum and chemicals generate enormous sales values, but they also consume expensive crude oil and other inputs. Once those costs are removed, Gujarat’s lead narrows. This is not a sign of weakness; it is a feature of the industries the state specialises in.

The bigger lesson is that capital deepening and value creation are related, but they are not the same thing. States gain more when machinery is combined with skills, supplier capabilities and activities that allow factories to retain a larger part of the final value.

There is one more part of the structure to consider: the size of the factories themselves.

The states with bigger factories also tend to have heavier factories

Haryana has the largest presence of big factories among the states shown. Around 20% of its operating factories employ 200 or more workers. Maharashtra follows at 19%, while Tamil Nadu and Karnataka are at 18%.

Andhra Pradesh sits at the other end, with only 8% of factories employing 200 or more workers.

The economic weight of the average factory broadly follows the same pattern.

An average Haryana factory creates about ₹19 crore of GVA, the highest among these states. Maharashtra and Madhya Pradesh follow at ₹18 crore, while Karnataka generates about ₹16 crore per factory.

Tamil Nadu is an interesting exception. It has one of the highest shares of large factories, but its GVA per factory is only around ₹8 crore. This fits the larger story: Tamil Nadu has a broad, employment-heavy manufacturing base. Its strength comes from the number of factories and workers, not from having the highest economic value per factory.

Haryana shows the opposite combination. It has fewer factories overall than Tamil Nadu, but a larger share of them are big, and the average factory creates more than twice as much GVA.

Factory scale therefore adds another layer to the state story. States with more large plants generally have economically heavier factories. Large anchor plants can support research, testing, exports and supplier networks that smaller factories could not build alone.

But the most successful cluster is not simply a collection of giant factories. It is a system in which large plants create demand for capable smaller suppliers. Those suppliers learn new standards, invest in machinery, hire workers and sometimes begin supplying other industries or exporting on their own.

This is why Haryana’s automobile belt matters beyond the assembly plants. The manufacturers in Gurugram, Manesar and Bawal sit inside a wider component and engineering network. Maharashtra and Tamil Nadu have built similar supplier depth around their automobile clusters.

Tamil Nadu’s lower GVA per factory should also be read in this context. Its strength lies in having a very broad base of factories across different industries, including many labour-absorbing and supplier units. Haryana’s average factory is economically heavier, while Tamil Nadu’s network is wider. One model concentrates value inside large plants; the other spreads production across more factory gates.

Conclusion

Taken together, the charts show that states are not competing through a single manufacturing model. Tamil Nadu leads through a broad, employment-heavy factory base; Gujarat through capital-intensive petroleum, chemicals and metals; Maharashtra through higher value creation; and Haryana through large automobile factories. Madhya Pradesh’s high capital and GVA per worker are notable, but may partly reflect a smaller number of large plants rather than an equally deep industrial network.

The next challenge is therefore different for every state. Tamil Nadu must convert its employment advantage into higher-value work and better wages, while Gujarat needs stronger downstream industries that spread jobs beyond large plants. Maharashtra and Karnataka must deepen their technology, skills and supplier capabilities, while automobile-dependent Haryana must adapt to changes such as the EV transition. For emerging states, attracting an anchor factory is only the beginning, the real test is whether it creates local suppliers, skills and jobs around it. India’s manufacturing strength will ultimately depend not just on adding factories, but on building deeper industrial ecosystems around them.

Subscribe now

Hello, welcome to this week’s edition of Points & Figures. This edition is done by Kulsum & Mridula.

The idea behind this newsletter is simple: to tell stories about the Indian economy, financial markets, and investing through data visualisations. The goal isn’t to send you a bunch of random, out-of-context charts, because pretty much everybody does that already.

Instead, we take large datasets that are often overlooked or under-appreciated and use them to tell stories about trends, shifts, and developments in the Indian economy and markets that you might not otherwise come across. And like everything else we do under Markets by Zerodha, the quality bar remains extremely high.

A factory is the only place where capital, labour and production converge at their strongest. Someone has invested in land, buildings and machinery. Workers have been hired. Raw materials have been bought from other businesses. Electricity, fuel and transport are being consumed. At the end of this process, something of far greater value than the sum of all these parts is expected to emerge.

That makes factories useful for examining several questions at once. Is manufacturing creating enough jobs? What kinds of jobs are being created? Is the sector becoming more dependent on machinery? And is all this investment actually helping factories create more value?

Unlike China, South Korea or Japan, India became a major services economy before it could build a comparably large manufacturing base. Services can generate high-value jobs and enormous economic value, but historically, they’ve not been as effective at absorbing large amounts of labor as manufacturing. So, if we want to know whether India is building a deeper industrial base, its factories are a useful place to look. But there is a complication: the data in this story do not describe every manufacturing unit in India.

What exactly does the ASI measure?

India has a large world of household units, tiny food processors, garment workshops, furniture makers and other small enterprises outside the registered factory system. The Annual Survey of Industries, or ASI, mainly covers the registered or organised factory sector. The unincorporated part of manufacturing is measured separately through surveys such as the Annual Survey of Unincorporated Sector Enterprises.

This distinction is important. The ASI can tell us a great deal about the capital, jobs, output and value created inside registered factories. It cannot, by itself, describe all of Indian manufacturing, most of which is still unorganized.

The ASI’s frame is built mainly from lists of registered factories maintained by state authorities. It covers factories registered under the Factories Act, along with certain bidi and cigar units, electricity undertakings and some other eligible establishments. Some categories, such as defence establishments, railway workshops, government mints, restaurants and oil storage depots, are outside its scope.

The word “factory” also has a legal threshold. Historically, the Factories Act covered units using power with at least 10 workers and units not using power with at least 20 workers. However, several states have amended those thresholds for themselves. For eight states, the 2023–24 ASI notes thresholds of 20 workers with power and 40 without power. This means coverage is not perfectly uniform across the country, and changes in the survey frame can affect comparisons over time.

The ASI does not visit every unit in the same way. It fully enumerates large factories and certain other categories, while drawing a representative sample from the remaining smaller units. In 2023–24, 63,981 units belonged to the census sector and 19,639 sampled units were selected from the sample sector. The responses are then weighted to estimate the wider registered factory universe.

So the national figures that follow are estimates, not a simple count of every factory return.

How many factories were actually operating?

The ASI estimated that India had about 2.13 lakh factories in operation in 2023–24. But “in operation” is more specific than “on the register”.

You see, the survey frame also contains units that have fixed assets but are not producing. Some of them maintain staff, and some do not. These units can remain in the frame in case they restart, but they are not counted as factories in operation for that year. In other words, the 2.13 lakh figure refers to factories functioning during their relevant accounting year, not every unit listed in the ASI frame.

Of these operating factories, roughly 1.97 lakh were in manufacturing — the core goods-producing part of the factory sector. The remainder came from the other activities covered by ASI, such as certain electricity and bidi units.

Where these factories operate has also changed. In 2001, rural areas accounted for about 35% of operating factories. By 2023, that share had risen to 44%, while the urban share declined from roughly 65% to 56%.

This does not mean rural India now produces 44% of factory output or employs 44% of factory workers; the chart counts factories, not their economic weight. A small food-processing unit and a giant urban refinery each count as one factory. Still, it suggests that the registered factory base has gradually spread beyond cities, potentially bringing industrial activity closer to workers, raw materials and smaller towns.

The number and location of factories tell us the breadth of the system. They do not tell us how economically important each factory is. For that, we need to look at what they produce.

What are India’s factories producing?

The ASI assigns each factory a principal industry under the National Industrial Classification, or NIC-2008, based on its main activity. It then places a rupee value on the factory’s output.

In 2023–24, the registered factory sector produced output worth about ₹153 lakh crore. Basic metals accounted for the largest share, at 14.5%, followed by food products at 13.2% and coke and refined petroleum products at 12.2%. Motor vehicles and chemicals were also among the largest industries by output.

This does not mean India physically produced more metal than food. Output here is measured in rupees, not tonnes, litres or units. A rise in the price of steel or petroleum can increase the value of output even when physical production changes much less. For instance, when certain inputs get more expensive, the price of the final good can be inflated.

That is the first double edge in the data: a high output value can reflect scale and production, but it can also reflect high input prices. We need employment and value added to understand what is happening inside the factory.

Factories added jobs, but what kind?

The ASI estimated that factories engaged 1.96 crore people in 2023–24, an increase of 5.9% from the previous year. Over the decade from 2014–15, the official series shows an increase of about 57 lakh persons engaged.

These figures should not be read as India’s entire manufacturing workforce. They cover employment within the ASI factory universe, which is formal. And long-term changes can reflect the opening and closing of factories as well as updates to the survey frame. Still, they show that employment in registered factories has expanded.

Within the 1.96 crore total, around 1.55 crore people were classified as workers. In ASI language, workers are people directly or indirectly involved in the manufacturing process. “Total persons engaged” is broader: not only does it refer to assembly line workers, but it also includes supervisory and managerial staff, other employees, and unpaid family members or working proprietors.

The average operating factory engaged about 92 people in 2023–24. But this average can be misleading. A small number of very large plants pull it upward, while most factories are much smaller. The median factory is therefore likely to employ fewer people than the average suggests.

It is also tempting to say manufacturing creates jobs that require no formal qualifications. That is too broad. Some factory-floor roles can absorb workers without university degrees, which is one reason manufacturing matters for a developing economy. But modern factories also need specialised technicians, machine operators, engineers and managers. The sector creates a range of jobs rather than one uniform kind of work.

The contract-worker question

The ASI separates people hired directly by a factory from workers employed through contractors. As a share of everyone engaged by factories, contract workers increased from about 16% in 2001 to 32% in 2023. Over the same period, the share of directly employed production workers declined to about 47%, while managers and other employees made up the remainder.

There is another valid way to calculate the contract share. If we exclude managers and other employees and compare contract workers only with all production workers, contract labour accounted for about 42% in 2023–24. The two percentages use different denominators: 32% of everyone engaged, versus 42% of workers.

Why do factories use so much contract labour?

One of the reasons is flexibility. Production can rise and fall with orders, seasons and demand. Contractors allow factories to adjust staffing without expanding permanent payroll at the same pace. They may also provide specialised workers or manage recruitment and compliance for large plants.

For businesses, that flexibility can reduce costs and make production easier to scale. For workers, the same arrangement may mean fewer direct ties to the factory and potentially weaker job security or benefits. That is the double-edged question: does contract labour help factories expand employment, or does it transfer more risk to workers?

The ASI cannot answer that on its own. It tells us how workers are engaged and what factories pay in aggregate, but it does not provide enough information on tenure, benefits, safety, job security or working conditions to judge the quality of these jobs. Nor should every contract job automatically be treated as inferior; the category can include different skills and arrangements.

What the data does establish is that contract labour is not marginal. A large part of registered factory production depends on workers who are not directly employed by the factories where they work.

The industries that produce the most are not always the ones that employ the most

At a detailed industry level, garments and clothing were the largest employers in 2023–24, engaging about 7.62 lakh people. Motor-vehicle parts employed roughly 7.54 lakh, followed by allopathic pharmaceuticals at 5.09 lakh. Cotton spinning, hot- and cold-rolled steel, bidi manufacturing and rice milling were also among the ten largest employers.

Compare this with the output chart. Refined petroleum was one of the largest industries by output, but it was nowhere near as important for employment. Garments and auto parts, instead, become far more prominent when we look at the jobs metric.

See, different industries use labour and machinery in very different proportions. A refinery can process enormous value using costly equipment and relatively few people. A garment factory needs far more workers to produce a similar rupee value of output. Even within a broad division such as motor vehicles or pharmaceuticals, it is particular activities, like components or allopathic medicines, that absorb large workforces.

So there is no single answer to which manufacturing industry is most important. Some industries matter because of how much they produce. Others matter because of how many workers they absorb. The policy trade-off is equally important: should India prioritise labour-intensive industries that can create more jobs now, or capital-intensive industries that may build scale, technology and higher value added? A strong industrial base probably needs both.

Some industries need far more capital per worker

On that note, how do factories employ “capital”?

The ASI valued fixed capital in the factory sector at about ₹46.2 lakh crore in 2023–24. Fixed capital includes productive assets such as land, buildings, plant and machinery, transport and computer equipment. Spread across everyone engaged by factories, that works out to an all-industry average of roughly ₹23.6 lakh per person.

But the average hides enormous differences. Coke and refined petroleum had about ₹365.5 lakh of fixed capital per person — more than 15 times the all-industry average. Basic metals stood at roughly ₹52 lakh, while chemicals were around ₹40 lakh. At the other end, wearing apparel had only about ₹3.8 lakh of fixed capital per person.

This helps explain the earlier gap between output and employment. Oil refining can generate huge output with costly plants and relatively few workers. Garment factories rely much more heavily on labour.

The measure still needs care. Fixed capital is the book value of assets already in place; it is not fresh investment during the year. A high figure can reflect expensive machinery, land and large plants, while a low figure can reflect a labour-intensive production process. It does not, by itself, tell us which industry is performing better.

Machines can help workers produce more, improve quality and reduce dangerous manual tasks. At the same time, capital-intensive industries may create fewer jobs for every rupee of assets. The relevant question is not whether India should choose machines or workers, but whether different industries are using their capital effectively.

Is factory capital creating more value?

One way to approach that question is to compare the Gross Value Added (GVA) factories generate with their fixed capital. In FY2020, factories generated about ₹0.41 of GVA for every rupee of fixed capital. The ratio improved to ₹0.55 in FY2022, before settling at ₹0.53 in FY2023 and FY2024.

This helps explain the earlier gap between output and employment. Oil refining can generate huge output with costly plants and relatively few workers. Garment factories rely much more heavily on labour.

The measure still needs care. Fixed capital is the book value of assets already in place; it is not fresh investment during the year. A high figure can reflect expensive machinery, land and large plants, while a low figure can reflect a labour-intensive production process. It does not, by itself, tell us which industry is performing better.

Machines can help workers produce more, improve quality and reduce dangerous manual tasks. At the same time, capital-intensive industries may create fewer jobs for every rupee of assets. The relevant question is not whether India should choose machines or workers, but whether different industries are using their capital effectively.

The broad improvement since FY2020 suggests that each rupee of factory capital is associated with more value added than it was at the beginning of the period. But this is not the same as calculating the return on a new machine or plant.

Fixed capital is a stock measured at book value, while GVA is a flow produced during the year. The ratio is also influenced by capacity utilisation, input prices, product prices and industry mix. The jump in FY2022 partly reflects the recovery from the pandemic-affected base of FY2021, when factories faced shutdowns and disrupted demand.

The careful conclusion is that capital productivity recovered after FY2020 and has remained around ₹0.53 of GVA per rupee of fixed capital for the last two reported years. To understand why, we need to look more closely at GVA itself.

Why did GVA grow twice as fast as output?

Suppose a factory produces goods worth ₹100. To make them, it buys ₹70 worth of raw materials, components, fuel, electricity and services from elsewhere. The factory has therefore created ₹30 of new value.

In simplified form:

GVA = Output − Intermediate inputs

Output can be enormous simply because a factory processes expensive raw materials. GVA removes the value of those purchased inputs and shows how much value the factory itself added through labour and capital.

In 2023–24, factory output was about ₹153.3 lakh crore, while GVA was about ₹24.6 lakh crore. Output per person engaged was ₹78.2 lakh, but GVA per person was only ₹12.6 lakh. The large gap represents the value of materials, fuel, components and services used in production.

Output per person was almost unchanged from the previous year: it moved from ₹78.3 lakh to ₹78.2 lakh. GVA per person rose from ₹11.9 lakh to ₹12.6 lakh, or about 5.6% in nominal terms.

It would be tempting to call this a clear jump in worker productivity. That goes too far. GVA per person is a useful proxy for labour productivity, but it is not a physical measure of what each worker produced. It is affected by prices, profit margins, input costs, capital intensity and industry mix. A fall in the price of raw materials, for instance, can lift GVA even if the number of goods produced barely changes.

What can we say safely? Because output rose by 5.8% while GVA rose by 11.9%, intermediate inputs grew more slowly than output. The GVA-to-output ratio consequently improved from roughly 15.1% to 16%. That may reflect better margins, easing input costs, a shift towards higher-value industries, greater operating efficiency or some combination of these. The aggregate ASI data cannot isolate how much each factor contributed.

This is the second major double edge in the story: creating more value from the same gross output can signal greater efficiency, but in nominal data it can also reflect movements in prices and margins.

A lot of small factories, but most value comes from large ones

Most of this addition to GVA comes from larger rather small factories.

Factories employing 100 people or fewer accounted for about 78% of factories, but only 22% of the workforce, 14% of fixed capital and 12% of value added. Factories with more than 100 employees represented only 22% of factories, yet accounted for 79% of the workforce and 88% of value added.

Why is the gap so large? Big plants combine more workers with much more machinery, larger production runs, specialised processes and extensive supplier networks. Scale can allow each factory to generate far more output and value than a small unit.

But concentration is again double-edged. Large factories can bring scale, productivity, exports and formal employment. Smaller factories can spread entrepreneurship and jobs across more places and may serve as suppliers to larger companies. If small firms remain permanently small because they lack finance, technology, infrastructure or market access, the economy misses an opportunity. If they grow into capable suppliers, the apparent divide between small and large factories can become an industrial network instead.

The right question is therefore not simply whether large factories are better. It is whether India has enough pathways for small factories to become more productive, connect to larger supply chains and scale when they are ready.

Where are the jobs, and where is the value?

Comparing each industry’s share of employment with its share of GVA brings the story together. Food products accounted for 11.1% of factory employment but only 7.4% of GVA. Textiles and wearing apparel also contributed much larger shares of employment than value added. These are relatively labour-intensive industries. Basic metals show the reverse: they accounted for 7.8% of employment but 11.6% of GVA.

Pharmaceuticals produced 7.2% of GVA with 4.9% of employment, while coke and petroleum generated 6.2% of GVA with only 0.9% of jobs. These are more capital-heavy or higher-value activities.

Different industries combine workers, machinery, intellectual property and expensive inputs in very different ways. A high GVA share relative to employment can indicate valuable products or capital-intensive production, but it does not make a labour-intensive industry unproductive or unimportant. Labour-intensive industries may be doing a different economic job: absorbing workers and spreading income more widely.

The most useful way to read this chart is not as a ranking from “good” to “bad”, but as a map of different roles within manufacturing. Some industries are employment engines. Some are value engines. A few manage to be important on both measures.

So, what kind of factory economy is India building?

The 2023–24 ASI presents a factory sector that is expanding employment, becoming more capital-heavy and creating a larger amount of value relative to its gross output. But each positive headline comes with a second question.

Employment increased, but so did the engagement of contract workers, who now account for nearly a third of everyone engaged, or 42% when measured only against production workers.

Capital intensity varies enormously, as oil refining uses more than 15 times the capital per person of the all-industry average.

GVA grew almost twice as fast as output—but prices, margins and input costs can all influence that gap.

Large factories created most of the value, but a healthy industrial system also needs smaller suppliers that can become more productive and scale.

None of this makes the expansion meaningless. It simply changes what we should ask of the data. The goal is not to maximise a single number such as factories, output, jobs or capital. A deeper industrial base needs to combine several things: enough labour-intensive production to absorb workers, enough capital and technology to remain competitive, rising real wages, stronger domestic supply chains and routes for smaller firms to grow.

The ASI gives us one of the clearest views of that process inside registered factories. It shows that India’s factory system is larger and more valuable than it was a decade ago. It also shows why the next stage of the manufacturing story cannot be judged by output alone.

That’s it for the eighth edition.
We hope you enjoyed what we’ve put together for you. This is a new format for us, and we fully understand that there are rough edges and plenty of things that can be improved.

And that’s where your end of the bargain begins.

Please send us your feedback, ideas, suggestions, criticisms, and anything else you think can help make this publication better. Unlike many people who say this as a ritual and then promptly ignore what follows, we genuinely do read the comments, emails, and messages we receive.

Points & Figures is still a work in progress, and we’d love your help in shaping what it becomes.

So let us know what you liked, what you didn’t, what you’d like to see more of, and what you’d like us to do differently. We’ll read it all and use it to make future editions more useful, more insightful, and hopefully a lot more fun.

Until next time.

Subscribe now

Hello, welcome to this week’s edition of Points & Figures. This edition is done by Kulsum & Mridula.

The idea behind this newsletter is simple: to tell stories about the Indian economy, financial markets, and investing through data visualisations. The goal isn’t to send you a bunch of random, out-of-context charts, because pretty much everybody does that already.

Instead, we take large datasets that are often overlooked or under-appreciated and use them to tell stories about trends, shifts, and developments in the Indian economy and markets that you might not otherwise come across. And like everything else we do under Markets by Zerodha, the quality bar remains extremely high.

Every country relies on the rest of the world in one way or another. India is no different. We import things we need to power our homes, run our factories and build our economy. We export the products and services we’ve become good at.

That balance isn’t always even, though. India imports far more physical goods than it exports, resulting in a sizable goods trade deficit. In 2024–25, that deficit reached $283.5 billion.

But goods are only one side of India’s trade story. In services, we regularly see a surplus. In FY 2025, it covered nearly $189 billion of that gap.

India’s services sector, once synonymous with basic IT and back-office work, has evolved far beyond basic outsourcing. Today, we export everything from software and consulting to financial and engineering services. India’s Global Capability Centres, meanwhile, support the day-to-day operations of multinational companies.

The foreign exchange earned from these exports helps offset a large part of what we lose because of the country’s merchandise (goods) trade deficit.

Why does India import so much?

A large import bill isn’t necessarily a sign of weakness. Every economy imports things it either cannot produce in sufficient quantities or simply finds more efficient to source from elsewhere. Some imports are consumed directly, while others become inputs for businesses and industries. The real question isn’t how much India imports, but what role those imports play in the economy.

Take energy, for example. Crude oil, coal and natural gas power homes, factories and transport systems. India simply doesn’t produce enough to meet domestic demand, making these imports essential.

Then there are imports that help create more economic activity. Electronic components, industrial machinery, chemicals and pharmaceutical ingredients are brought into the country so Indian factories can manufacture finished products, many of which are eventually exported.

There are a couple of specific things worth noting, here.

For one, our coal imports have declined about 20%. Why is that? We have two guesses. For one, India’s own coal production has increased. Two, a huge portion of electricity generation has shifted from thermal to renewable energy.

It’s also worth noting that our import of precious stones like pearls, precious and semi-precious stones has declined, although the trade data does not attribute a cause, industry reports suggest the decline was likely driven by weaker global demand for natural diamonds and lower purchases of rough stones by Indian processors.

India’s goods exports are changing

India’s merchandise exports have remained broadly unchanged at $437.7 billion in 2024–25. What we sell to the world, however, is slowly changing. India has long been one of the world’s biggest exporters of refined petroleum products, but newer industries are now catching up.

For instance, India now assembles and exports millions of smartphones every year. Many of the chips, displays and other high-value components still come from overseas, but the finished product is assembled and exported from India. As we cement ourselves in such value chains, electronics, engineering goods and pharmaceuticals are steadily becoming a bigger part of our export basket.

Government policies have played a role in this shift, although not all with the same success. Production Linked Incentive (PLI) schemes have helped attract large investments into sectors such as electronics, turning India into one of the world’s largest smartphone assembly hubs. Special Economic Zones (SEZs) and other export-support measures have also encouraged companies to manufacture in India and serve global markets.

Buying from the East, selling to the west

India imports machinery, electronics, chemicals and pharmaceutical ingredients from China, and large quantities of crude oil from Russia. Meanwhile, the US and Europe remain major markets for Indian services, pharmaceuticals and engineering goods.

Simply put, India relies on global supply chains for many of its energy needs and industrial inputs, while earning foreign exchange by exporting services and an increasingly diverse range of manufactured goods.

Conclusion

Trade isn’t a scoreboard where a country wins by exporting more and importing less. Every major economy imports what it needs and exports what it does best. India’s imports keep homes powered, factories running and industries supplied, while its exports, from services and refined petroleum to pharmaceuticals, engineering goods and electronics, bring foreign exchange and create jobs.

That said, there is something disproportionate to our persistent trade deficit in goods. There’s a simplistic way of looking at this: that we should cut imports, substituting it with our own produce to remove our dependency. That answer has rarely helped any country rise through the world economy.

It is better to ask: how do we make more of what the world needs? That is one of the defining challenges of the Indian economy, today.

That’s it for the seventh edition.

We hope you enjoyed what we’ve put together for you. This is a new format for us, and we fully understand that there are rough edges and plenty of things that can be improved.

And that’s where your end of the bargain begins.

Please send us your feedback, ideas, suggestions, criticisms, and anything else you think can help make this publication better. Unlike many people who say this as a ritual and then promptly ignore what follows, we genuinely do read the comments, emails, and messages we receive.

Points & Figures is still a work in progress, and we’d love your help in shaping what it becomes.

So let us know what you liked, what you didn’t, what you’d like to see more of, and what you’d like us to do differently. We’ll read it all and use it to make future editions more useful, more insightful, and hopefully a lot more fun.

Until next time.

Hello, welcome to this week’s edition of Points & Figures. This edition is done by Kulsum & Mridula.

The idea behind this newsletter is simple: to tell stories about the Indian economy, financial markets, and investing through data visualisations. The goal isn’t to send you a bunch of random, out-of-context charts, because pretty much everybody does that already.

Instead, we take large datasets that are often overlooked or under-appreciated and use them to tell stories about trends, shifts, and developments in the Indian economy and markets that you might not otherwise come across. And like everything else we do under Markets by Zerodha, the quality bar remains extremely high.

If you had the power to improve the economy, what would you do? Make financing easier? Build more highways? One answer may not come to mind immediately: make people healthier.

A healthy person can work, earn and save. Someone who is frequently ill may miss work and spend a large part of their income on treatment. Multiply that difference across millions of households, and health becomes an economic story.

The National Family Health Survey, or NFHS, gives us India’s household health check-up. It tracks how children grow, the diseases adults live with, when women marry and have children, and whether families have financial protection against illness. The latest survey asks a simple question: as India grows, are Indian families becoming healthier?

What we eat is changing

Indian households are spending less of their food budgets on cereals and pulses, and more on fruit, dairy, eggs, meat and packaged food. From what we can tell, diets are becoming more varied, but also more processed. So eating differently does not automatically mean eating better.

The old problem: child undernutrition

Stunting means a child is too short for their age because poor nutrition or repeated illness has held back their growth over a long period. It can affect immunity, learning and future health. Wasting is more immediate: the child is dangerously thin for their height, often after a rapid loss of weight. It raises the risk of illness and death.

There is some good news. Stunting fell from 35.5% in NFHS-5 to 29.3% in NFHS-6. But wasting barely moved, from 19.3% to 19.0%. Long-term improvements in food security, sanitation and healthcare may gradually reduce stunting. Wasting is more sudden and harder to prevent.

The problem can begin even before birth. Indian research found that wasting was already present in many newborns, pointing to poor foetal growth and maternal nutrition. That means prevention must begin before and during pregnancy, not only after a child is born.

The gap continues after birth: only 15.3% of children aged 6–23 months receive frequent and varied meals. Food-support programmes may have improved access to basic calories, but calories alone are not enough. Many children still do not get enough protein, micronutrients or diverse foods. This may be one reason wasting remains stubbornly high. The national average also hides sharp differences between states.

Blood pressure down, blood sugar up

India’s metabolic health is moving in two directions. Elevated blood pressure has declined among both men and women, while high blood sugar has become more common. The survey cannot tell us exactly why: diet, physical activity, age, genetics and better diagnosis can all influence the numbers. But the broader direction is clear, India is carrying a growing burden of chronic disease that requires years of treatment.

Social health is improving

There are also ways in which Indian society has distinctly improved.

Marrying later can give girls more time to finish school, delay childbirth and participate in the workforce, which improves the prospects of the entire household.

UNICEF notes that girls who marry later are more likely to stay in school and less likely to become pregnant at a very young age. This gives them more choice over their lives and a better chance to work and earn, changes that can improve the future of the entire household.

And, India’s total fertility rate remains around two children per woman.

More families have insurance, but gaps remain

As chronic illness becomes more common, families also need protection from its financial cost. More households now have at least one member covered by health insurance or a health-financing scheme.

This can prevent a hospitalisation from immediately turning into debt or forcing a family to sell assets.

But coverage does not guarantee affordable healthcare.

Conclusion

India has made real progress against some of its oldest health problems, but a new bill is arriving. Fewer children are stunted, fewer women are marrying early and more households have insurance. Yet wasting persists, obesity is rising and high blood sugar is becoming more common. Growth alone cannot turn India’s population into an economic advantage. That will depend on whether people are healthy enough to learn, work and build financial security, and whether falling sick still has the power to undo years of progress.

That’s it for the sixth edition.

We hope you enjoyed what we’ve put together for you. This is a new format for us, and we fully understand that there are rough edges and plenty of things that can be improved.

And that’s where your end of the bargain begins.

Please send us your feedback, ideas, suggestions, criticisms, and anything else you think can help make this publication better. Unlike many people who say this as a ritual and then promptly ignore what follows, we genuinely do read the comments, emails, and messages we receive.

Points & Figures is still a work in progress, and we’d love your help in shaping what it becomes.

So let us know what you liked, what you didn’t, what you’d like to see more of, and what you’d like us to do differently. We’ll read it all and use it to make future editions more useful, more insightful, and hopefully a lot more fun.

Until next time.

Subscribe now

Hello, welcome to this week’s edition of Points & Figures. This edition is done by Mridula & Kulsum.
This newsletter is a rebranding of what we used to publish as The Chatter, and the idea behind it is simple: to tell stories about the Indian economy, financial markets, and investing through data visualizations. The goal isn’t to send you a bunch of random, out-of-context charts, because pretty much everybody does that already. Instead, we take large datasets that are often overlooked or underappreciated and use them to tell stories about trends, shifts, and developments in the Indian economy and markets that you might not otherwise come across. And like everything else we do under Markets by Zerodha, the quality bar remains extremely high.

For decades, India built its power system around one goal: delivering electricity to more people quickly enough to sustain economic growth. Coal primarily provided the answer. India possesses large domestic reserves, and unlike solar and wind, coal plants can operate on command, regardless of the time of day or weather conditions. However, coal was never the entire story; hydro, gas, and nuclear power have always been part of the mix. Yet, because the grid, regulations, and decades-long power-purchase contracts were centered on coal, that legacy continues to shape almost every decision India makes today.

Now the objective has evolved. India must meet rapidly climbing electricity demand while simultaneously cleaning up its power generation. The government has set a target of 500 GW of non-fossil-fuel capacity by 2030, complementing its formal climate pledge to derive about half of its installed capacity from non-fossil sources by the same year. The long-term horizon is net-zero emissions by 2070.

The central theme of this piece is a contradiction: India has already met its capacity target. It surpassed the 50% non-fossil mark in mid-2025, five years ahead of schedule. Yet, coal still generates roughly seven out of every ten units of electricity the country consumes. We aim to explain how these two facts coexist: half of the capacity is clean, while the majority of the power generated remains coal-based.

1. The Coal Engine: Why the Grid Still Depends on Coal

It is crucial to understand that capacity and generation are distinct concepts. Capacity represents what has been built, whereas generation represents what actually runs. By January 2026, non-fossil sources accounted for approximately 52% of installed capacity, with coal and lignite reduced to about 44%. However, coal still supplied roughly 69% of the electricity generated in FY2025-26 through January, and 72.4% in April 2026. This is because a coal plant runs for significantly more hours per day than a solar panel or wind turbine.

This is why India cannot simply switch off coal. The grid relies on it for the bulk of its electricity and, crucially, for power that can be scheduled after dark. There is also a more subtle reason: large coal, hydro, and nuclear plants spin heavy turbines. This “spinning mass”—or inertia—cushions the grid against sudden frequency swings when supply and demand fall out of balance. While this does not keep the grid perfectly steady on its own, it acts as a stabilizing force that solar arrays cannot naturally provide.

2. The Sunny Day, Strained Night Paradox

On April 25, 2026, India’s electricity demand reached a record 256.1 GW in the afternoon, which the grid met without shortage. Solar did the heavy lifting; at midday, it supplied nearly one-third of all generation. Across April, renewables covered 16.5% of India’s electricity, the highest monthly share since July 2025.

Yet, that same evening, after solar output faded, the available supply fell short of demand by about 4.2 GW. The previous night, at 10:34 PM, the shortfall was 5.4 GW—roughly the consumption of 27 lakh rural homes. Such shortfalls occurred on 13 of 15 nights in the second half of April. The chart below provides broader context: while solar and wind capacity have grown quickly, their output is concentrated in specific hours and seasons.

The paradox is not that India lacks aggregate generating capacity. Instead, the system can produce a surplus of variable renewable power at midday but fails to provide enough dispatchable generation, storage, or imported power after sunset.

That evening shortfall is, at its core, a flexibility problem. Once solar vanishes, the system cannot bring enough coal, hydro, gas, or storage online quickly enough to keep pace. There is a second, related problem: during peak solar hours, congestion and limited flexibility can force renewable plants to throttle their production. India therefore requires two simultaneous solutions: enough firm supply for the evening and a grid nimble enough to move and store the daytime solar power it already produces.

3. The Clogged Pipelines: Grid and Hardware Bottlenecks

Building renewable capacity is only the first step; electricity must also be moved from resource-rich regions to demand centers at the right time.

India’s renewable resources are geographically lopsided. Rajasthan, Gujarat, Karnataka, and Andhra Pradesh lead in solar, while Gujarat and Tamil Nadu dominate wind power. Demand, however, is dispersed across industrial corridors, cities, and towns. Therefore, renewable-rich states require robust interstate links to transmit power, and building these links has become a bottleneck.

The scale of the planned infrastructure is enormous. A 2022 blueprint for integrating over 500 GW of renewables envisioned approximately 51,000 circuit kilometers of interstate lines and 433,500 MVA of transformation capacity, at an estimated cost of ₹2.44 lakh crore. Separately, the National Electricity Plan aims to increase inter-regional transfer capacity from 119 GW to 168 GW by 2032. These represent related, parallel efforts.

Execution is lagging on four fronts. The most stark example is in Rajasthan, India’s top solar state. In April 2026, the Central Transmission Utility informed the regulator that it was struggling to find transmission routes for roughly 60 GW of renewable projects seeking connectivity. This does not represent 60 GW of stranded power, but rather proposed capacity awaiting a viable way to connect. Furthermore, there is the issue of curtailment: when the grid cannot safely absorb or move renewable power, operators instruct plants to reduce output. An Ember analysis identified complete curtailment of some short-term renewable trades, especially between 11 AM and 2 PM, while separate reporting for April 2026 estimated average curtailed solar at about 23 GWh per day—clean power that was built and paid for, but wasted because the grid could not accommodate it.

Hardware is the third choke point. Transformers are in short supply worldwide. In India, lead times for 220-kV transformers have stretched from about eight to 14 months, with copper and specialized electrical steel adding cost pressure. Finally, the lines themselves face delays due to right-of-way disputes, land acquisition issues, forest and wildlife clearances, litigation, and procurement delays. Building a transmission corridor is often more challenging than constructing the solar farm that feeds it.

4. Shifting Electricity Through Time—and Fixing the Money Flow

Transmission moves electricity across geography; storage moves it across time. Pumped-storage hydropower and batteries can absorb cheap solar power at midday and release it after sunset, smoothing the steep evening ramp that conventional plants must otherwise manage.

India’s storage base remains small relative to the need. CEA data for January 2026 recorded about 276 MW / 792 MWh of grid-scale battery storage. Pumped storage accounts for several gigawatts, with a larger pipeline planned, but these projects are capital-intensive and require years to secure approvals and complete construction.

Funding all of this—storage, local upgrades, and the flexible grid—is complicated by the finances of the state-owned distribution companies (DISCOMs). Many of these companies lose money on every unit sold, spending more to supply electricity than they recover. The causes are complex: subsidized tariffs, delayed government subsidy payments, high technical and commercial losses, and weak billing and collection. Attributing these losses solely to free electricity is an oversimplification. Smart meters can help by improving measurement, billing, and enabling time-of-day pricing that shifts consumption to sunny hours. However, a meter cannot fix a balance sheet; tariff reform, timely subsidies, and loss reduction must accompany these measures. Without healthy DISCOMs, there is no entity to pay for the flexible grid required for this transition.

5. The Hidden Climate Cost of Continued Coal Use

The reliance on firm power explains why renewable growth has not yet displaced coal, and this dependency carries two distinct climate costs: one obvious and one largely invisible.

The obvious cost is power-plant carbon dioxide emissions, where the data tells a blunt story. Coal accounts for the overwhelming majority of the power sector’s emissions. Even as generation and capacity have climbed sharply since 2019, the carbon intensity of each unit has barely budged. This indicates that the increased demand was met largely by burning more fossil fuel rather than displacing it.

The hidden cost is methane. India produced about 1.048 billion tonnes of coal in FY2024-25, and the government expects output to grow by 6–7% annually to about 1.5 billion tonnes by FY2029-30. This includes an initiative to increase underground output beyond 100 million tonnes annually by FY2028-29, starting from a small current base. Mining coal releases trapped gas; unless captured or destroyed, this methane escapes. This is significant because methane has more than 80 times the warming potential of CO₂ over a 20-year period and has driven roughly 30% of the temperature rise since the Industrial Revolution.

Global Energy Monitor has modeled India’s coal-mine methane at 1.6 million tonnes by 2029—approximately 138 million tonnes of CO₂-equivalent on a 20-year basis. While this is a modeled projection rather than an official inventory, it highlights a large, undercounted source of emissions that is not captured when analyzing only power-plant emissions.

6. Conclusion: One Transition, Many State-Level Realities

There is no single Indian energy transition; there are dozens, because resource quality, demand, legacy plants, and procurement processes vary by state.

Some states are well advanced. Karnataka and Rajasthan generate large shares of their electricity from clean sources; Rajasthan leads in installed solar capacity, while Gujarat and Tamil Nadu are wind powerhouses. Nuclear energy provides a smaller but reliably low-carbon layer. India had 8.78 GW installed by January 2026, with an ambition of reaching 100 GW by 2047—an ambitious goal requiring significant investment, regulatory reform, and time.

Coal-heavy industrial states face the opposite scenario. Uttar Pradesh and Chhattisgarh rely primarily on coal for their electricity, and Chhattisgarh—home to a vast fleet of coal plants—is among the largest state-level emitters in the power sector. Bihar and Jharkhand also operate carbon-intensive grids. These comparisons are only meaningful with context: total emissions indicate scale, while grams of CO₂ per kilowatt-hour indicate intensity.

Nationally, the trend is clear. Renewable electricity’s share rose from about 14% in 2000 to 19.8% in 2024 according to Ember’s data. This represents real progress, yet it remains well behind global peers when measured on a like-for-like basis (wind, solar, hydro, and other non-nuclear renewables).

India’s central challenge is no longer just adding megawatts—it is ahead of schedule in that regard. The challenge is stitching the pieces together: integrating clean generation, firm evening supply, necessary transmission, time-shifting storage, and ensuring DISCOMs are solvent enough to pay for the system. If the energy reaches the right place at the right time, the transition works. If any of these pieces fail, the grid will continue to rely on coal after dark.

That’s it for the fifth edition.

We hope you enjoyed what we’ve put together for you. This is a new format for us, and we fully understand that there are rough edges and plenty of things that can be improved.

And that’s where your end of the bargain begins.

Please send us your feedback, ideas, suggestions, criticisms, and anything else you think can help make this publication better. Unlike many people who say this as a ritual and then promptly ignore what follows, we genuinely do read the comments, emails, and messages we receive.

Points & Figures is still a work in progress, and we’d love your help in shaping what it becomes.

So let us know what you liked, what you didn’t, what you’d like to see more of, and what you’d like us to do differently. We’ll read it all and use it to make future editions more useful, more insightful, and hopefully a lot more fun.

Until next time.

Hello, welcome to this week’s edition of Points & Figures. This newsletter is a rebranding of what we used to publish as The Chatter, and the idea behind it is simple: to tell stories about the Indian economy, financial markets, and investing through data visualizations. The goal isn’t to send you a bunch of random, out-of-context charts, because pretty much everybody does that already. Instead, we take large datasets that are often overlooked or underappreciated and use them to tell stories about trends, shifts, and developments in the Indian economy and markets that you might not otherwise come across. And like everything else we do under Markets by Zerodha, the quality bar remains extremely high.

If you want to understand the true state of the Indian economy, don’t look at GDP growth—look at how people borrow.

To understand today’s numbers, you have to rewind a decade. In 2015, the RBI launched its Asset Quality Review (AQR), forcing banks to finally admit they were drowning in toxic corporate debt. Burned by massive corporate defaults, banks pivoted hard to the perceived safety of retail credit. Then, in 2018, the IL&FS collapse triggered a brutal liquidity crisis, paralyzing the NBFC sector. Just as the dust was settling, the pandemic hit.

For ten years, Indian lenders have been stumbling from one systemic shock to the next. The result? Institutional paranoia.

Today, India’s retail lending market is violently split by two opposing forces. At the top, a post-pandemic “K-shaped” economic recovery has armed affluent households with the capital to purchase premium assets. At the bottom, stagnant wages and inflation have forced working-class families into borrowing just to sustain consumption.

Watching this divide widen, the Reserve Bank of India (RBI) recently intervened again. Fearing a bubble in unsecured debt, the central bank aggressively hiked risk weights on retail loans to 125%, effectively slamming the brakes on easy credit.

To map exactly how this is playing out on the ground, this edition of Points & Figures dives into the data from CRIF High Mark’s May 2026 “How India Lends” report. The numbers capture the fallout perfectly: lenders are retreating to safety. The era of loose, unsecured credit is over; we are now in the era of premiumisation and physical collateral.

The Big Picture: Safety Over Scale

By March 2026, Indians owed ₹170.2 lakh crore in retail loans, a 16.6% annual jump. But that single number masks a dramatic pivot in capital allocation.

Secured gold loans exploded by over 50%. Unsecured credit card balances, meanwhile, flatlined at exactly zero. Lenders are no longer betting on borrower intent; they are demanding hard assets.

1. Housing: The Premiumisation Illusion

If you glance at the housing market, it looks like a boom. It isn’t. It is an exclusivity trend.

The home loan market isn’t expanding; it is gentrifying. Wealthy households are leveraging credit for asset creation, while developers have largely abandoned affordable housing. The speed of this shift is staggering: before the pandemic in FY20, loans exceeding ₹75 lakh made up just 19% of new home loan value. Today, they account for 40.7%.

For banks, this premiumisation is a feature, not a bug. Delinquency data exposes a stark divide: smaller-ticket home loans show significantly higher default rates than the ₹75-lakh-plus bracket. Wealthy borrowers use debt to build long-term equity; working-class borrowers, squeezed by real-wage stagnation, are increasingly defaulting on the dream of ownership.

Public Sector Banks (PSUs) dominate this space. Armed with vast rural networks and cheap deposit capital, they are locking in safe, secured, 20-year yields, avoiding the volatility of unsecured lending entirely.

2. The Gold Rush: Collateral as Survival

When the RBI squeezed unsecured personal loans, working-class borrowers turned to the ultimate socio-economic safety net: family jewelry. Inflation drove living costs up, but it also drove gold prices to record highs, meaning the same household assets could suddenly unlock vastly larger loans for basic liquidity.

But this surge is equally driven by lender paranoia. Terrified of unsecured defaults, banks and NBFCs pivoted hard to the zero-risk security of physical gold, transforming it from a last-resort emergency measure into an aggressively marketed mainstream financial tool.

This is a profoundly regional phenomenon, deeply tied to rural and agricultural distress. South India—Tamil Nadu, Andhra Pradesh, Karnataka, and Telangana—holds the lion’s share. In these states, gold is highly financialized. As erratic monsoons and rising input costs squeeze farming incomes, pledging family gold has become the ultimate rural safety net to cover working capital for seeds and fertilizers, or simply to survive the months between harvests. What looks like massive credit growth on a bank’s balance sheet is often just a rural distress signal.

Specialized NBFCs absolutely dominate here. While traditional banks treat gold loans as a peripheral product, NBFCs have spent decades building hyper-efficient infrastructure to appraise, store, and liquidate gold collateral.

3. Credit Cards: The End of the Unsecured Party

The flip side of the gold rush is the credit card freeze.

Post-COVID, lenders aggressively expanded unsecured credit to “new-to-credit” users. Spotting a localized credit explosion, the RBI intervened heavily: hiking capital risk weights, mandating strict consent norms, and penalizing unsolicited card issuance.

The music stopped instantly. Facing rising delinquencies in subprime segments and realizing that high customer acquisition costs weren’t yielding profitable usage, banks froze mass-market issuance to focus exclusively on proven, high-quality borrowers. Total outstanding card balances grew 0.0% year-on-year, and new issuance plummeted 15.2%.

Credit cards remain a heavily gated, hyper-urban product. A staggering 80% of all outstanding dues sit in just ten states, led by Maharashtra with over ₹50,000 crore.

Private banks, which issue over 70% of new cards, have ring-fenced their prime customers. By locking out marginal borrowers, they have kept card delinquencies (31–90 days past due) at just 1.46%, vastly outperforming the rest of the industry’s 1.97%.

4. NBFCs: The Shadow Banks Winning the Middle

With traditional banks retreating to the safety of wealthy urbanites and secured housing, NBFCs are aggressively capturing the rest of the country.

They have become the undisputed kings of the consumption economy. In Q4 FY26, NBFCs originated a massive 70.1% of all new consumer durable loans, funding the digital lending explosion for smartphones and appliances.

This isn’t a niche play; it’s a structural takeover. By aggressively targeting lower-income brackets that borrow to consume, NBFCs now command the lion’s share of originations across almost all small-ticket retail segments.

More importantly, they are managing this risk far better than banks. In consumer durables, NBFCs run a late-payment rate of just 0.67%, less than half the private banks’ 1.49%. The outperformance extends to two-wheeler loans as well (2.70% versus private banks’ 2.93%). They have successfully cracked the code of small-ticket, high-volume underwriting.

The Geographic Divide and the Cracks at the Bottom

Zoom out, and India’s credit market maps perfectly onto its socio-economic divide. Home loans and credit cards belong to the big cities. Over half of all home loan capital never leaves the top eight metros.

Conversely, the loans that fund daily mobility and aspiration—personal, two-wheeler, and consumer durables—skew heavily rural, with half their value originating outside the top 100 cities.

The underlying narrative of this data is deeply cautionary. When you map out the delinquency rates across the entire market, a clear pattern of stress emerges at the bottom of the pyramid.

Look closely at the ticket sizes. Across almost every loan category—from personal credit to housing—the absolute smallest loans are flashing red with the highest rates of default.

Look closely at the ticket sizes. Across almost every loan category—from personal credit to housing—the absolute smallest loans are flashing red with the highest rates of default.

Conclusion: The Two Indias

The Indian credit engine is still humming, but the CRIF data reveals a system that is increasingly engineered to serve only those who already have a safety net.

For the wealthy, debt is a tool to buy bigger homes and accumulate more assets. But for the working class and rural farmers, debt is becoming a mechanism for survival—whether that means relying on NBFCs to buy a smartphone or pledging generational gold to buy seeds. Lenders, burned by the past and boxed in by the RBI, have made their choice. They are locking their vault doors to the vulnerable, and handing the keys to the affluent.

That’s it for the fourth edition.

I hope you enjoyed what we’ve put together for you. This is a new format for us, and I fully understand that there are rough edges and plenty of things that can be improved.

And that’s where your end of the bargain begins.

Please send us your feedback, ideas, suggestions, criticisms, and anything else you think can help make this publication better. Unlike many people who say this as a ritual and then promptly ignore what follows, we genuinely do read the comments, emails, and messages we receive.

Points & Figures is still a work in progress, and we’d love your help in shaping what it becomes.

So let us know what you liked, what you didn’t, what you’d like to see more of, and what you’d like us to do differently. We’ll read it all and use it to make future editions more useful, more insightful, and hopefully a lot more fun.

Until next time.

Hello, my name is Krishna and this is our third edition of Points & Figures.

This is a rebranding of a newsletter we used to publish under The Chatter. In its new avatar, the idea is simple: to tell stories about the Indian economy, financial markets, and investing through data visualizations (and anything else that caught our attention).

This has been a long-pending idea. We simply couldn’t do it earlier because the team that runs all the initiatives under Markets is barely six or seven people. Considering that we already do more than many much larger publishing teams, the idea of telling stories through data visualizations remained just that—an idea.

But thanks to the affordances offered by tools like Claude and ChatGPT, our small, close-knit team is able to do more than ever before.

The goal of this publication isn’t to send you a bunch of random, out-of-context charts—although we might occasionally do that. Pretty much everybody does that already, and there’s little value in us doing the same thing.

Instead, our aim is to take large datasets that are often overlooked or underappreciated and use them to tell stories about important aspects of the Indian economy and financial markets. In the editions to come, you’ll see data-driven stories about trends, shifts, and developments that you might not otherwise come across. And like everything else we do under Markets, the quality bar will remain extremely high.

Our last 2 editions:

What are we looking at today?

About ten years back, none of this existed. Being online all the time wasn’t a thing. Ordering food, paying a bill, catching a cab, streaming a show on your phone: none of it worked the way it does now, because none of it really could. Then Jio entered the market in 2016 and changed the game almost overnight, pricing calls and data so aggressively that most of the other players in the market simply couldn’t keep up. One by one, they either shut down or got absorbed into someone bigger. Today, the market has come down to roughly two players controlling almost everything.

All of this comes out of the data in Jio’s IPO documents. We’ve already covered the IPO itself over on The Daily Brief. What we’re doing here, for Points & Figures, is using the charts in that filing to trace how India’s digital economy actually got to where it is today.

Side note: Big thanks to my colleague Mridula, who helped me put all of this together.

More chronically online than ever before

The first thing we start with is that how big has our digital economy gotten. When we say “digital economy,” we simply mean everything people are buying, paying for, or transacting through online: UPI payments, e-commerce, subscriptions, all of it. In 2019, this made up about 9% of India’s total economic output, worth roughly ₹15.3 lakh crore.

By 2026, that number has moved to 14%, or ₹49.6 lakh crore. On its own, a 5% jump might not sound like much. But it’s worth pausing on what’s actually sitting inside that jump. In this same window, digital wallet users went from 6% of the population to 38%. Government-to-citizen digital services, things like applying for a certificate or accessing a subsidy online, grew 25x. That’s the scale of behaviour change hiding inside what looks, on paper, like a modest shift in a percentage.

Projections for 2031 put the digital economy at ₹125.8 lakh crore, which would mean digital activity accounts for a meaningfully larger share of everything India produces than it does even today.

You can see the same acceleration if you trace how people’s relationship with their phones has changed over the last thirty years, not as a series of distinct eras but as one long, continuous slide from luxury to habit.

Between 1995 and 2005, owning a phone at all was a luxury: calls could cost anywhere from ₹8 to ₹17 a minute, ringtones were still monophonic, and anything resembling mobile internet was barely usable, let alone common.

The decade that followed, from 2005 to 2015, brought cheap dual-SIM phones and basic 2G and 3G data, and with it a culture built around the missed call, where people treated data as something scarce enough to be used sparingly rather than left running in the background.

It’s really only after 2015 that this changes completely: 4G gives way to 5G, UPI payments become routine, streaming becomes the default way people spend their time online, and data gets cheap and constant enough that nobody thinks about it anymore. That last shift is really what rewired how people in India pay for things, work, and stay in touch with each other.

How data got this cheap

None of this happened on its own, and it’s worth being precise about why. It was Jio’s entry in 2016 that forced the price down, not some industry-wide strategy. Jio priced data so aggressively that the rest of the market had no choice but to match it just to hold on to customers. In 2016, before that shock hit, 1GB of data cost around ₹228, expensive enough that the average Indian used barely 0.2 GB a month.

Once the price war that followed Jio’s entry kicked in, that cost fell sharply: by 2019, 1GB cost just ₹9.9, and monthly usage had already climbed to 7.5 GB. Today, in 2026, 1GB costs ₹7.9, and the average user gets through 25.7 GB a month, which is the natural result of data becoming cheap enough that people stop keeping track of how much they’re using and just stay connected.

That same shift is visible in how mobile broadband has spread, where mobile broadband simply means a high-speed wireless connection on your phone, as opposed to a basic voice-only SIM with no real data attached.

In 2019, only 39% of Indians had a mobile broadband connection; by 2026, that number had climbed to 68%, which works out to over a billion people online. That’s the base everything else sits on top of, from the UPI payments to the government services we just looked at, to the e-commerce we’re about to get into.

Where India still lags compared to the world

India’s absolute numbers look impressive on their own, but they mean more once you remember where the country started. Not long ago, cash on delivery was close to the only way most people would buy anything online, because paying upfront for something that might not show up wasn’t something people trusted.

About 41% of Indians use streaming apps, which compares reasonably well to China’s 71% and the US’s 88% once you account for how much lower average incomes are in India relative to those two markets. E-commerce is where the real story is. Only 18% of Indians shop online regularly, against 82% of Americans and 69% of Chinese citizens. That gap is real, but so is the shift behind it: going from a market that ran almost entirely on cash on delivery to one where close to a fifth of the country now pays digitally for what it buys online is genuine progress, even with a long way still to go. The Indian internet today is still used mainly for chatting, watching things, and proving who you are, more than for buying things, and closing that gap is probably the single biggest opportunity left in the market over the next decade.

Part of why India managed to bring so many people online this quickly is that its telecom operators stopped waiting for global technology to fully mature before deploying it themselves. For most of its history, Indian infrastructure trailed the rest of the world by years.

3G, which went global in the early 2000s, took nearly a decade to arrive properly in India, and with 4G that gap had narrowed to five or six years. With 5G, it nearly closed altogether: global rollouts began around 2019 and 2020, and Indian operators had full nationwide 5G networks running by 2022, a lag of just two to three years. That timing matters because it means Indian developers are now building on largely the same generation of network technology as their counterparts in Silicon Valley, even if actual speeds on the ground still have some catching up to do.

How rural India got online

That rapid rollout reshaped who actually uses the internet in India.

Before 4G arrived at scale, broadband access was mostly an urban and upper-income phenomenon, and the numbers from 2019 reflect that.

There were 54.3 crore broadband users against 61.6 crore people still on older, voice-only connections, with broadband penetration sitting at just 39%. By 2026, voice-only users had fallen to 26.3 crore, broadband had crossed the one-billion mark, and penetration had climbed past 68%. Looking ahead to 2031, voice-only connections are expected to nearly disappear altogether, with broadband projected to cover 98% of all mobile customers. What that tells you is that future growth isn’t going to come from cities, which are already mostly saturated, but from rural India, which still has real room to grow.

Supporting a billion broadband users who are each burning through 25 GB a month takes an enormous amount of spectrum, which is simply the radio frequency that carries wireless signals through the air. It helps to think of spectrum as a set of lanes on a highway: every telecom company has to buy its own lanes to reach your phone, and if it doesn’t own enough of them, its network slows down and gets congested no matter how good its towers are. This is why telecom wars are actually decided at the spectrum auction, long before any customer notices a difference in their app.

Jio currently holds the largest share of total spectrum, at 45% of the country’s airwaves, with Airtel close behind at 38%.

But the number that matters most strategically is low-band spectrum, which covers frequencies below 1 GHz and travels further while cutting through concrete more effectively, making it the backbone of both indoor coverage in dense cities and long-range reach across rural areas. Jio has cornered 39% of this layer on its own, compared to Airtel’s 16% and Vodafone Idea’s 13%, and that low-band advantage is a significant part of why Jio has ended up leading the market.

The gap in home broadband

Mobile is largely a solved problem at this point.

Home broadband, meaning a fixed internet connection running into a specific house or office rather than something that travels with you, is a completely different story. Of India’s 31.2 crore homes, only 6.36 crore currently have a fixed internet connection, which leaves an enormous and mostly untapped market.

Most of the growth so far has come through fibre, which refers to the physical glass cables buried underground that move data as pulses of light and remain the fastest, most stable way to transmit data that exists today. Fibre connections have grown from 14 lakh homes to 4.27 crore over this period.

But the more significant recent shift has been the arrival of Fixed Wireless Access, or FWA, which exists because digging up roads to lay physical fibre into every individual house is expensive and slow. Instead, telecom companies run fibre to a tower somewhere near your neighbourhood and beam that connection wirelessly to a small receiver mounted on your roof or window, so that it behaves like ordinary home Wi-Fi except for the fact that the last stretch of the journey is wireless rather than cabled.

Starting from almost nothing, FWA already connects 1.66 crore homes

Even accounting for that growth, India’s home broadband penetration sits at only around 20%, which looks small set against the rest of the world. China, South Korea, the UK, and the US all have penetration rates above 100%, meaning more active home connections than actual households, while Brazil, a far more realistic comparison for a market like India’s, sits at 61%. That gap between India and a country like Brazil looks like one of the more promising openings left for Indian telecom, even if the exact economics of capturing it still need to play out.

Comparing Jio, Airtel, and Vodafone Idea

With the price wars largely behind them, Jio and Airtel have built two genuinely different, and genuinely profitable, businesses on top of the same market.

Airtel has leaned into a premium strategy, with a heavier skew toward urban users, postpaid plans, and family bundles, and that shows up directly in its numbers: an India mobile ARPU of ₹257.2 a month against Jio’s ₹214, and ~₹34,000 crore in net profit.

Jio has taken the opposite approach, competing on scale rather than price, carrying 52 crore customers who each use an average of 42.3 GB a month, and converting that sheer volume into roughly ₹30,000 crore in profit.

Vodafone Idea, meanwhile, hasn’t managed to build either kind of advantage: it lost 54 lakh customers over the past year, its ARPU remains stuck at ₹174, and its margins are still underwater, leaving it without the cash to compete with the scale of investment Jio and Airtel are both able to make.

The bottom line

The first decade of India’s digital economy was really about one thing: getting as many people online as cheaply as possible. The next decade looks set to be about something different, home broadband and the margins that come with owning the physical connection into people’s houses. As Jio and Airtel tighten their hold on the residential market through 5G FWA, and as bills across the industry start to drift upward again, the value in this business is likely to keep flowing toward whichever two companies actually own the wires and towers that run into your home.

That’s it for the third edition.

I hope you enjoyed what we’ve put together for you. This is a new format for us, and I fully understand that there are rough edges and plenty of things that can be improved.

And that’s where your end of the bargain begins.

Please send us your feedback, ideas, suggestions, criticisms, and anything else you think can help make this publication better. Unlike many people who say this as a ritual and then promptly ignore what follows, we genuinely do read the comments, emails, and messages we receive.

Points & Figures is still a work in progress, and we’d love your help in shaping what it becomes.

So let us know what you liked, what you didn’t, what you’d like to see more of, and what you’d like us to do differently. We’ll read it all and use it to make future editions more useful, more insightful, and hopefully a lot more fun.

Until next time.

Hi, my name is Bhuvan, and welcome to the second edition of Points & Figures.

This is a rebranding of a newsletter we used to publish under The Chatter. In its new avatar, the idea is simple: to tell stories about the Indian economy, financial markets, and investing through data visualizations (and anything else that caught our attention).

This has been a long-pending idea. We simply couldn’t do it earlier because the team that runs all the initiatives under Markets is barely six or seven people. Considering that we already do more than many much larger publishing teams, the idea of telling stories through data visualizations remained just that—an idea.

But thanks to the affordances offered by tools like Claude and ChatGPT, our small, close-knit team is able to do more than ever before. Coincidentally today’s edition is around this :)

The goal of this publication isn’t to send you a bunch of random, out-of-context charts—although we might occasionally do that. Pretty much everybody does that already, and there’s little value in us doing the same thing.

Instead, our aim is to take large datasets that are often overlooked or underappreciated and use them to tell stories about important aspects of the Indian economy and financial markets. In the editions to come, you’ll see data-driven stories about trends, shifts, and developments that you might not otherwise come across. And like everything else we do under Markets, the quality bar will remain extremely high.

This was our last edition:

First, what are we looking at?

At the start of 2025, almost nobody outside a few research circles had heard of DeepSeek. Eighteen months later, models built by Chinese labs handle more of the world’s developer AI traffic than American ones do. Reversals that size usually take a decade in infrastructure markets. This one took about a year, and you can watch it happen frame by frame in the charts that follow.

The data comes from OpenRouter, and it’s worth understanding why that’s a useful vantage point. OpenRouter is a switchboard for AI models: instead of wiring your app directly to OpenAI or Anthropic, you point it at OpenRouter and it forwards each request to whichever of several hundred models you choose, across dozens of labs, behind one key and one bill. Developers use it precisely because it lets them swap models without rewriting anything — which means that when a cheaper or better model appears, the traffic can move almost overnight. Because every request passes through one chokepoint, OpenRouter can publish what almost no single lab will: anonymised rankings of which models, apps, languages and kinds of task are actually being used, week by week, measured in tokens.

A token is the unit throughout — the chunk of text, very roughly three-quarters of a word, that a model reads and writes. “Share of usage” means share of tokens.

Now the honest part, because it changes how you should read everything below. OpenRouter is not the whole AI market; it’s one revealing slice of it. It sees developers wiring models into software through an API. It does not see the hundreds of millions of people typing into ChatGPT, Claude or Gemini directly, the big enterprise contracts signed lab-to-customer, or the models companies download and run on their own machines — which also means the smallest, cheapest open models, the ones quietly doing on-device translation and summarisation, are nearly invisible here. The crowd it does capture leans a particular way: toward people building agents and coding tools, who run models in long automated loops where price per token is the dominant cost. OpenRouter describes its own data as “a representative lens on large-scale LLM usage dynamics,” and within its slice that’s fair. Just keep the slice in mind: this is the open, developer-facing, price-sensitive end of AI, which happens to be exactly where the most interesting competition is playing out.

A few smaller caveats:

• Money is an estimate. Where we talk about spend, it’s tokens multiplied by each model’s list price. Real prices are negotiated, discounted and uneven, so read the money charts as relative scale, not as anyone’s actual bill.

• Model traits come from a catalogue. Whether a model reasons, can see images, or has a long memory is read from OpenRouter’s model catalogue. About three-quarters of tokens map cleanly to a catalogued, priced model; the rest sit out those charts.

• App categories are inferred from app names, so the line between “coding” and “chat” is approximate at the edges.

With that settled, here is the story the numbers tell.

Chinese labs now supply most AI usage

Start with the single fact everything else hangs off: where the models people use are built. Through early 2025 this was an American chart — OpenAI, Anthropic and Google, with everyone else rounding to noise. The amber band climbing from the floor is China, and by mid-2026 it is the largest single block on the page. The cause is partly geopolitical: cut off from Nvidia’s top chips by US export controls, Chinese labs were pushed to compete on efficiency rather than raw scale, then to give the results away as free downloads to win adoption. The constraint became a strategy, and the strategy shows up here as a colour inversion.

Chinese models went from niche to the lead in a year

The same shift, narrowed to two lines so the crossover is unmistakable: Chinese-built models against the four large US “closed” labs (OpenAI, Anthropic, Google, xAI). “Closed” simply means a model you can only reach through the maker’s paid service, never download. China begins as a rounding error and ends in front. Hold one caveat in your head as you read it, though, because the next few charts turn on it: this is a contest for usage, and usage is not the same thing as revenue.

A crowded field — and most of it is now Chinese

Drop from countries down to companies and a second pattern appears: nobody is running away with this. A dozen labs share the market, and the field has grown more crowded over time, not less. We colour the Chinese labs amber and the US labs blue, fold the long tail of smaller names into an “Other” band of the matching colour, and the picture resolves into two blocks — with the amber one now heavier than the blue.

The usage leaderboard reshuffled around Chinese labs

Rank the labs by usage each month and follow the lines as they cross. DeepSeek, Qwen, Kimi, MiniMax and the rest climb from the bottom of the table to the top while the US incumbents slide; every crossing is one lab overtaking another. The churn is the point. A position on this board is rented, not owned — won with the last strong release and lost to the next one.

Every major lab’s rise and fall, side by side

Stacked charts flatter the whole and hide the parts, so here each lab gets its own panel drawn on the same axis. Side by side you can see how different the trajectories are: steady climbers, slow faders, and the occasional lab that spikes on a hot launch and gives most of it back within weeks. It is a useful corrective to any narrative that treats “the market” as a single smooth trend.

It’s not just DeepSeek — a whole wave of Chinese labs

DeepSeek became the headline in early 2025, but reading “China” as “DeepSeek” misses what actually happened. The amber block is built from many names — Qwen, Kimi, MiniMax, GLM and more — each shipping capable models, most of them free to download. Alibaba’s Qwen alone has passed a billion downloads, overtaking Meta’s Llama as the most-used open model family in the world. The strategy across the group is the same: give the model away, win the developers, monetise around the edges. The breadth is what makes it durable rather than a single lucky release.

No single lab owns the market anymore

This one cuts against software’s usual grain. Platforms normally tend toward winner-take- all: the leader’s advantages compound and concentration rises. Here it is doing the opposite — the top three labs’ combined share has fallen from roughly four-fifths to under half. When capable models are cheap, open and interchangeable, no one can build a moat out of being the only option. Switching costs collapse, and with them the usual gravity of the market.

Downloadable models overtook locked-down ones

This is the mechanism behind the deconcentration. An “open” model is one whose weights anyone can download and host; a “closed” one runs only through its maker’s service. Open-weight models have crossed from minority to majority of usage. The economics are unforgiving in the open model’s favour: once weights are public, dozens of providers compete to serve the same model and the hosting price is bid down toward cost. A closed model has exactly one seller, who keeps the pricing power — and, increasingly, a shrinking share of the volume.

A token is cheap from China, expensive from America

Weight the price by what people actually run, and the US frontier sits roughly an order of magnitude above the Chinese field. This is a deliberate split, not a failure to compete: America’s leading labs have chosen not to enter the price war, betting that buyers will pay a premium for the top of the capability range. It’s worth being precise about what’s falling, too. Across the industry the price per token has dropped fast for ordinary models, but the price of the very best reasoning has held remarkably firm. What collapsed isn’t the cost of a token so much as the cost of a given level of intelligence.

China wins the tokens; the US still wins the money

Here is where usage and revenue part ways. We’ve watched China take the volume; multiply each model’s usage by its price and the chart turns inside out. US closed labs still collect the great majority of the dollars, because they charge so much more per token that a sliver of usage still adds up to most of the money. (Spend here is estimated from list prices, so treat it as the shape of the thing rather than a ledger.) Two businesses are hiding inside one market: a volume business and a margin business, and they are not run by the same people.

The models that win on usage lose on revenue

The same split, frozen into a single frame. For each region one dot marks its share of usage and the other its share of estimated spend, and the gap between them is the whole story. China sits far to the usage side and far from the money; the US is the mirror image. One picture, two economies.

The price of intelligence collapsed

Step back and look at the entire field at once. Every leading model is placed by its list price (left to right, on a scale that runs from cents to dollars) and by how much text it can take in at once (bottom to top), with each bubble sized by how heavily it’s used. The amber Chinese models pack into the cheap left; the blue US frontier — the Claudes and GPTs — sits almost alone on the expensive right, often fifty to a hundred times pricier for broadly comparable work. A buyer’s real question is no longer “can I get a capable model?” but “how much am I willing to pay for the last increment of quality?” — and for most jobs the honest answer has moved a long way left.

OpenRouter runs on coding tools, not chatbots

Who is generating all these tokens? Overwhelmingly software, not people in conversation. Coding assistants and autonomous “agents” dwarf chat. This is the single most important thing to hold in mind while reading the rest: OpenRouter’s world is builders wiring models into products, and that crowd behaves differently from consumers. They run models in bulk, they watch the per-token price like a hawk, and they switch the instant a better deal appears — which is exactly why cheap, open models took this market first.

A coding agent burns 5× the tokens of a chat

Coding doesn’t just have more users; each request is far heavier. A coding agent reads whole files, runs tools and feeds the output back on itself before it answers, so a single overnight run can call a model thousands of times. A chat turn is a sentence or two. That difference in appetite is why per-token price dominates a developer’s bill, and why the cheapest competent model tends to win the work rather than the most capable one.

Chinese models win coding; the US keeps law and health

Demand isn’t uniform across the kinds of work people do. Split usage by task and the Chinese models lead where iteration is constant and cost compounds — coding, translation, general technical work. US labs hold their ground in the regulated, high-stakes fields: legal, health, finance, where a buyer is paying as much for trust, accountability and data assurances as for raw capability. The 50% line is the divide; which side a task falls on says a lot about what its users are really buying.

Chinese models, but the prompts are mostly English

A small fact that punctures a big assumption: “built in China” is not “used in Chinese.” Even with Chinese labs supplying most of the models, the people running them are a global, overwhelmingly English-speaking developer crowd, and Chinese-language requests are under a tenth of the total. A good open model is stateless about its origins — it goes wherever it’s cheapest and best, regardless of the flag on the lab that made it.

Reasoning models now do almost all the work

The last three charts are about how the models themselves changed, and this is the largest shift of the three. A “reasoning” model works a problem through in hidden intermediate steps before it answers — slower and more expensive per call, but markedly more reliable on anything with structure. The idea went mainstream with OpenAI’s o1 in late 2024 and DeepSeek’s open R1 soon after; in the year since, this style has gone from a tenth of usage to over nine-tenths. The default expectation of a model has quietly moved from “answer immediately” to “think first, then answer.”

Half of AI usage now runs on models that can see

Models also grew senses. Around half of usage now flows to models that can take in images, audio or video alongside text. One careful distinction: this tracks what a model can do, not what each request happens to use. A model that can read images counts here even when the prompt is plain text — which is the point. Multimodality has stopped being a special feature and become part of the standard kit, whether or not a given job needs it.

Models that read a whole book at once took over

Finally, memory. A model’s “context window” is how much text it can hold in mind before answering; a million-token window is roughly a 700-page book at once. Usage has shifted almost entirely to these very-long-context models. Handing an AI a whole codebase, a stack of contracts or a year of documents in a single prompt has gone from a stunt to a routine, and the short windows that defined 2023 now barely register.

Background and further reading: OpenRouter’s own State of AI study and the accompanying 100-trillion-token analysis; Stanford HAI on China’s open-weight ecosystem; and Epoch AI on how inference prices have fallen.

That’s it for the second edition.

I hope you enjoyed what we’ve put together for you. This is a new format for us, and I fully understand that there are rough edges and plenty of things that can be improved.

And that’s where your end of the bargain begins.

Please send us your feedback, ideas, suggestions, criticisms, and anything else you think can help make this publication better. Unlike many people who say this as a ritual and then promptly ignore what follows, we genuinely do read the comments, emails, and messages we receive.

Points & Figures is still a work in progress, and we’d love your help in shaping what it becomes.

So let us know what you liked, what you didn’t, what you’d like to see more of, and what you’d like us to do differently. We’ll read it all and use it to make future editions more useful, more insightful, and hopefully a lot more fun.

Until next time.

Hi, my name is Bhuvan, and welcome to the first edition of Points & Figures.

This is a rebranding of a newsletter we used to publish under The Chatter. In its new avatar, the idea is simple: to tell stories about the Indian economy, financial markets, and investing through data visualizations.

This has been a long-pending idea. We simply couldn’t do it earlier because the team that runs all the initiatives under Markets is barely six or seven people. Considering that we already do more than many much larger publishing teams, the idea of telling stories through data visualizations remained just that—an idea.

But thanks to the affordances offered by tools like Claude and ChatGPT, our small, close-knit team is able to do more than ever before. And that’s the reason you’re reading this edition.

The goal of this publication isn’t to send you a bunch of random, out-of-context charts—although we might occasionally do that. Pretty much everybody does that already, and there’s little value in us doing the same thing.

Instead, our aim is to take large datasets that are often overlooked or underappreciated and use them to tell stories about important aspects of the Indian economy and financial markets. In the editions to come, you’ll see data-driven stories about trends, shifts, and developments that you might not otherwise come across.

And like everything else we do under Markets, the quality bar will remain extremely high.

For our first edition, we thought we’d dive into the Draft Red Herring Prospectus (DRHP) filed by the National Stock Exchange for its IPO and pick out some of the most interesting data points that help explain how Indian capital markets have evolved.

The NSE is the dominant exchange for almost every major asset class in India—equities, equity derivatives, debt securities, ETFs, and more. Its DRHP is a goldmine of insights into both the evolution and current state of Indian capital markets.

My young colleagues, Mridula and Kulsum, went through the 600-page document and picked out the most interesting observations, so you don’t have to.

With that said, here’s our first data story on how Indian capital markets have evolved.

Points & Figures is still in its infancy. While we have a reasonably clear idea of what we want it to become, we’re still figuring out many things along the way. So, as always, we’d love to hear what you think. Read it, send us your feedback, suggestions, and criticism, and we’ll keep making it better.

History of Indian stock market

A quick history of the Indian capital markets.

How India invests

Although this chart shows that close to 50% of Indian household assets are held in financial assets, investments in equities and mutual funds still account for only a small share of that. The bulk of household financial savings continues to be allocated to insurance policies, pension funds, bank deposits, and similar instruments.

In that sense, it is still very early days for Indian capital markets when it comes to households investing in direct equities, mutual funds, bonds, and other market-linked products. That said, there has been a clear shift in attitudes toward investing. The post-COVID period saw a large influx of younger, first-time investors, helped by a spectacular bull market and, inevitably, the lure of quick gains.

While much of the post-COVID froth and speculative fervor has now left the market, investing has become more mainstream than ever. Mutual funds, in particular, have emerged as the go-to instrument for Indians seeking exposure to financial markets, whether through equities, gold, debt, or other asset classes.

It’s important to remember that India is still a relatively poor country. A large number of Indians still do not have enough surplus income to save and invest meaningfully. In that sense, the growth of Indian capital markets is ultimately a bet on rising incomes and increasing prosperity.

The opportunity ahead remains enormous precisely because so many Indians are yet to become participants in formal financial markets. As incomes rise, savings increase, and more households move beyond meeting their immediate needs, a larger share of those savings is likely to find its way into financial assets.

That doesn’t mean the path ahead will be smooth. India still has its work cut out for it. The geopolitical environment is arguably more uncertain than it has been in decades. Technological shifts, particularly those driven by AI, could reshape industries, jobs, and economic structures in ways that are difficult to predict. And there are plenty of domestic challenges that remain unresolved.

But there is also reason for optimism.

When you step back and look at how far India has come over the last few decades—in terms of incomes, financial inclusion, digital infrastructure, access to markets, and economic opportunity—the progress has been remarkable.

Ultimately, the long-term growth of Indian capital markets is inseparable from the long-term growth of India itself. If India becomes richer, more productive, and more prosperous, that prosperity will inevitably flow through to its capital markets, financial institutions, and the broader financial ecosystem.

Total unique investors

Although you’ll see plenty of data points about how large Indian capital markets are—turnover figures, the number of derivatives contracts traded, mutual fund AUM, cash market volumes, and so on—ultimately, the single most important metric is much simpler: the number of unique investors participating in the markets.

By that measure, India today has nearly 13 crore unique investors, with “unique” defined as unique PAN holders. Interestingly, this is a relatively new data point. NSE only started disclosing it in recent years, and until around 2021–22, we didn’t really have a clear picture of the true size of India’s investor base.

What’s remarkable is how quickly this number has grown. Prior to COVID, the unique investor base was only around 1–1.5 crore. In the span of roughly six years, that number has grown to nearly 13 crore.

Here’s a fun fact: in the last six years alone, India has added more investors than it did in the entire history of its capital markets from the 1950s up until COVID.

To my mind, the history of Indian capital markets is now divided into two distinct eras: the period before COVID and the period after COVID. The first runs roughly from the 1960s to 2020. The second begins in 2020 and continues today.

In just a few years, an extraordinary amount of change has been compressed into the system. The scale and speed of the transformation—in participation, awareness, accessibility, and adoption—are such that I still find it difficult to fully wrap my head around how much has changed in such a short span of time.

The spectacular growth of Indian capital markets after COVID wasn’t driven solely by rising interest in investing. Equally important was the fact that, by the time the pandemic arrived, India had already built much of the digital infrastructure required to support mass participation in financial markets.

Opening a broking account, investing in mutual funds, completing KYC requirements, signing documents, and verifying identity could all be done entirely online, thanks to the combination of Aadhaar, e-signatures, and DigiLocker. Payments had become seamless through UPI. Mobile internet was affordable and widely accessible thanks in large part to the disruption brought about by Jio. And for the first time in India’s history, hundreds of millions of people had access to the internet through a smartphone in their pocket.

In hindsight, this digital infrastructure was one of the most important enablers of the post-COVID investing boom.

Had the pandemic occurred a decade earlier, it is difficult to imagine Indian capital markets witnessing the same scale of participation. The friction involved in opening accounts, transferring money, completing paperwork, and accessing market information would have been far higher.

Instead, when millions of Indians became curious about investing during the post-COVID bull market, the infrastructure was already in place. The rails had been built. All people had to do was download an app, complete a few digital verification steps, and start investing.

The rise of Indian capital markets over the last few years is therefore not just a story about markets. It is also a story about India’s digital public infrastructure and how the combination of Aadhaar, DigiLocker, e-signatures, UPI, affordable smartphones, and cheap mobile internet dramatically reduced the barriers to financial participation.

Another illustration of the post-COVID expansion in investor participation is the explosion in cash market activity.

This chart shows the total turnover in the equity cash market. On the eve of COVID, annual turnover stood at roughly ₹100 lakh crore. Since then, it has nearly tripled, reflecting both the sharp increase in the number of market participants and the broader rise in trading activity.

The slight decline at the end of the series is largely a function of market conditions. Indian equities began correcting around September 2024, and cash market turnover has yet to recover to the peaks seen during the post-COVID bull market.

What’s worth noting is that turnover figures tend to fluctuate with market sentiment and activity levels. But even after accounting for the recent slowdown, the scale of participation in the cash market today is dramatically higher than it was just a few years ago, underscoring how profoundly Indian capital markets have changed in the post-COVID era.

The relentless growth of mutual funds

The star of the post-COVID phase of Indian capital markets has undoubtedly been mutual funds.

Mutual funds have become the default way for the average Indian to get exposure to the markets, and we’ve seen a spectacular increase in assets under management over the last few years. What’s particularly interesting is that this growth has been broad-based. It’s not just that more money is flowing into mutual funds; the number of unique investors has increased dramatically as well, and investors are increasingly using a wider variety of products.

The mutual fund industry has also evolved alongside investor demand. Today, investors have access to everything from low-cost index funds to all-in-one asset allocation products like multi-asset funds. More recently, regulators have approved life cycle funds, which are broadly India’s version of target-date funds.

Perhaps the biggest change is just how easy it has become for investors to build diversified portfolios through mutual funds. Today, investors can get exposure to equities, debt, gold, commodities, and even international markets through a single investment platform. International investing has become somewhat more difficult because of the overseas investment limits, but the broader point still stands: mutual funds have made diversification dramatically simpler than it was a decade ago.

For the average salaried investor with some money to save every month, mutual funds have effectively become the default investment vehicle. But it’s not just retail investors. Mutual funds are also widely used by HNIs, family offices, and corporate treasuries, partly because of the convenience and flexibility of the structure and partly because of the tax efficiencies that mutual funds have historically offered relative to many alternatives.

In many ways, if the post-COVID era belongs to any one product category, it belongs to mutual funds.

Increase in options volumes

As popular as direct equities and mutual funds have become, the real star of the post-COVID era has been the spectacular rise in options trading activity.

In many ways, this has been the defining trend in Indian capital markets over the last few years, and exchanges, brokers, and other market participants have all been beneficiaries of this surge in activity.

While futures trading has also grown, options have emerged as the preferred instrument for most traders. One reason is cost. Futures have become relatively more expensive to trade because of the impact of the Securities Transaction Tax (STT), whereas options offer a more capital-efficient way to take directional or hedged positions.

As a result, options have become the go-to instrument for retail participation in derivatives markets, especially in the post-COVID period.

That said, activity has moderated from its peak. Trading volumes and turnover have cooled somewhat due to a combination of factors, including tighter regulations around derivatives trading, higher margin requirements, and a broader lull in market sentiment as equities have largely moved sideways. Even so, the scale of options trading today remains far higher than it was before COVID, making it one of the most significant structural shifts in the evolution of Indian capital markets.

Indian derivatives trading is big yet small

One of the most misleading charts you’ll come across if you’ve spent any time following Indian markets is the one that proudly proclaims India to be the world’s largest derivatives market based on the number of contracts traded.

To put it mildly, this is a deeply flawed comparison.

The problem is that the absolute number of contracts traded is a largely meaningless metric when comparing derivatives markets across countries. What matters is not how many contracts are traded, or even the notional turnover they represent, but the premium turnover—the actual amount of money changing hands and being risked by traders.

Yet it’s common to see charts comparing the number of options contracts traded in India with those traded in the United States and then drawing sweeping conclusions about the size of the respective markets. I’ve written about this before, but the comparison simply doesn’t hold up.

When you compare India and the US using premium turnover instead of contract count, the picture changes dramatically. India’s derivatives market is still a fraction of the size of the US market. In fact, the difference is so large that India’s numbers barely register on the same chart.

This distinction also helps put another popular narrative into perspective—that vast numbers of Indians are speculating recklessly in derivatives markets. While options trading has undoubtedly exploded in recent years, the actual number of unique individuals actively trading options remains relatively small.

The NSE DRHP doesn’t provide this number, but various disclosures and industry estimates suggest that only around 15–20 lakh unique individuals actively trade options in India. That’s a meaningful number, but it’s a far cry from the impression one gets from headlines about India being the world’s largest derivatives market.

As with many things in finance, the choice of metric can completely change the story being told. In this case, counting contracts creates a dramatically distorted picture of both the size of India’s derivatives market and the extent of retail participation within it.

Bonds, coporate bonds

India has historically been a bank-dominated economy when it comes to funding businesses. While the corporate bond market may look large in absolute terms, it remains a relatively small percentage of GDP, which gives you a sense of how much room there is for growth.

What’s also worth noting is that the Indian bond market has traditionally been dominated by government securities. If I remember the numbers correctly, more than 60% of the bonds traded are G-Secs. In other words, the bond market has largely been a government bond market, with corporate bonds accounting for a much smaller share of activity.

Corporate bonds have historically been unpopular for a variety of structural, regulatory, and taxation-related reasons. As a result, companies have generally found it easier to raise money from banks rather than directly from debt markets. For a long time, growth in the corporate bond market was fairly stagnant.

That said, there have been signs of life in recent years. Over the last six or seven years, a number of regulatory changes have been aimed at deepening debt markets and improving access. At the same time, awareness of bonds as an asset class has increased, and we’re beginning to see more interest from retail investors as well.

This is one area of the market that I would keep an eye on. A deeper corporate bond market doesn’t just create another asset class for investors. More importantly, it gives companies an additional avenue to raise capital, especially those that may not have easy access to bank funding. It also creates competition for banks, which in turn could make borrowing costs more competitive.

It’s still early days, and India’s corporate bond market remains underdeveloped relative to the size of the economy. But if there is one segment of Indian capital markets that has significant room to grow over the coming decade, it is probably this one.

Projected growth of turnover

There’s a funny saying that goes something like this: only two people know what will happen in the future—God and a liar.

The reason I bring that up is that the NSE’s DRHP includes projections for the future growth of turnover across various market segments. Now, if there’s one thing anyone who has spent time in financial markets learns, it’s that predicting the future is extraordinarily difficult. Markets have a habit of humbling anyone who becomes too confident about what comes next.

That said, it’s still interesting to see the assumptions built into these projections. The NSE is clearly quite bullish on the long-term growth of Indian capital markets, whether in cash equities, derivatives, or other segments.

Of course, one could argue that a certain degree of optimism is to be expected. After all, the exchange is in the process of going public and is telling its story to prospective investors. But even with that caveat, the projections offer a useful glimpse into how the NSE itself thinks about the future trajectory of Indian markets.

Whether reality ultimately matches those expectations is another matter entirely. If history teaches us anything, it’s that markets are very good at surprising both optimists and pessimists alike.

The IPO boom

Thanks to the exuberance in the markets after 2020, IPO activity increased spectacularly. Both the number of IPOs and the total amount of capital raised hit record levels, finally giving many of the PE and VC funds that had invested in India after 2012–13 an opportunity to exit.

And that’s actually a good thing.

As an aside, this is one of the reasons why I think people often misread the recent FDI outflows. Many see them as a negative, but a portion of those outflows simply reflects investors exiting through public markets. Investors need deep and liquid capital markets to realize returns on their investments. In fact, the existence of a credible exit mechanism is one of the reasons they’ll continue to invest in the first place. They’ll only come back if they know they can eventually get out.

But anyway, I digress.

The IPO numbers still surprise me. Even after having watched the markets closely through this period, it’s remarkable just how much they’ve increased in such a short span of time.

Nifty vs S&P 500 returns in $

Despite all the noise about rupee depreciation, etc. Over the long term, Indian markets, even in dollar terms, have delivered returns similar to the US market.

Now, despite all the hoopla about foreign investors fleeing India, falling FII ownership, and the endless headlines about foreign money leaving the country, foreign investors still remain a significant part of Indian capital markets.

If you look at the total assets held in the market, foreign investors continue to account for a substantial share of ownership. While their percentage ownership has declined over time, this is often less a story of foreigners abandoning India and more a story of domestic participation growing much faster

NSE is a near monopoly

NSE is a near monopoly with pretty much close to 100% market share in pretty much all the major segments. Although BSE seems to be catching up in the derivative segment, it’s still very, very early days.

NSE’s journey

NSE is really profitable

It’s kind of crazy how profitable NSE is, how much money it makes, and how much of it pays out as dividends.

How exchanges grow

This was an interesting data point from the DRHP. It shows how exchanges in the US and Europe have grown over time, and the number of acquisitions undertaken by some of these exchanges is kind of crazy. It speaks to both the maturity of those markets and the sheer number of opportunities available for consolidation.

In contrast, India has a much smaller exchange ecosystem. Broadly speaking, we have four major exchanges: NSE, BSE, MCX, and MSE. I’m not counting exchanges like NCDEX, IEX, and others, but the point still stands.

What’s also interesting is that SEBI has generally not been very keen on allowing exchanges to become larger through acquisitions. As a result, inorganic growth is not really an option for Indian exchanges in the way it has been for their Western counterparts.

That means Indian exchanges largely have to grow the old-fashioned way: by increasing volumes, adding new products, attracting more participants, and deepening the markets they already operate in. Whether that’s a good thing or a bad thing is a separate debate, but it does make the growth trajectory of Indian exchanges very different from what we’ve seen in the US and Europe.

That’s it for the first edition.

I hope you enjoyed what we’ve put together for you. This is a new format for us, and I fully understand that there are rough edges and plenty of things that can be improved.

And that’s where your end of the bargain begins.

Please send us your feedback, ideas, suggestions, criticisms, and anything else you think can help make this publication better. Unlike many people who say this as a ritual and then promptly ignore what follows, we genuinely do read the comments, emails, and messages we receive.

Points & Figures is still a work in progress, and we’d love your help in shaping what it becomes.

So let us know what you liked, what you didn’t, what you’d like to see more of, and what you’d like us to do differently. We’ll read it all and use it to make future editions more useful, more insightful, and hopefully a lot more fun.

Until next time.