BharatNama

#47: Why does India lose 15% of its power, China 4%?

Jayant Mundhra — Fri, 04 Sep 2026 01:30:38 GMT

I keep coming back to one number, and it unsettles me every time.

For every hundred units of electricity India pushes into its distribution grid, it collects money for only about eighty-five. The rest, nearly one in seven units, vanishes before a single rupee is billed against it.¹

Now here is the part that should stop you. China, running a grid larger than India's across a country every bit as vast, loses barely one unit in twenty-three.³

That is not a rounding error. It is a chasm. And I have become convinced the reason it exists is not the reason most people assume. This is not fundamentally a story about old wires and weak transformers. It is a story about who the state chooses to bill, and who it quietly chooses not to.

What the number actually measures

The metric here is called AT&C loss, aggregate technical and commercial loss. It is the single cleanest gauge of how well a power distribution company, a DISCOM, actually runs. It bundles two very different things into one figure:

Technical loss. The electricity that genuinely disappears as heat in wires and transformers over long distances. This is physics, and every grid on earth pays this tax.
Commercial loss. The electricity that is delivered but never billed, or billed and never collected. This is theft, unmetered supply, and collection failure. This is not physics. This is governance.

India's national AT&C loss came in at 15.04% in FY25, down from 15.97% a year earlier.¹ That is real, and it is the best reading India has posted in over a decade, down from nearly 25% around FY13 and 21.9% as recently as FY21.⁵

There is even a genuine milestone buried in the FY25 data. For the first time ever, India's DISCOMs as a whole turned a profit after tax at the all-India level.⁵ After decades of red ink, that deserves to be said plainly.

But hold the celebration next to the benchmark. China's State Grid, the largest utility on the planet, runs a comprehensive line loss of just 4.4%.³ India, even at its best-ever number, loses more than three times as much of its power. The gap is the whole point of knowing where you stand.

The physics is not the problem

Here is the thing. If India's losses were mostly technical, this would be an engineering bill: heavier conductors, better transformers, high-voltage lines closer to homes. Expensive, but solvable with capital.

They are not mostly technical.

When a well-run grid pays its physics tax, technical losses land somewhere between 3% and 6%. India's pure transmission and distribution loss sits closer to 19%,⁴ and its AT&C loss at 15% is dominated by the commercial layer sitting on top of the physics: power that is stolen, hooked, unmetered, or simply never collected.

That distinction matters enormously, because the two halves of the number ask for completely different fixes:

The technical half wants money and steel.
The commercial half wants meters, audits, and the political will to disconnect a non-paying consumer.

China's 4.4% is essentially all physics. India's 15% is physics plus a vast, off-the-books subsidy that nobody will admit is a subsidy. That is why I keep saying this is an administrative choice, not an infrastructure sentence.

The agricultural alibi

So why does the commercial leak persist, year after year, when it would bankrupt any private business inside a quarter?

The honest answer sits in the fields. After the Green Revolution, states raced to electrify rural India and win the farm vote. Rather than hand farmers transparent cash support, they gave them free or flat-rate, unmetered electricity for irrigation pumps. It was popular, it protected vulnerable growers, and it quietly destroyed the grid's accounting.

Once a large slice of the grid is legally unmetered, a convenient blind spot opens up:

A substation can measure exactly how much power flows out onto a rural feeder.
But if the farm connections downstream are unmetered, nobody can reconcile what was actually consumed against what was billed.
So theft by residential, commercial, and industrial users on that same feeder gets quietly booked as "agricultural consumption".

In other words, unmetered farm supply is not just a subsidy. It is the ledger where a lot of ordinary theft goes to hide. Field research has long found power theft is most intense precisely in India's most agricultural districts, which is exactly what you would expect if the farm category is doing double duty as a cover story.

This state-versus-state, feeder-versus-feeder detective work is the kind of thread I love pulling apart, and it is the sort of thing I go back and forth on with readers every morning in my BharatNama WhatsApp community, where I share a smaller India deep-dive like this daily (t.ly/h2jq1).

Join the BharatNama community

What the leak costs the exchequer

None of this is free. A DISCOM that cannot recover the cost of the power it buys eventually defaults on the generators and lenders it owes, and the hole lands on the state's books.

The balance sheets are still brutal, even after every bailout:

Accumulated losses of India's distribution utilities stood at ₹6.92 lakh crore as of March 2024.²
Their combined net worth was negative, at around ₹1.73 lakh crore.²
Total borrowings climbed to ₹7.52 lakh crore, up from ₹6.84 lakh crore a year earlier.²
Outstanding loans to the distribution sector had reached about ₹5 lakh crore by December 2025, nearly 40% of the entire loan books of state-backed lenders PFC and REC.⁶

The recent good news is real but fragile. The gap between what power costs to supply and what utilities earn, the ACS-ARR gap, has narrowed to just 6 paise a unit in FY25 from 89 paise in FY21.⁶ The trouble is that much of that improvement came from the centre forcing states to clear subsidy backlogs and pay up, not from organic efficiency. The liabilities did not vanish; they moved from the generators to the state-backed banks.

And it drags on sovereign finances directly. Power purchase costs eat roughly three-fourths of a typical state's cost of supply, and interest alone consumes more than 10% of it in states like Rajasthan, Tamil Nadu, Andhra Pradesh, and Jharkhand.⁶

The Chinese playbook: total visibility

Now to the country on the other side of the gap, because the lesson is instructive even where it is not portable.

China's State Grid faced 5% to 15% losses across many regions in the early 2000s, not unlike India today. It has since driven overall grid losses down from 6.8% in 2015 to 4.4% in 2024,³ and it did so less by laying copper than by building an architecture of total measurement.

The core moves, all from State Grid's own reporting:

It wired the grid with over 650mn smart meters, from ultra-high-voltage lines down to individual homes, into a single cloud ledger it calls the "One Data Chain", collecting readings at 15-minute intervals across hundreds of millions of end-users.³
On top of that data it runs an AI "loss engine" that continuously reconciles power dispatched against power billed, flags anomalies that signal theft or tampering, and pushes recovery orders straight to field crews without waiting for managerial sign-off.³
Crucially, it ties the pay and promotion of some 13,000 county-level managers directly to line-loss targets, ranked publicly every month. A manager cannot blame "technical constraints" when the algorithm can see the leak.³

The payoff has been enormous: State Grid says it cut its high-loss lines from 44% of the network to under 1%, saving roughly 236 TWh of electricity, an efficiency gain worth close to $12bn over 2016 to 2024.³

Where India stands, against the world and against itself

Set the numbers side by side and the picture is stark, but the more revealing comparison is India against India.

Against the world:

China (State Grid): 4.4% grid loss.³
Global average, per the World Bank, sits at roughly 8% of output.⁴
India: 15.04% AT&C, and about 19% on pure T&D.¹ ⁴

Against itself, and this is the part that proves the loss is a choice:

Gujarat's DGVCL posted AT&C losses of just 1.31%, among the most efficient utilities anywhere on earth.¹¹
Delhi's private DISCOM Tata Power-DDL ran 5.54% in FY25, down from 53% when it took over the network in 2002.⁷
The national average is 15%.¹
Punjab, meanwhile, was still bleeding around 19% in FY25, having installed zero of its 9mn sanctioned smart meters.⁵

A single country contains both a 1.3% utility and a 19% one. That roughly fifteen-fold spread is not explained by geography or weather, because these utilities serve the same kind of consumers under the same union rules. It is explained by governance.

The private turnaround India already has

This is why I find the "India is just structurally leaky" argument so unconvincing. India has already solved this, at home, more than once.

Delhi is the textbook case. In 2002 the state-owned Delhi Vidyut Board was collapsing, billing barely 57% of the power it distributed, with aggregate losses of 45% to 60%. The distribution business was unbundled and handed to private joint ventures, Tata Power-DDL and the two BSES DISCOMs, under a hard contractual mandate to slash losses.

The results speak for themselves:

Tata Power-DDL: from 53% AT&C in 2002 to 5.54% in FY25.⁷
The BSES DISCOMs settled into the mid-6% range on the same trajectory.
Odisha, which privatised four DISCOMs, saw AT&C losses fall by 11 to 16 percentage points between FY20 and FY25.¹²

Across India, where public and private utilities operate side by side, the private ones run AT&C losses about 8 percentage points lower.⁶ SBICAPS has estimated that privatising DISCOMs in 11 viable states, including Uttar Pradesh, Punjab, Haryana and Bihar, could cut their combined fiscal deficit from about 3.5% to 3.0% of state GDP, with individual states gaining anywhere from 25 to 281 basis points of headroom.⁶ That aligns almost perfectly with the fiscal consolidation the 16th Finance Commission is pushing for.

The mechanism is not magic. Private capital demands cost recovery, cost recovery demands metering, and metering demands that theft finally be treated as theft.

Can India copy the playbook?

So what actually transfers, and what does not?

The technology is the easy part, and India is buying it. The Revamped Distribution Sector Scheme (RDSS), a ₹3.04 lakh crore programme running to FY26, is built on exactly China's logic: meter the feeder, the transformer, and the consumer, then let automated energy audits pinpoint the leak. It mandates about 23 crore prepaid smart meters, of which roughly ₹1.31 lakh crore is earmarked for metering alone.⁸

The catch is that a meter is only as good as the will to act on what it sees:

As of early 2026, only about 5.83 crore smart meters were installed, just 25% of the target, and the deadline has already been pushed back two years.⁸
The rollout is patchy, strong in reform-minded states, stalled where it matters most.
Where consumers or politicians resist, as in Gujarat's billing backlash or Punjab's total non-start, the capital spent on meters earns nothing.

This is the honest limit of the China lesson. State Grid's closed-loop accountability works because a county manager cannot claim a political exemption. India's linemen get assaulted attempting disconnections, and local leaders still demand illegal connections stay live to keep a vote bank happy. The hardware is importable. The political will to disconnect a defaulter is not, and that is the real frontier.

This is the tension I keep circling in these deep-dives, and if you want the daily version, the running thread on India's political economy, do come join the BharatNama WhatsApp community (t.ly/h2jq1).

Join the BharatNama community

What to watch

A few concrete markers over the coming quarters:

The RDSS meter count. Whether installations move meaningfully past 25% before the extended deadline, and whether laggards like Punjab and Tamil Nadu start metering agricultural feeders at all.
Fresh privatisation moves. Watch UP, which has floated plans to privatise its worst DISCOMs, as the next big test of the Delhi-Odisha template.
Whether the FY25 profit holds. The first-ever all-India DISCOM profit⁵ is fragile; if subsidy payments slip again, the ACS-ARR gap will reopen.
The demand backdrop. India just met a record peak of about 256 GW in April 2026,⁹ and crossed 500 GW of installed capacity with more than half from non-fossil sources.¹⁰ A leaky, insolvent distribution grid is the weakest link under all of it.

The one line

India's grid does not leak because it lacks wires; it leaks because, for decades, it has been cheaper for the state to look away than to send the bill.

And well that is it for today's edition. That said, do check out my core WhatsApp community Biz News+ where I share 4-5 deepdives from the world of business, economics & public economics daily: https://t.ly/h2jq1

And if you want to understand where China stands and what it means for India, do check out my companion newsletter, Decoding the Dragon: https://t.ly/t7uhs

And, do check out my work on the following platforms as well: Instagram, LinkedIn and Youtube

Best,
Jayant

References

1. Ministry of Power, 14th Integrated Rating & Ranking of Power Distribution Utilities (FY25 AT&C 15.04%): tndindia.com
2. Power Finance Corporation, Report on Performance of Power Utilities 2023-24: powerline.net.in
3. State Grid Corporation of China, via SASAC, "Stronger, Smarter, Greener": en.sasac.gov.cn
4. World Bank, World Development Indicators, electric power T&D losses: data.worldbank.org
5. The Core, "India's Discoms Turn A Profit In FY25": thecore.in
6. Down To Earth, on SBICAPS DISCOM privatisation study: downtoearth.org.in
7. Tata Power-DDL, FY2024-25 performance (AT&C 5.54%): tatapower-ddl.com
8. The Secretariat, on RDSS smart-meter rollout: thesecretariat.in
9. PIB, India meets all-time peak power demand of ~256 GW: pib.gov.in
10. Enerdata, India hits 500 GW installed, over 50% non-fossil: enerdata.net
11. Business Standard, Integrated Rating, Gujarat and Adani DISCOMs top ratings: business-standard.com
12. Saur Energy, "Can DISCOM Privatisation Solve The Weakest Link": saurenergy.com

#46: Why does heat cost India $194bn in lost work?

Jayant Mundhra — Thu, 03 Sep 2026 01:30:48 GMT

Here is a number that has stayed with me all week. In 2024, India lost roughly 247bn potential hours of work to extreme heat.¹

That is not a projection for some distant, air-conditioned future. That is last year.

Put a price on those hours and it comes to about $194bn of income that simply never got earned.¹ To give that scale, it is larger than the annual budget of almost any Indian state, evaporated into the afternoon air.

And the part that unsettles me most is this: almost none of it shows up in our official numbers.

What we are actually talking about

Let me set the scene, because the structure matters more than any single figure.

About 75% of India's workforce, roughly 380mn people, works in heat-exposed jobs: farming, construction, brick kilns, salt pans, street vending, delivery riding.² This is not a fringe. This is the real economy.

Heat-exposed work produces about half of India's GDP.²
It drives about 30% of India's GDP growth.²

So when a farmer in May slows to half pace, or a rider shelters in the shade for twenty minutes every hour, that is not a human-interest story. That is the supply side of the economy quietly shrinking.

The baseline tells you the direction of travel. India's heat-driven labour losses in 2024 were 124% higher than the average across the 1990s.¹ We have more than doubled the damage in a single generation.

On average, each Indian lost about 419 hours of safe working capacity to heat last year.¹ Think of it as five working weeks, gone, per person, unbooked anywhere.

The drag is now structural, not seasonal

For years, heat was treated as an event. A bad fortnight in May, some headlines, then the monsoon arrives and everyone moves on. That framing is now simply wrong.

The economists have started saying so plainly. Without serious adaptation, heat is projected to shave 0.5 to 1 percentage point off India's potential annual GDP growth, on a recurring basis.³ For an economy straining to grow at 7 to 8%, quietly surrendering up to a full point to weather you can forecast but not yet manage is enormous.

And the forward projections are worse than the present:

McKinsey estimates 2.5% to 4.5% of India's GDP will be at risk by 2030 from lost labour hours, worth $150-250bn.²
The World Bank lands in the same band, up to 4.5% of GDP.³
The ILO projects India will account for 34mn of the world's 80mn heat-related job losses by 2030.⁴

Sit with that last one for a second. India has about 18% of the world's people, yet is projected to absorb roughly 43% of the world's heat-related job losses.⁴ That is the benchmark I would pin to a wall in every ministry.

The losses are concentrated, not spread evenly:

Agriculture took 66% of India's heat-related labour-hour losses in 2024.¹
Construction took another 20%.¹

Agriculture also shows how the damage compounds. In the 2022 heatwave, wheat yields fell 10 to 15%, cutting the crop from about 110mn to 99mn tonnes and pushing the government into an export ban.⁵ Lower labour capacity, lower yields, higher food inflation, and then the real wages of the very workers who lost the hours fall further still. It is a loop, not a straight line.

This kind of feedback loop, where a labour number, a food-inflation number and a real-wage number turn out to be the same story wearing three hats, is exactly the sort of thread I keep pulling apart in the BharatNama WhatsApp community, where I share a smaller India deep-dive like this one with thousands of readers every morning (t.ly/h2jq1).

Join the BharatNama communityJoin the BharatNama community

The frontline, where the body gives out first

The averages hide brutal margins. Between 80 and 90% of Indian workers are informal, with no paid breaks, no cooling, no health cover, no financial buffer for a lost day.⁶

Study the extreme trades and you see how thin the survivability line has become. In India's salt pans and brick kilns, researchers record wet-bulb globe temperatures that sit above the internationally accepted safe threshold for the workload, day after day, with many workers surviving a full shift on barely a litre of water.⁶ At that point the heart is spending its energy cooling the body rather than doing the job, so output falls even when the worker is standing right there.

The gendered version of this is worse, and more current than most people realise. A 2026 assessment found extreme heat now costs women informal workers about $57bn a year globally, with Indian cities among the hardest hit.⁷ Women make up a fifth of urban heat deaths in the cities studied, while carrying both outdoor paid work and indoor unpaid work over a stove.⁷ Roughly 81% of India's working women are in informal jobs, so they sit at the sharp end of every number above.⁸

And this is spilling into the new economy, not just the old one. India logged more than 40,000 suspected heatstroke cases in 2024, and a growing share of the exposed are gig and platform workers, the delivery riders and cab drivers we now treat as invisible infrastructure.⁹

The blind spot, a brand-new GDP series that still cannot see heat

Here is the part I find genuinely remarkable, and it is the heart of today's edition.

On 27 February 2026, India rolled out a brand-new GDP series, shifting the base year to 2022-23 and finally adopting "double deflation", which strips price effects out of inputs and outputs separately for a cleaner read on real growth.¹⁰ It is a real, overdue upgrade.

Yet even this rebuilt machine cannot see heat. And the reason is subtle.

The output of the informal half of the economy is estimated by leaning on the Periodic Labour Force Survey, the PLFS, which is the government's main employment survey. The PLFS captures "hours actually worked" in a reference week.

But "hours worked" is a measure of presence and duration. It says nothing about intensity or pace.

A construction worker on site for nine hours, taking undocumented micro-breaks in the shade to avoid collapse, still reports a full day.
A field labourer working at 30% slower pace in 44-degree heat still logs the same hours.

The national accounts then multiply those intact-looking hours by a baseline productivity weight to build value added. So the model quietly assumes optimal output was produced. Heat rarely destroys the job outright, which is what a survey would catch as unemployment. It destroys the productivity inside the job, which the survey cannot catch at all.

The result is a kind of phantom GDP: an economy that records a full day of work while the worker doing it produced markedly less. We have upgraded the ruler, but it still cannot measure the thing that is shrinking.

The law just got weaker, not stronger

While the statistics miss the problem, the regulation is actively retreating from it. This is the twist that should worry a policymaker most.

On 21 November 2025, India brought its four new labour codes into force, consolidating a sprawl of older laws.¹¹ The Occupational Safety, Health and Working Conditions Code is now the governing law for workplace safety.

Under the old Factories Act of 1948, states like Maharashtra, Tamil Nadu and Gujarat had notified hard wet-bulb limits, often 30°C, beyond which a factory was legally unsafe.⁹
The new OSH Code specifies no temperature, humidity or ventilation threshold at all, even for formal factories, and centralises the power to set such standards.¹²
Gig and platform workers, the group most exposed to outdoor heat, are left out of the OSH Code entirely and parked under the Social Security Code, which carries no occupational-hazard mandate.⁹

In plain terms, we replaced enforceable, inspectable numbers with a vague duty to keep conditions "adequate", and handed heat safety back to employer discretion just as the heat got worse.¹² Unions from the app-based transport and warehouse workforce are now demanding what the law removed: paid cooling breaks of 20 minutes every two hours on heatwave days, a heat-welfare fund, and protection from being penalised by an algorithm for stopping in the sun.⁹

Where India stands, against the world and against itself

Benchmark it, because the gaps are the whole point.

India versus the world:

India's 247bn labour hours lost in 2024 make it one of the most heat-exposed large economies on earth, and its 419 hours lost per person is far above the level in temperate advanced economies.¹
Half of India's GDP comes from heat-exposed work; in a rich economy that share is a fraction of it, which is why the same heatwave is a macro event here and a nuisance there.²
India is set to carry 43% of the world's heat-linked job losses with 18% of its people.⁴

India versus India, which is the sharper mirror:

Kerala has used its own Minimum Wages Act to mandate a rest ban for outdoor workers through the 12pm to 3pm window on hot days, and has issued guidance covering delivery riders.⁹
At the other end, a 2024 assessment of nine high-risk cities found several, including Gwalior and Ludhiana, running heat response with weak or absent long-term plans, while Haryana maps industrial hubs like Gurugram and Faridabad into its highest heat-risk zone.¹³
The nine cities studied, home to just over 11% of urban India, almost all did the cheap emergency things, water and shade and hospital readiness, and almost none did the structural things: cooling for workers, greening, heat insurance.¹³

The gap between what Kerala mandates and what Gwalior does not is exactly the state-versus-state spread we go back and forth on in the BharatNama group, where a quieter India deep-dive lands every single day (t.ly/h2jq1).

Join the BharatNama communityJoin the BharatNama community

What India can learn

The good news is that a cheap, working model already exists inside India. After a 2010 heatwave killed an estimated 1,344 people in a single week, Ahmedabad built South Asia's first Heat Action Plan in 2013: early warning, public outreach, trained hospitals, water and cooling points. It is credited with avoiding roughly 1,100 deaths a year.¹⁴

But Ahmedabad saves lives; it does not protect output, and it does not fix the law. For that, look abroad at the one thing India just deleted: enforceable thresholds.

Spain, after workers died in the fields, amended its safety law in 2023 so that authorities can prohibit outdoor work outright when the weather agency issues a red or orange heat alert.¹⁵
The United States has proposed a federal heat rule covering about 36mn workers, tying mandatory water, shade and paid rest breaks to a defined heat-index trigger.¹⁶

The transferable lesson is not "copy Spain". India's federal structure, informal labour and thin state capacity make a blanket ban hard. The portable idea is narrower and cheaper: a clear, humidity-aware temperature trigger that automatically switches on paid rest, water and shade, written into the OSH Code rules and extended to gig workers. India already knows how to run the warning system. It has simply chosen, for now, not to make the protection binding.

What to watch

Whether the Central Government notifies actual wet-bulb or heat-index thresholds in the OSH Code rules, or leaves "adequate" as the standard.
Whether MoSPI or the NSO builds any heat or productivity-intensity adjustment into the new GVA method, rather than trusting hours-worked alone.
Whether more states follow Kerala with legally binding rest-hour orders before the 2026 summer.
The next Lancet Countdown print, to see whether the 247bn-hour loss was a peak or a new floor.

Heat is no longer a weather story for India. It is a labour-supply story, an inflation story and a measurement story, all at once, and right now we are managing the one part of it that is easiest to count.

And if you want to understand where China stands and what it means for India, do check out my companion newsletter, Decoding the Dragon: https://t.ly/t7uhs

And, do check out my work on the following platforms as well: Instagram, LinkedIn and Youtube

Best,
Jayant

References

Lancet Countdown 2025, India data sheet, lancetcountdown.org
McKinsey Global Institute, Will India get too hot to work?, mckinsey.com
Business Standard, How extreme summers are starting to weigh on India's economic growth, business-standard.com
India.com, Around 34 million job losses in India due to severe heat waves by 2030 (World Bank, ILO), india.com
Likely impacts of the 2022 heatwave on India's wheat production, IOPscience, iopscience.iop.org
Down To Earth, Sweat for survival: how long can India's informal labour bear the heat, downtoearth.org.in
Ground Report, Extreme heat costs women informal workers $57 billion a year, groundreport.in
Climate Resilience Center, The scorching divide: how extreme heat inflames gender inequalities, onebillionresilient.org
Mongabay India, Workers' groups are pushing for legal protections against deadly heat, india.mongabay.com
Press Information Bureau, New series of GDP estimates with base year 2022-23, pib.gov.in
King Stubb & Kasiva, All four labour codes enforced from 21 November 2025, ksandk.com
LiveLaw, Heat as an occupational hazard: enforcement gap in India's labour and climate law, livelaw.in
Sustainable Futures Collaborative, Is India Ready for a Warming World?, sustainablefutures.org
PreventionWeb, India: new evidence shows pioneering heat action plan helps millions avoid killer heat, preventionweb.net
Idealista, Outdoor work ban due to extreme temperatures enters into force in Spain, idealista.com
US Department of Labor, OSHA proposed rule to protect indoor and outdoor workers from extreme heat, dol.gov

#45: Why India exports just $5bn of goods online?

Jayant Mundhra — Wed, 02 Sep 2026 01:30:49 GMT

Here is a number that should bother anyone who believes in the "Make in India, sell to the world" story. India sells roughly $5bn of goods online to the rest of the planet each year¹. China sells close to $300bn². That is a gap of nearly sixty to one, between two countries that both call themselves factories to the world.

This is quite unnerving for me. We have the sellers, the products, and the ambition. What we have not had is the machine that moves a handmade lamp from Moradabad to a doorstep in Munich without the maker drowning in shipping forms.

On 23 July 2026, India tried to buy that machine off the shelf. It changed a rule it had guarded for almost a decade³.

What actually changed

The government issued Press Note 3 (2026 Series), and it does one surgical thing. It allows 100% foreign direct investment in the inventory-based model of e-commerce, but exclusively for exporting goods made in India³.

To see why that is a big deal, you need the one distinction that has governed Indian e-commerce since 2016.

The marketplace model. The platform is a matchmaker. It connects an independent seller to a buyer, takes a commission, and never owns the goods. Foreign money has been allowed here, up to 100%, for years⁴.
The inventory model. The platform buys the goods itself, owns them on its balance sheet, and sells them directly. For any company with foreign equity, this was banned⁴.

That ban was not an accident. It was a wall built to protect millions of kirana stores from well-funded foreign giants buying stock in bulk and undercutting everyone on price⁴. Press Note 3 of 2016 drew that wall, and it held.

Press Note 3 of 2026 cuts a single door into it. Foreign platforms can now own inventory, on one condition: the goods must be exported, not sold to an Indian consumer³. The compliance is tied to the Foreign Trade Policy 2023 and FEMA rules, so it sits inside the existing export plumbing rather than beside it³.

How the new machine is meant to work

Strip away the jargon and the mechanism is elegant.

A foreign platform can now set up a separate, ring-fenced Indian export company. That company behaves like a principal buyer, not a middleman.

It buys textiles, handicrafts, auto parts or wellness products outright, in rupees, from an Indian factory or an artisan cluster.
The maker gets paid on delivery inside India, so the crushing wait for cross-border payments and the currency risk simply vanish for them.
The platform then swallows all the friction: freight, customs, foreign warehousing, marketing abroad, and returns from a customer in Ohio or Abu Dhabi.

This is the interesting bit. The policy quietly separates where the sale happens from who owns the stock. The Indian maker gets a domestic transaction. The platform gets a global one. That is the whole trick.

And the timing is not random. The government wants total exports to hit $2 trillion by 2030-31⁵, with e-commerce exports alone climbing from ~$5bn today to a $200bn target, which it frames as roughly a tenth of the projected global online-export market¹. You do not get from $5bn to $200bn by asking small sellers to fill out more shipping labels. You get there by handing the job to whoever already runs a planet-sized logistics network.

Why India is copying China, on purpose

The honest way to read this policy is as India studying its rival's homework.

China did not build the world's largest cross-border trade engine through millions of tiny sellers each battling customs alone. It did it through the inventory model. Platforms like Shein and Temu buy stock outright from dense manufacturing clusters, predict demand with data, pool freight, and ship at a scale that collapses the per-unit cost.

The results are not close.

China's cross-border e-commerce trade was worth ~$375bn in its latest full year, with the export half alone at ~$300bn, growing nearly 17% a year².
India's online exports sit at ~$5bn¹.

This state-versus-rival gap is exactly the kind of thing I keep pulling apart, and it is the sort of thread we go back and forth on in the BharatNama WhatsApp community, where I share a smaller India deep-dive like this with thousands of readers every morning (t.ly/h2jq1).

Join the BharatNama community

There is a second reason the timing works in India's favour. As global brands hunt for a "China plus one" supplier to de-risk their supply chains, India suddenly has both the factories and, now, the legal permission for a foreign platform to buy from them at scale. The door is open exactly when buyers are looking for a second room.

The giants have already put numbers on it

You can measure how much foreign platforms wanted this by how fast they moved once it landed.

Amazon runs its Global Selling programme since 2015. It has onboarded over 200,000 Indian exporters across 28 states and 200-plus cities, and it crossed $20bn in cumulative exports from India, hitting its 2025 goal early. Its new target is $80bn in cumulative exports by 2030⁶.
Small towns are already in the data. Karur in Tamil Nadu, a home-textiles cluster, did about $147mn of exports through the platform in a single year⁶.
Walmart, which owns Flipkart, has committed to sourcing $10bn of Indian goods every year by 2027⁷. It has already sourced around $30bn from India over two decades⁷.

The knock-on effect is showing up in warehouses before the goods even move. India's industrial and warehousing demand jumped 22% year on year in the first quarter of 2026, a four-year high, with about 11mn sq ft leased across the top eight cities⁸.

Delhi NCR took the largest share at 28%, followed by Chennai at 21%⁸.
E-commerce and automobile firms together drove 32% of the leasing, and third-party logistics players nearly a third of it, with their activity running 1.8 times the level of a year earlier⁸.

Investor certainty is the quiet win here. As one tax adviser at EY put it, the note "removed uncertainty" and gave foreign investors policy predictability, because the old rules were silent on whether inventory-led exports were even allowed⁹.

Where India stands: against the world, and against itself

BharatNama's job is to place India next to something. Two comparisons matter here.

India against the world (online exports, latest available):

China: ~$300bn, export cross-border e-commerce²
India: ~$5bn¹
The gap: ~60 times, which is the single number this whole policy exists to shrink

India against itself:

The build-out is clustering, not spreading. Delhi NCR (28%) and Chennai (21%) alone soaked up nearly half of Q1 warehousing demand⁸, while the tier-two and tier-three makers the policy is meant to lift sit far from those hubs.
India's overall FDI held up well through this, with inflows of about $59bn in FY26, up 18%¹⁰, so the country is not short of foreign capital. It has been short of the kind of foreign operation that owns stock and moves it out.

The story India tells itself is of a lamp-maker in Moradabad reaching Munich. The infrastructure, for now, is being poured in Gurugram and Sriperumbudur. Both things are true at once, and BharatNama exists to hold them together rather than pick one.

What India can actually learn from China

The temptation is to say "India should just copy China". The reality is more careful.

What China actually did was sequence three things:

It set up over 165 cross-border e-commerce pilot zones with simplified customs and tax rebates.
It built dense bonded warehousing so stock could sit near ports, cleared for export.
It let platforms own inventory and integrate logistics end to end, from factory to foreign doorstep.

Press Note 3 of 2026 is India reaching for the third lever. The first two are the harder, less glamorous ones, and they are where the real work sits.

What transfers: the inventory-led export model, bonded warehousing near ports, and a single foreign platform absorbing logistics risk for a scattered supplier base. This policy makes all three legal.
What does not transfer easily: China's manufacturing density and logistics velocity took two decades and enormous state spending to build. India cannot legislate that into existence. A rule change opens the lane; it does not lay the road.

The lesson is that the FDI note is necessary but nowhere near sufficient. Without cheaper freight, faster customs and export incentives that match China's, India gets the permission slip without the performance.

The risk nobody at DPIIT can fully price

Now let me be honest about the other side, because a BharatNama edition that only cheers is not doing its job.

The sharpest critique comes from the Global Trade Research Initiative. Its founder, Ajay Srivastava, makes two arguments worth sitting with¹¹.

India gave away a negotiating chip for free. Access to India's vast digital and retail market is one of the biggest cards India holds in trade talks, especially with the United States. Srivastava argues this concession should have been traded inside a deal for lower US tariffs on Indian goods, not handed over unilaterally in advance¹¹.
The inventory model hands the platform the whip. Once a platform owns the goods rather than just listing them, it has every incentive to push its own stock, read the data on which Indian products sell best, and either source those directly or launch cheaper own-label versions. Independent sellers and traditional export houses get squeezed out of the middle¹¹.

Then there is the fear on the ground. Traders' bodies, led by figures like CAIT's Praveen Khandelwal, call this a Trojan horse¹². Their worry is simple: once foreign giants are allowed to own inventory locally "for exports", how do you stop that stock leaking into the domestic market, and how long before the lobbying to sell it to Indians too becomes irresistible¹²?

CAIT has flagged that the current rules do not even mandate physical audits to check that export inventory stays separate from domestic stock¹².
Some in the trade body have gone further and demanded an outright rollback, arguing the change favours foreign platforms over Indian retailers and small exporters².

This is where I keep landing, and it is the sort of policy trade-off we argue out in the BharatNama community rather than pretend is settled (t.ly/h2jq1). The enforcement problem is real: keeping one identical product legal-for-export and another legal-for-domestic, inside the same fungible supply chain, is an administrative nightmare. It works only if the customs-bonded, physically segregated, digitally tracked firewall actually holds. On paper it can. In a system this large, holding it is the whole game.

Join the BharatNama community

What to watch

The fine print from DGFT. Watch for the operational directive on physical segregation, customs-bonded warehousing and GST refunds. That document decides whether the firewall is real or rhetorical.
The first inventory-led export hubs. Track whether Amazon and Flipkart actually break ground on dedicated export warehouses near ports, and where.
Any leakage case. The first proven instance of "export" stock turning up in the domestic market will be the political flashpoint that decides the policy's future.
The US trade talks. Watch whether India extracts anything in return, or whether GTRI's "gave it away for free" warning proves right.

The one line

India has handed the world's biggest platforms the keys to its export engine, and the only thing standing between a genuine trade breakthrough and a backdoor into the kirana store is a firewall that has never been tested at this scale.

And if you want to understand where China stands and what it means for India, do check out my companion newsletter, Decoding the Dragon: https://t.ly/t7uhs

And, do check out my work on the following platforms as well: Instagram, LinkedIn and Youtube

Best,
Jayant

References

1. Outlook Business, India sets $200bn e-commerce exports target by 2030: outlookbusiness.com
2. Xinhua, China's cross-border e-commerce exports reach new high: english.news.cn
3. Business Today, Govt permits FDI in inventory-based e-commerce for exports: businesstoday.in
4. Vajiram & Ravi, FDI in e-commerce, government liberalises policy: vajiramandravi.com
5. IBEF, India's trade strategy to achieve US$2 trillion exports by 2030: ibef.org
6. Amazon, India e-commerce exports top $20bn, sets $80bn goal for 2030: aboutamazon.in
7. Sourcing Journal, Walmart commits to sourcing $10bn in Indian exports by 2027: sourcingjournal.com
8. Colliers via NBM&CW, 3PL firms drive one-third of warehousing demand in Q1 2026: nbmcw.com
9. Medianama, India eases e-commerce FDI rules, allows foreign-funded firms to own export inventory: medianama.com
10. India Briefing, India FDI inflows reach US$58.85bn in FY 2025-26: india-briefing.com
11. Outlook Business, FDI easing could pave way for wider opening of online retail: GTRI: outlookbusiness.com
12. Business Standard, Trade body calls for immediate rollback of new FDI export policy: business-standard.com

#44: Why is INDIA building a 2mn-tonne urea plant in RUSSIA?

Jayant Mundhra — Tue, 01 Sep 2026 01:30:42 GMT

I think India just made one of the most quietly revealing decisions of the decade, and almost nobody is talking about why.

We are going to build a giant urea plant. A 2mn-tonne one. But not in India. In Russia.¹

In December 2025, three Indian fertiliser firms, the state-owned RCF and NFL and the cooperative-sector Indian Potash, signed a memorandum with Russia's Uralchem to build a roughly $2bn urea complex at Tolyatti, in the Samara region.¹ ² The ammonia will come from Togliattiazot next door, and the urea is meant to sail almost entirely to Indian farms.³

Here is what stopped me. India is the largest urea importer on earth. We are also, this very year, spending public money to become self-sufficient in urea at home.⁴ And yet the single biggest new bet we have placed is a plant on another country's soil.

That is not a contradiction. Once you see why, it tells you something uncomfortable about where India actually stands.

Let me set the scene with the numbers, because they are the whole story.

India consumes about 40mn tonnes of urea a year.⁴
We produce roughly 30mn tonnes at home, from 33-odd plants.⁴
The remaining ~10mn tonnes, about a quarter of what we use, is imported.⁴

Urea matters more here than anywhere else. Over 76% of all the fertiliser India made in 2024-25 was urea.⁵ It is the cheap nitrogen that keeps the wheat and rice coming. It is also politically radioactive, which is why a 45-kg bag has been frozen at ₹242 for the farmer for years.⁴

The catch sits under all of it. Urea is, in plain terms, solidified natural gas. You make ammonia from gas, then turn ammonia into urea, and gas is both the raw material and the fuel. It works out to 60-80% of the cost of every tonne.⁴ ⁶

So the real question behind fertiliser security is not "where do we build the factory". It is "where is the gas".

And that is exactly where India is exposed.

We do not have enough gas of our own. Our fertiliser plants lean heavily on imported LNG, which has run as high as 63% of the sector's gas use.⁶ Imported gas is dear and violently unstable.

In early 2022, spot LNG spiked to around $36 per unit.⁶
Even in calmer times, delivered gas to Indian urea plants sits near $10-15 per unit.⁶
India imports roughly half of the LNG it burns, much of it through the Strait of Hormuz.⁴

Russia is the mirror image. It sits on the largest gas reserves on the planet, its regulated gas costs low single digits, and since Europe stopped buying, a lot of that gas has nowhere to go. Stranded gas is cheap gas.

Now the Tolyatti choice makes sense. Togliattiazot is one of the world's biggest ammonia makers, and for decades it exported through a pipeline to the Black Sea that ran via Ukraine. That pipeline has been shut since the war began. So Russia is sitting on world-scale, low-cost ammonia with no easy buyer, and India walks in as exactly that buyer.

Read plainly, the deal is this. India is not really importing urea. India is importing Russia's cheap gas, in a solid form our farmers can spread on a field.

This shift in who feeds India is the kind of thread I keep pulling at in the BharatNama WhatsApp group, where I share a smaller India deep-dive like this with thousands of readers every morning (t.ly/h2jq1).

Join the BharatNama community

To feel how far we have travelled, rewind the tape.

At Independence, India barely made any nitrogen fertiliser and imported almost all of it. The Green Revolution of the late 1960s was built on subsidised urea, and India spent the next fifty years building a domestic industry that now makes ~30mn tonnes a year. That is a genuine achievement, and it is why the import share is a quarter, not three-quarters.

But the arc is now bending the wrong way. Because gas is the binding constraint, and India's gas problem has not been solved, our import dependence for fertilisers overall is projected to climb past 41% in 2025-26.⁵

The map of who supplies us has also flipped in just a few years, and this is the part that should make policymakers sit up:

The Gulf, once two-thirds of our urea, has fallen from about 75% of sourcing to 38%.⁵
Russia and China together now account for nearly 40% of India's urea.⁵
China's slice alone jumped from 1.5% in 2024-25 to about 20% in 2025-26.⁵

And this is not new for Russia. Its share of India's overall fertiliser imports has climbed from under 8% in 2017-18 to around 27% by 2023-24, one of the fastest supplier shifts in the trade.¹¹

We have quietly swapped dependence on the Gulf for dependence on two countries with whom the strategic weather can change fast. China, remember, curbs its own urea exports whenever domestic prices threaten to rise, precisely to shield its farmers.⁷ A supplier that shuts the tap to protect its own is not a supplier you can build food security on. Which is exactly why a plant India part-owns starts to look less strange.

Now set India against the country that actually solved this. China.

China is largely self-sufficient in urea. It will make about 77mn tonnes in 2026 against demand near 66mn, and it exports the surplus.⁷ More importantly, it makes its urea a completely different way. About 78% of Chinese urea is made from coal, not gas.⁷

That one choice changes everything, because China has cheap domestic coal and does not have to import expensive gas. The result, in the same world market where India pays through the nose:

Chinese urea sold at roughly $260 a tonne in early 2026.⁷
The global benchmark at the same moment was around $740 a tonne.⁷
That is close to a 3x cost advantage, sitting inside China's borders.⁷

So the benchmark is stark. China insulated its farmers by building self-sufficiency on a feedstock it owns. India cannot copy that, because we have neither China's coal-to-chemicals base nor cheap gas of our own. Our version of the same idea is to go and rent someone else's cheap feedstock. That is what Tolyatti is.

There is a second gap, and this one is inside India, not across a border.

Look at who pays for a bag of urea. The farmer pays ₹242. The government pays the rest, and the rest is enormous: the true cost of a bag runs ₹2,200 to ₹4,000, so the state absorbs the difference through subsidy.⁴ For 2025-26, the fertiliser subsidy is budgeted at ₹1.71 lakh crore.⁴

That gap does something quietly destructive. Because urea is kept artificially cheap while phosphorus and potassium are not, farmers pour on nitrogen. India's N:P:K use ratio has skewed to about 9.8:3:1, against a recommended 4:2:1.⁴ We are subsidising soil degradation.

And here is the sting for the Russia plant. Cheaper imported urea eases the subsidy bill in the short run, which is a real fiscal win. But it does nothing about the underlying distortion that makes us over-consume urea in the first place. We are treating the symptom, expensive gas, and leaving the disease, mispriced fertiliser, untouched.

So what can India actually learn here, given we cannot become China?

The honest answer is that self-sufficiency in the factory is not the same as self-sufficiency in energy, and India keeps conflating the two. In July 2026 the Cabinet cleared a new urea policy, NIPU-2026, to build 8 to 9 new plants at home and add about 10mn tonnes of capacity, on a 12-16% return-on-equity band.⁴ ⁸ It is sensible industrial policy. But every one of those domestic plants will still run on gas India mostly has to import.⁴

That is the deeper lesson from both China and Russia. The country that wins fertiliser security is the one that owns its feedstock. China owns coal. Russia owns gas. India owns neither at the scale it needs. So India is doing two rational things at once:

Building at home for physical resilience, so a blockade or a sanction cannot cut the supply entirely.
Building in Russia for cheap economics, renting the feedstock advantage it cannot manufacture domestically.

There is even a name being floated for the second idea, "Making in Russia for India".⁹ It flips Make in India on its head: put the energy-hungry step where the energy is cheap, and ship the finished product home. For a product that is 80% embodied gas, that is not heresy. It might be the only honest arithmetic.

The precedent, worth remembering, is Oman. For two decades the OMIFCO plant, part-owned by IFFCO and Kribhco, has shipped its urea to India under a buy-back deal.¹⁰ Tolyatti is that model again, just bigger, and inside a sanctioned economy. This is exactly the sort of state-versus-state and country-versus-country contrast I dig into daily with the BharatNama community (t.ly/h2jq1), because the numbers rarely say what the headline says.

Join the BharatNama community

What I would watch over the next few quarters:

Whether Tolyatti clears its final investment decision. As of now it is still a memorandum, and financing plus sourcing equipment inside a sanctioned Russia is genuinely hard. "Being built" is, for now, "agreed to be built".¹ ²
The sanctions exposure. Uralchem's ownership has faced Western sanctions, and state-owned RCF and NFL taking equity carries real secondary-sanctions risk.¹
How the payment is settled. Watch for a rupee or rouble mechanism that quietly uses the rupee balances Russia has struggled to spend from our discounted-oil trade.
Whether NIPU-2026 plants actually break ground, or whether cheap imported urea makes domestic capacity look unattractive all over again.⁴ ⁸

India's fertiliser problem was never really about factories. It was always about gas, and we are answering a gas problem by borrowing someone else's cheapest molecule and calling it food security.

And if you want to understand where China stands and what it means for India, do check out my companion newsletter, Decoding the Dragon: https://t.ly/t7uhs

And, do check out my work on the following platforms as well: Instagram, LinkedIn and Youtube

Best,
Jayant

References

Business Today, "Uralchem signs MoU with RCF, NFL, IPL to build major urea plant in Russia": link
Sputnik India, "India-Russia 2 MT Urea Mega Plant to Be Ready in 2 Years": link
Argus Media, "Russia's Uralchem agrees urea JV deal with Indian firms": link
Business Standard, "New urea policy tackles imports, but ignores fertiliser subsidy distortions": link
East Asia Forum, "India's fertiliser dependence leaves its agriculture vulnerable": link
IEEFA, "LNG in India's fertiliser sector: a trillion-rupee subsidy burden": link
The Daily Star / Reuters, "China's coal-based urea shields farmers from global fertiliser turmoil": link
Drishti IAS, "National Investment Policy for Urea-2026 (NIPU-2026)": link
Vision IAS / Economic Times, "Making in Russia for India": link
IFFCO, Oman India Fertiliser Company (OMIFCO): link
Business Standard, "Russia's share in India's fertiliser imports rose to 27% in 2023-24": link

#43: Why does 19% of Kerala's budget now go to elders?

Jayant Mundhra — Sat, 29 Aug 2026 01:31:33 GMT

The state that reached India's future first

I keep coming back to one number, and it unsettles me a little every time. In 2026, 18.7% of Kerala's people are over the age of 60.¹ For the country as a whole, that figure is roughly 11%.²

So Kerala is not a slightly older version of India. It is a different demographic country altogether, living a future the rest of us have only read about in reports.

And this year, Kerala did something no Indian state has ever done. It stopped treating old age as a corner of the social justice department and built an entire machinery around it: a dedicated Department of Senior Citizens' Welfare, a statutory Elderly Commission, and a separate, tracked Elderly Budget worth ₹46,236 crore, about 19.07% of the entire state budget.³

That is the number in the title, and it is the one I want to sit with you today. A state has just decided to route nearly a fifth of everything it spends towards its old. Not by accident. On purpose, in a printed budget document, the first of its kind in India.

What Kerala actually built

The pieces landed fast, and in a specific order. The Left government presented the Elderly Budget in January 2026.⁴ Then Kerala voted, the Congress-led UDF swept in with 102 of 140 seats, and the new government moved quickly on the institutions.⁵

Three things now exist that did not a year ago:

A Department of Senior Citizens' Welfare, created on 20 May 2026 via Government Order 63/2026/GAD, the first dedicated department for the elderly set up by any state government in India.⁶ It is the nodal agency, meant to pull eldercare out of the silos it was scattered across (health, housing, local bodies) and run it as one.⁷
The Kerala State Elderly Commission, set up under the Kerala State Elderly Commission Act, 2025 (Act 5 of 2025), a statutory, quasi-judicial body with the teeth to intervene on neglect, eviction, and economic abuse of the old.⁸
The Elderly Budget itself, a document that forces every department to show exactly how much of its money reaches seniors, separating what the state chooses to spend from what it is legally bound to.⁴

Here is why this matters beyond Kerala. The traditional Indian safety net for old age was never the state. It was the family. Kerala is the first place in India where that contract has visibly frayed enough that the government has had to step in and formalise it. This is the demographic future of Tamil Nadu, of Punjab, of Himachal, arriving early in one corner of the country.

How Kerala got old so fast

The speed is the part that gets lost. This was not gradual.

In 1961, just 5.1% of Kerala was over 60, slightly below the national average of the time.¹
By 2001, it was 10.5%, already well ahead of India.
By 2021 it hit 16.5%, and by 2026 it is 18.7%.¹
The state's own projections put it at 22.8% by 2036, when Kerala will have around 8.4mn people over 60.⁹

The engine was Kerala's own success. Early public health, near-total female literacy, low fertility and rising life expectancy are exactly the things development is supposed to deliver. Kerala delivered them decades before anyone else, and the demographic bill for that head start is now arriving.

The dependency maths is where it bites. The old-age dependency ratio, the number of elderly per 100 working-age people, is climbing towards 34.3% in Kerala over the coming years, against a national ratio less than half that.⁹ In plain terms, Kerala is heading towards a world where one in three working-age adults is effectively supporting an elderly dependent.

This state-versus-state gap is exactly the kind of thread I love pulling apart, and it is the sort of thing we go back and forth on every morning in the BharatNama WhatsApp community, where I share a smaller India deep-dive like this with thousands of readers (t.ly/h2jq1).

👉 Join the BharatNama community

Where Kerala stands: against India, and against itself

Two comparisons make this real. Both matter, because a single number for Kerala tells you almost nothing until you set it beside something.

Kerala against the rest of India. On the 2021 official state shares of the 60-plus population, the spread inside one country is enormous:¹⁰

Kerala: 16.5% (the oldest, and now 18.7% in 2026)
Tamil Nadu: 13.6%
Himachal Pradesh: 13.1%
Punjab: 12.6%
Andhra Pradesh: 12.4%
Bihar: 7.7% (the youngest large state)

So Kerala's elderly share is more than double Bihar's. India is not ageing as one country. It is a young north and centre carrying an old, prosperous south and a greying Punjab, and the national average of ~11% hides that split entirely.

Kerala against the world. Put Kerala on the global map and it stops looking like an Indian state and starts looking like East Asia:

India (60+): ~11%²
Kerala (60+): 18.7%¹
China (60+): 22.0%, around 310mn people at end-2024¹¹
Japan (65+): a record 29.4%, the oldest society on earth¹²

Kerala today sits closer to China than to India, and it is walking the same road Japan walked a generation ago. The difference, and this is the uncomfortable bit, is that Japan grew rich before it grew old. Kerala is growing old on a middle-income budget, with borrowing limits set in Delhi.

The empty houses behind the numbers

Ageing in Kerala does not look like ageing in Japan, and the reason is migration.

Kerala exports its young. The Kerala Migration Survey 2023 counted 2.2mn emigrants, most of them working-age, alongside 1.8mn return migrants and remittances of about ₹2.17 lakh crore flowing home in a single year.¹³ Student emigration has doubled, with young people now leaving as early as 17.¹⁴

The money is a lifeline. The social cost falls on the people left behind:

The share of older adults living alone or only with a spouse has climbed sharply, the "empty nest" now a mass phenomenon.
Ageing here is heavily feminised. Women outlive men by years, and a large majority of Kerala's elderly women are widows, often with no service pension, few property rights and thin savings.
Loneliness and psychological distress rise where the adult children have gone, precisely because the informal caregiver, the family, is on a plane to the Gulf or Canada.

Why the budget is mostly a trap

Now to the number in the title, and why 19% of a budget going to the elderly is less generous than it sounds.

The Elderly Budget is split into two very different things. There is Plan spending, the developmental money that builds new services, day-care centres, mobile clinics, the "silver economy". And there is Non-Plan spending, the statutory money the state is legally bound to pay whether it likes it or not.

The split is brutal:⁴

Total Elderly Budget: ₹46,236 crore.
Of which Non-Plan (mostly pensions): about ₹44,895 crore, roughly 97% of the whole thing.
Leaving the State Plan for elderly welfare at only around ₹1,342 crore.

Read that again. For every ₹100 Kerala routes to the old, roughly ₹97 is locked in pension cheques it cannot legally stop paying, and barely ₹3 is left to actually build anything new. Service pensions for retired staff alone run to tens of thousands of crores.

This is the crowding-out problem in one state, ahead of schedule. Committing a fifth of the budget to the elderly, almost all of it non-negotiable, is a fifth that cannot go to schools, jobs, or the young. And the pressure is only rising: the discontinuation of the revenue-deficit grant and a tighter 16th Finance Commission settlement are squeezing Kerala's fiscal room just as these bills come due.

The new government has still promised more. Its flagship Oommen Chandy Free Health Insurance Scheme offers cover of up to ₹25 lakh per family, with an initial ₹10 crore to get it going.¹⁵ The intent is right, because out-of-pocket costs are the real killer here. But funding rich health cover for an ageing population, on a shrinking tax base, is the exact squeeze the whole edition is about.

The health bill nobody budgeted for

Ageing does not just mean more pensioners. It means a different, far more expensive kind of patient.

Over 70% of Kerala's elderly live with at least one chronic disease.
Multimorbidity, two or more chronic conditions at once, affects 52.2% of older adults in Kerala, against a national average of 23.3%.¹⁶ That is more than double the country's rate.
Kerala spends more on health than any Indian state, yet a large majority of health spending is still out-of-pocket, and private hospitals dominate the specialised care the old need most.

So you get a cruel combination: the sickest elderly population in India, the highest treatment intensity, and a payment system that still lands much of the bill on the family. This is why the health-insurance promise is politically irresistible and fiscally terrifying at the same time.

What India can actually learn

The good news is that other societies have been here, and Kerala is openly studying them. The honest question is not "should India copy Japan", it is "what actually transfers".

Japan built universal Long-Term Care Insurance in 2000 and shifted the whole model to community-based care, deliberately keeping the old out of expensive hospital beds.¹² Kerala's day-care centres (Pakalveedu) and mobile Vayomithram clinics are the same instinct, done on a fraction of the money.
South Korea threw technology at the problem: AI companions, monitoring devices and telecare for elders living alone. For a state defined by empty nests and absent children, subsidised remote monitoring is arguably a better fit for Kerala than for anyone.
Scandinavia offers the ethic, not the model. Its high-tax, fully state-funded institutional care is out of reach for a middle-income state with Kerala's debt. What is portable is the standard: dignity, regulation, and a statutory body to enforce it, which is precisely what the new Elderly Commission is for.

What does not transfer is the money. Japan, Korea and the Nordics all had the fiscal depth to fund this. Kerala's real innovation is being forced to design an ageing-society welfare state on a budget none of those countries had to work with. If it cracks that, it writes the manual for the rest of India.

That tension between a serious long-term vision and a genuinely tight balance sheet is the thread I find most interesting, and it is the kind of thing I dig into every day with the readers in the BharatNama community (t.ly/h2jq1).

👉 Join the BharatNama community

What to watch

A few concrete things over the next few quarters:

The Oommen Chandy scheme's real cost. Watch whether the ₹25 lakh cover survives contact with the actuaries, or gets quietly trimmed to critical illness only once the premium bill is totted up.
Whether the Department gets teeth or stays a letterhead. Its first real test is auditing and consolidating the patchwork of existing schemes, not just existing on paper.
The Elderly Commission's first interventions. Does it actually act on property disputes and elder abuse, or become another advisory body?
The other southern states. Tamil Nadu and Andhra are a decade behind Kerala on this curve. Watch whether they start copying the institutional triad before their own bills arrive.

The one line

Kerala has become the first Indian state rich enough to grow old and too stretched to fully afford it, which is exactly the paradox the rest of India is walking towards.

And if you want to understand where China stands and what it means for India, do check out my companion newsletter, Decoding the Dragon: https://t.ly/t7uhs

And, do check out my work on the following platforms as well: Instagram, LinkedIn and Youtube

Best,
Jayant

References

Onmanorama, Kerala's Elderly Budget 2026-27, link
UNFPA India, India Ageing Report, link
Business Standard, Kerala becomes first state to unveil 'Elderly Budget', link
The South First, Kerala's 'Elderly Budget': an ageing state braces to support its seniors, link
Business Standard, Kerala Assembly election results 2026, link
Wikipedia, Department of Senior Citizens Welfare (Kerala), link
NextIAS, Kerala's Senior Citizens Welfare Department, link
PRS India / India Code, Kerala State Elderly Commission Act, 2025, link
The South First, Rising lifestyle diseases in ageing Kerala: Economic Review, link
ForumIAS / MoSPI 'Elderly in India 2021', Kerala highest share of elderly, Bihar lowest, link
Government of China, over one-fifth of Chinese population older than 60, link
Japan Today, Share of population aged 65 or older hits record high 29.4%, link
StudyIQ, Kerala Migration Survey 2023 key findings, link
The South First, Kerala youth exodus, rise in student emigration, link
Onmanorama, Kerala Budget 2026: Oommen Chandy health insurance ₹25 lakh coverage, link
BMC Geriatrics, The burden of disease-specific multimorbidity among older adults in India and its states (LASI), link

#42: Why did India build 30 FIGHTER JETS it cannot fly?

Jayant Mundhra — Thu, 27 Aug 2026 01:30:32 GMT

Here is a fact that should stop any Indian defence planner cold. Right now, on the tarmac at Hindustan Aeronautics Limited in Bengaluru, sit around 30 fully built Tejas Mk1A fighters. Airframes complete. Avionics fitted. Radar installed. Test-flown, in some cases. And not one of them can be handed to the Indian Air Force.¹

The reason is absurdly simple. They have no engines.

India can design a modern fighter jet. It can build the wings, the fly-by-wire, the indigenous Uttam radar, the whole complex machine. What it cannot build, after 37 years of trying, is the one part that actually makes a jet a jet: the engine.² That single gap is the most expensive sentence in India's defence story, and in 2026 it finally forced a reckoning.

The anatomy of a captive

The engine India is waiting on is the GE F404, made by General Electric in the United States. In 2021, HAL signed a ₹5.4k crore (~$716mn) deal for 99 of them to power the Tejas Mk1A.³

Then the deliveries did not come.

GE had shut its F404 line around 2016 after an earlier order, and restarting a cold production line proved brutally slow.⁴
As of July 2026, GE has handed over just seven engines against a schedule that was meant to deliver 24 to 26 a year.⁵
HAL has publicly slapped GE with liquidated-damages penalties, but a penalty does not build a fighter squadron.⁵

Now sit with the strategic meaning of that. India's frontline light-fighter programme, the backbone meant to replace the retiring MiG-21s and Jaguars, can be paused by a supplier 12,000 km away missing a delivery slip. This is what dependency actually looks like when the bill comes due.

It is worth being precise about the arithmetic, because it is stark. HAL has built the airframes faster than GE can supply the engines to put in them. So the bottleneck is not Indian factories or Indian engineers. It is a single imported component that India is not allowed to make, and cannot yet make itself. Even the maintenance depot for the F404 is only now being set up on Indian soil, years into the programme.⁴

And the bill is coming due at the worst possible moment. The IAF is down to roughly 29 to 30 fighter squadrons against a sanctioned strength of 42, its thinnest in decades.⁶ Every grounded Tejas is a hole in that number.

This is exactly the kind of structural vulnerability I keep coming back to in the BharatNama WhatsApp community, where I share a smaller India deep-dive like this with thousands of readers every morning (t.ly/h2jq1). A country that cannot power its own weapons does not fully own them.

Join the BharatNama community

The 37-year shadow

To understand why India is here, you have to go back to 1989, when the Cabinet cleared the Kaveri engine programme, meant to power the Tejas by 1996.²

It never happened. And the reasons are a masterclass in how not to fund hard technology.

The programme was sanctioned with roughly ₹383 crore. A serious military turbofan needs multiples of that, spent on destructive testing, exotic metallurgy, and high-altitude rigs.
The often-cited illustration of the gap: where GE reportedly tested dozens of alloy formulations to perfect the F404's turbine, India's materials budget stretched to a mere handful.² So Kaveri's turbine blades kept failing at combat temperatures.
India had no high-altitude test facility of its own, so prototypes were shipped to Russia for validation, at great cost and delay.²
After US sanctions following the 1998 Pokhran tests cut off collaboration, GTRE was left reinventing several complex domains at once, alone.

By 2008, the engine was overweight and underpowered, topping out near 73 kN of wet thrust against the 80-plus a Tejas needs.² The government blinked. It formally delinked Kaveri from the Tejas in September 2008 and bought GE engines instead.² Kaveri became a technology demonstrator: a museum piece with a pulse.

The lesson is painful and it is the whole point. India spent decades starving the programme to save money, then spent far more importing engines to cover the gap. Short-term thrift, long-term capture.

The thing that changed in 2026

Which is what makes this year genuinely different. On 16 February 2026, with Defence Minister Rajnath Singh watching, GTRE ran a full afterburner test of an upgraded Kaveri and finally crossed the line it had missed for a generation: 81 to 83 kN of wet thrust.⁷

Let that land. That is broadly the same league as the GE F404 (~84 kN) that India has been importing all along.⁷ The number India could not reach in 37 years, it reached this year.

How? Partly by finally spending, and partly by borrowing brains from an unlikely place: BrahMos Aerospace, the supersonic-cruise-missile joint venture, whose engineers helped redesign the afterburner's flame holders and combustion.⁷ India had the propulsion talent all along. It just kept it in different silos.

The upgraded engine leans on modern single-crystal superalloy blades and digital engine control, the exact metallurgy the 1990s programme could never afford.⁷
Even quieter, and arguably more important, is the "Dry Kaveri", a 49 to 51 kN non-afterburning variant now closing on certification, targeted for the end of 2026.⁸
That dry engine is the one meant to power the DRDO Ghatak, India's stealthy flying-wing strike drone, and a whole planned family of unmanned combat aircraft.⁸

And this is where the payoff gets interesting. The Ghatak is built for a doctrine called manned-unmanned teaming: a piloted Tejas Mk2 acting as the quarterback, directing a pack of stealthy Ghatak drones ahead of it as "loyal wingmen" to scout, jam and take the first hits.⁸ The IAF wants well over 100 such drones. Crucially, none of that fleet flies unless the engine is Indian, because no one exports you an engine for a drone meant to penetrate a rival's air defences.

None of this means the Kaveri flies on a fighter tomorrow. Certification, endurance testing and flying testbeds will take years. But for the first time, the core physics is no longer the obstacle.

Where India stands: against the world, and against itself

Here is where the benchmarking gets uncomfortable, because it always does. The club of countries that can independently design and build a fighter jet engine has maybe five members: the United States, Russia, the United Kingdom, France, and now China.⁹ India wants to be the sixth. Look at the gap to the newest entrant:

China. Went from importing Russian AL-31 engines to fielding its own WS-15 in serial production, powering 300-plus J-20 stealth fighters flying on fully domestic engines.¹⁰ Its indigenous turbofan sits in the ~140-160 kN class.
India. After 37 years, has just touched 83 kN on a test bed, and has never flown an indigenous engine on a fighter.²

That is the real scoreboard. China and India both started their serious engine journeys as importers. One now flies hundreds of stealth jets on home-built engines; the other is celebrating a bench test. The gap is not talent. It is decades of sustained funding and institutional patience that India chose not to give itself.

Now turn the lens on India against itself, because the more revealing comparison is India-then versus India-now:

The old model: a single state lab, GTRE, starved of cash, working in isolation, shipping prototypes abroad to test.
The new model: a funded push with the private sector inside the tent. Godrej Aerospace and Larsen & Toubro are building engine modules and the Ghatak airframe; BrahMos is fixing the afterburner; and in July 2026 Hyderabad's Azad Engineering handed over India's first indigenous expendable turbojet, a 350 kg-thrust-class engine built in 24 months under a public-private model.¹¹

The same country, two entirely different theories of how to build hard technology. The second one is finally the one being funded.

The price of the trap, and the French escape hatch

The 2026 crisis was not only about the F404. It was about leverage. When India went to buy the heavier GE F414 for the Tejas Mk2 and the fifth-generation AMCA, GE knew India's airframes were frozen around that exact engine, and priced accordingly.

The F414 quote reportedly jumped to around ₹210-240 crore (~$24mn) a unit, roughly three times the earlier ₹70-80 crore (~$9mn) projection.¹²
Swapping the engine out was near impossible: the AMCA's stealth shaping and intakes were built around the F414's geometry, and changing it would reset years of flight testing.¹²

So New Delhi did the one thing that actually breaks a monopoly. It went shopping for the technology itself, not the product. France's Safran is now close to a landmark pact to co-develop a clean-sheet 120 kN engine for the AMCA Mk2, with what is described as 100% technology transfer, full intellectual-property rights, and no export strings, a roughly $7bn programme targeting 400-plus engines.¹³ Britain's Rolls-Royce has pitched a rival 110-130 kN engine on the same 100%-transfer terms.¹⁴

This is the part worth sitting with in the BharatNama community too (t.ly/h2jq1): the difference between buying an engine and buying the ability to build engines. One keeps you a customer. The other, eventually, ends the dependency for good.

Join the BharatNama community

What India can learn

The lesson is not abroad this time. It is in the mirror, and in China.

China's engine programme succeeded because it was funded relentlessly for two decades through repeated, humiliating failures, without the state losing its nerve or delinking the engine to save face. India delinked Kaveri in 2008 the moment it got hard. That single instinct, cut the losing programme, is exactly what kept India a captive customer.

What transfers:

Fund through the failures. Engine development is destructive testing at industrial scale. There is no cheap version. The 2026 re-funding of Kaveri and the NAETC national engine-test complex being built at Ramanagara so prototypes never ship abroad again are, finally, an acceptance of that.¹⁵
Keep the private sector inside. The Azad, Godrej, L&T and BrahMos involvement is the structural change that matters more than any single test.¹¹
Own the IP, not just the assembly. The Safran and Rolls-Royce offers are valuable only because they hand over the metallurgy and the design authority, the exact things a licence-build never gives you.

What does not transfer is the shortcut. There isn't one. A jet engine is the one technology money and patience buy, and nothing else does.

What to watch

Whether the Dry Kaveri clears final certification by the end of 2026, the trigger for the CCS to fund full-scale Ghatak drone prototypes.⁸
Whether the Safran co-development pact actually reaches CCS approval, or stalls like so many before it.¹³
Whether GE's F404 deliveries finally hit the promised 24-a-year run rate and let those 30 grounded Tejas reach the IAF.⁵
Whether the 130 kN twin-engine test bed and the NAETC break ground on schedule, the infrastructure that decides whether any of this is repeatable.¹⁵

The one line

India spent 37 years learning that you cannot import your way to air power, and 2026 is the year it finally started paying the real price of building it at home.

And if you want to understand where China stands and what it means for India, do check out my companion newsletter, Decoding the Dragon: https://t.ly/t7uhs

And, do check out my work on the following platforms as well: Instagram, LinkedIn and Youtube

Best,
Jayant

References

#41: MODI vs STATES: Why Indian states get 32%, not the promised 41%?

Jayant Mundhra — Tue, 25 Aug 2026 01:30:47 GMT

The nine points that go missing

Here is a number that should bother anyone who cares about how India is run. The Finance Commission says India's states are entitled to 41% of the Centre's tax pool. The states actually get about 32%.¹⁰

That nine-point gap is not an accounting slip. It is the single most important fact about Indian fiscal federalism today, and almost nobody frames it this way.

The trick is elegant. The Constitution lets the Centre keep two kinds of tax entirely to itself. One is a cess, a tax levied for a stated purpose, say education or clean energy. The other is a surcharge, a tax on a tax. Neither is shared with the states.⁵

So the Centre has a lever. Raise money through ordinary income tax or GST, and 41 paise of every rupee flows to the states. Raise the same money through a cess instead, and the states get nothing.

Guess which one a revenue-hungry Centre reaches for. The headline devolution rate has never looked more generous. The effective one has rarely looked worse. That is the whole story.

How the pool actually works

Start with the plumbing, because the plumbing is the point.

India's Constitution hands the Centre the big, buoyant taxes: corporate tax, income tax, GST, customs. It hands the states the big spending jobs: hospitals, schools, police, roads, agriculture. That mismatch between who collects and who spends has a name, vertical fiscal imbalance, and every large federation lives with some of it.

To bridge it, the Constitution orders a Finance Commission every five years to decide what share of the Centre's taxes the states get. That shared portion is called the divisible pool.

The 14th Finance Commission raised the states' recommended share from 32% to a landmark 41-42%.¹
The 15th kept it at 41%.¹
The 16th, whose report covers 2026 to 2031 and was accepted in the 2026-27 Budget, kept it at 41% again, resisting demands from 18 states to push it to 50%.¹ ²

On paper, then, nearly half the Centre's taxes go to the states. In practice, far less does. The reason sits in one line of the Constitution.

The loophole was written in the year 2000

Before 2000, the sharing arrangement was narrower and messier. The 80th Amendment fixed that by pooling all central taxes together and giving states a single share of the lot.⁵

It was sold as a win for the states, and in one way it was. But tucked inside the same amendment was a carve-out. Cesses raised for a specific purpose, and surcharges under Article 271, were placed outside the divisible pool.⁵

The original logic was reasonable. A cess is money collected for a named job, so the argument went that it should not be handed to states as untied cash to spend on anything. The Fourth Finance Commission said as much back in 1965.

Here is where it gets strange. That logic only holds if the money actually reaches the job it was collected for. Increasingly, it does not.

The share of cesses and surcharges in the Centre's gross tax revenue tells the story of a loophole growing into a pillar:

Around 10.4% in 2011-12.³
A peak of 20.2% in 2020-21, roughly double.³
Still 14.5% in 2023-24, well down from the peak but far above where it started.³

Every rupee inside that rising share is a rupee the 41% formula never touches. The pool shrinks before the split even happens.

The auditor's inconvenient numbers

The one defence of keeping cesses out of the pool is that they are ring-fenced for their stated purpose. India's own auditor has demolished that defence.

The Comptroller and Auditor General, in its Report No. 16 of 2025, found that ₹3.69 lakh crore of cess money had simply never been moved to the reserve funds it was earmarked for, as of March 2024.⁴ It was collected in the name of highways, health, education and the rest, and then left sitting in the Centre's general account.

The detail is worse than the headline:

The oil industry cess raised about ₹2.9 lakh crore since 1974-75. Just ₹902 crore, or 0.3%, reached the fund it was meant for.⁴
Health and education cesses worth ₹37,537 crore collected between 2018-19 and 2023-24 never reached their dedicated funds.⁴

This is not new. An earlier CAG audit found that in 2018-19 alone, 35 different cesses raised about ₹2.75 lakh crore, of which only ₹1.64 lakh crore went to the funds they were levied for.¹²

So the sequence is this. The Centre taxes citizens in the name of a specific cause. It keeps the states out of that money by calling it a cess. Then it often does not spend it on the cause either. The earmark is a label, not a ring-fence.

This is exactly the kind of thread I keep pulling on, and it is the sort of thing we go back and forth over in the BharatNama WhatsApp community, where I share a smaller India deep-dive like this with thousands of readers every morning (t.ly/h2jq1).

Join the BharatNama community

December 2025: the loophole gets a new lease

If you think this is a story about old habits, watch what just happened.

The GST compensation cess was a special levy, brought in when GST launched to make good the revenue states lost in the switch. Its proceeds flowed back to the states. It was always meant to be temporary, and its statutory life runs out on 31 March 2026.⁶

Instead of letting that revenue fold into ordinary, shareable GST, Parliament acted in December 2025. It passed the Central Excise (Amendment) Bill and moved the Health Security to National Security Cess Bill, both aimed at tobacco and pan masala.⁶

The excise duties rise to offset the ending compensation cess.⁶
A new production-based cess targets pan masala and gutkha machines.⁶
Demerit goods like tobacco sit at the 40% GST slab, but the top-up now comes through excise and a cess, not a shared tax.⁶

Read the manoeuvre plainly. A levy that used to flow to the states is expiring. It is being replaced by levies that stay entirely with the Centre, wrapped in the language of public health and national security. A state-friendly tax dies, and a Centre-only tax is born in its place.

Where India stands, against itself

Now break the national number open, because the pain is not shared evenly.

Cesses fall heaviest on consumption, and consumption is highest in India's richer, more industrialised states. So the states that generate the most tax are the ones that lose the most when revenue is routed through a cess rather than the pool. The loophole quietly punishes the states doing the economic heavy lifting.

The 16th Finance Commission tried to answer this on the distribution side, and this is where its real changes sit. It left the 41% share alone but rewired the formula that splits the pool between states:²

A new "contribution to GDP" criterion, weighted 10%, rewards high-output states for the first time.²
The old "tax and fiscal effort" weight was scrapped.²
Income distance, the equity measure, was trimmed from 45% to 42.5%; population nudged up to 17.5%; area cut from 15% to 10%.² ¹

The result is a tilt towards the productive south and west:

Karnataka rises from 3.65% to 4.13%, the single biggest gainer.⁸
Maharashtra edges from 6.32% to 6.44%.⁸
All five southern states gain share under the new formula.⁸

But notice what this really is. Rewarding rich states with a bigger slice partly offsets the fact that those same states lose the most to cesses. It treats the symptom, the division of the pool, while leaving the disease, the shrinking of the pool, untouched.

And the 16th FC went further in a way that stings the weakest states. For the first time, it discontinued the Article 275 revenue deficit grants that used to plug the gap for states like Himachal Pradesh and Punjab, betting that state finances will run surpluses.⁹ It did expand local body grants sharply, to about ₹7.9 lakh crore.⁹ But that shifts money from untied constitutional entitlements towards conditional, Delhi-designed grants. That is centralisation dressed as cooperative federalism.

What India can learn from Australia

India is not alone in facing a wide gap between who taxes and who spends. Australia's federal government collects roughly 80% of public revenue and lets its states carry a huge share of the actual services. The imbalance is as steep as India's. The difference is entirely in the machinery.

Australia's answer is a ring-fenced pool. The entire GST take is dedicated to the states and territories, and Canberra cannot invent a "national security cess" to siphon money out before the split. There is no loophole to grow.¹¹

The sharing is run by the Commonwealth Grants Commission, an independent body that does something India's system does not. It measures each state's real capacity to raise revenue and its real cost of delivering services, then tops up the weaker states so every state can offer a comparable standard.¹¹

Two features are worth India's attention:

A "no worse off" guarantee, legislated to 2026-27 and extended by agreement to 2029-30, protects every state during reform. Crucially, Canberra funds these top-ups from its own money, not by raiding the shared pool.¹¹
A relativity floor, now 0.75, ensures no state's per-person GST share collapses below a set level.¹¹

The lesson is not "copy Australia". India's scale, its federal politics and its fiscal room are all different. The transferable idea is narrower and sharper. A federation can run a huge vertical imbalance without wrecking trust, but only if the shared pool is broad, ring-fenced, and immune to unilateral carve-outs. India has the independent body in the Finance Commission. What it lacks is the sealed pool.

The state-versus-state spread and the Centre-versus-state tug are the kind of tension I unpack most days for the BharatNama community, and if this edition is your sort of thing, that is where the daily version of it lives (t.ly/h2jq1).

Join the BharatNama community

What to watch

Whether the 2026-27 Budget's numbers hold: fiscal deficit is targeted at 4.3% of GDP, down from 4.4%, and tax devolution to states is pegged at ₹15.3 lakh crore.⁷ A tight deficit target is exactly the pressure that pushes the Centre towards cesses.
The rollout of the Health Security to National Security Cess from April 2026, and how much tobacco and pan masala revenue it locks away from the pool.⁶
Whether the discontinued revenue deficit grants force hill and special-category states into fresh borrowing.⁹
Any move, however unlikely, to bring cesses and surcharges back inside the divisible pool. Without that, the 41% will keep meaning 32%.

The one line

India's states were promised 41% and handed 32%, and the missing nine points are not lost to arithmetic but to a loophole the Centre has every incentive to keep widening.

And if you want to understand where China stands and what it means for India, do check out my companion newsletter, Decoding the Dragon: https://t.ly/t7uhs

And, do check out my work on the following platforms as well: Instagram, LinkedIn and Youtube

Best,
Jayant

References

1. PRS Legislative Research, Report of the 16th Finance Commission for 2026-31: https://prsindia.org/policy/report-summaries/report-of-the-16th-finance-commission-for-2026-31
2. Business Standard, 16th Finance Commission retains 41% devolution, introduces GDP criterion: https://www.business-standard.com/budget/news/16th-finance-commission-retains-41-percent-devolution-gdp-criterion-126020101042_1.html
3. Factly / Dataful, Share of cesses and surcharges in gross tax revenue falls to 14.5% in 2023-24: https://factly.in/data-share-of-cesses-surcharges-in-gross-tax-revenue-falls-to-14-5-in-2023-24-down-from-the-high-of-20-2-in-2020-21/
4. Comptroller and Auditor General of India, Report No. 16 of 2025, Union Government: https://cag.gov.in/en/audit-report/details/128299
5. Drishti IAS, Cess and its role in Union finance (Articles 270-271, 80th Amendment): https://www.drishtiias.com/daily-updates/daily-news-analysis/cess-and-its-role-in-union-finance
6. Deccan Herald, What is the Health Security to National Security Cess Bill, 2025: https://www.deccanherald.com/india/explained-what-is-health-security-se-national-security-cess-bill-2025-introduced-in-lok-sabha-today-3815460
7. PRS Legislative Research, Union Budget 2026-27 Analysis: https://prsindia.org/files/budget/budget_parliament/2026/Union_Budget_Analysis-2026-27.pdf
8. ICRA, State Government Finances, February 2026 (16th FC inter-se shares): https://www.icra.in/Research/ViewResearchReport/6763
9. The India Forum, Continuity and Change in the 16th Finance Commission: https://www.theindiaforum.in/economy/continuity-and-change-16th-finance-commission
10. EnsureIAS, Centre-State Financial Devolution: https://www.ensureias.com/blog/current-affairs/centre-state-financial-devolution
11. Commonwealth Grants Commission, 2018 GST distribution legislation and no-worse-off guarantee: https://www.cgc.gov.au/about-gst-distribution/2018-gst-distribution-legislation
12. Shankar IAS Parliament, CAG report on cess collections (2018-19): https://www.shankariasparliament.com/current-affairs/cag-report

#40: Why does India lend just 55% of GDP to business?

Jayant Mundhra — Sat, 22 Aug 2026 01:30:37 GMT

Here is a number that should stop you. India runs one of the fastest-growing large economies on earth, with a deposit base that has nearly tripled in a decade, and yet its banks lend barely 55% of GDP to the private sector.¹

China lends 194%.² Vietnam, which started poorer and reformed later than India, lends about 145%.³ South Korea sits near 170%.⁴

So the obvious story writes itself: India is under-banked, Indians do not save, financial inclusion has failed. Except every part of that story is wrong.

The more I dig into this, the more I think India's credit problem has almost nothing to do with mobilising money, and almost everything to do with who gets first claim on it. And that changes what the fix has to be.

The banking system that quietly doubled

Start with the thing everyone gets wrong. India is not short of deposits.

Total bank deposits crossed ₹254 lakh crore by end-December 2025, up from about ₹225 lakh crore a year earlier and roughly ₹88 lakh crore a decade ago.⁵
Non-food bank credit reached ₹208 lakh crore over the same period, growing 14.5% year-on-year by December 2025.⁶
Total assets of India's formal lending system more than doubled as a share of GDP, from 58.8% in 1997 to 118.2% by 2022.⁷

Let that sink in. The pool doubled. The plumbing that carries that pool into the hands of factories, exporters and entrepreneurs did not. That is the whole puzzle in one sentence.

If deposits have tripled and the system has doubled, but private credit is stuck near 55% of GDP, the leak is not on the funding side. It is on the asset side, in how a bank is allowed to use the money once it has it.

How a ₹100 deposit gets carved up

This is where financial repression stops being a textbook phrase and becomes a set of very specific rules. Follow a single ₹100 deposit as it enters an Indian bank.

₹3 goes to the RBI as the Cash Reserve Ratio (CRR), earning nothing. The RBI cut CRR in phases through late 2025, from 4.5% down to 3%, which freed some cash, but the principle stands: a slice is locked before any lending happens.⁸
Around ₹18 to ₹22 goes into government bonds to meet the Statutory Liquidity Ratio (SLR), the rule that forces banks to park a fixed share of deposits in central and state government securities. The statutory floor is 18% today, down from a peak of 38.5% decades ago, but banks routinely hold well above it.⁹
Of whatever is left, 40% is ring-fenced for Priority Sector Lending (PSL), the mandate that directs credit to agriculture, micro enterprises and weaker sections rather than to whoever the market would price best.¹⁰

Add it up and less than half of every ₹100 is left for unconstrained commercial lending to industry, infrastructure and high-value services. The sovereign and a set of social mandates get first call on the nation's savings. Private enterprise gets the residue.

None of this is an accident. It is a deliberate, decades-old architecture, and each piece was built for a reason. SLR gives the government a captive buyer for its debt. PSL pushes credit to farmers and small firms the market would ignore. The intent is defensible. The macroeconomic side-effect is that industrial credit in India is made artificially scarce, and therefore artificially expensive.

Because their mandated holdings earn less than a market-priced loan would, banks recover the difference by charging fatter margins on the shrinking pool of commercial credit they are free to give. The unconstrained borrower, the factory or the exporter, ends up subsidising the repressed rest of the balance sheet. That is why Indian industry borrows dear even though Indian banks are unusually safe.

The banks are healthy. That is part of the problem.

Here is the twist. India's banks have rarely been in better shape, and that has made them more cautious, not less.

Gross non-performing assets (GNPA) have fallen to a multi-decade low of about 2.2% as of September 2025, with net NPAs at just 0.5%.¹¹
That is down from a peak near 11.5% in 2018, when a wave of bad infrastructure and corporate loans nearly broke the public sector banks.¹¹

Join the BharatNama community

After that trauma, and the brutal clean-up that followed, bank managements learned a lesson: government bonds carry zero risk weight and never turn into a non-performing loan, while a corporate loan can. So even beyond the statutory minimum, banks lean into sovereign debt as a defensive crouch. Safety and repression point the same way, and the private borrower loses twice.

This tension, between a banking system that is finally sound and an industrial sector that still cannot get cheap capital, is exactly the kind of contradiction I keep pulling apart in the BharatNama WhatsApp community, where I share a smaller India deep-dive like this with thousands of readers every morning (t.ly/h2jq1).

Where India stands: against the world

Now put India on the same chart as the countries it wants to compete with. Using the World Bank's private-credit-to-GDP measure:

India: ~55% of GDP. Barely half the average for emerging markets, which sits around 135%.¹
Vietnam: ~145%. A country that liberalised after India and from a lower base.³
South Korea: ~170%. A high-income economy that used deep credit to industrialise, and now wrestles with household debt near 89% of GDP.⁴
China: ~194%. Decades of state-directed credit, now tipping into overcapacity and hidden defaults.²

The lesson is not "more credit is always better". China is the cautionary tale at the top end: credit poured into loss-making local government vehicles and a property bubble, and its true bad-loan pile is far larger than the official numbers admit. Korea shows the mid-life risk, where deep credit curdles into household leverage.

But India is nowhere near those problems. India is on the opposite failure: too little productive credit, not too much. It is starving, not bingeing. And a country cannot build a manufacturing base on the residue of its own savings.

Where India stands: against itself

Now break the national number open, because the average hides the real story.

Bank credit-to-GDP in India's richer, industrialised states runs up to three times higher than in its poorer, agrarian ones.
Bihar and Uttar Pradesh, home to roughly a quarter of all Indians, show credit-to-GDP ratios of just 25% to 30%, dragging the national average down.
The industrial and metro clusters of Maharashtra, Tamil Nadu, Karnataka and Delhi soak up a hugely disproportionate share of formal credit.

So the system is not just shallow, it is uneven. It mobilises deposits from the whole country and then re-deploys them into a handful of commercial hubs. A saver in Patna funds a factory in Pune. That is fine as far as it goes, but it means the credit gap is really a set of gaps, and the states that most need investment get the least of it.

The shadow-banking bypass

When banks will not or cannot lend, someone else does. India's answer has been the explosive rise of Non-Banking Financial Companies (NBFCs), the shadow banks.

NBFC assets grew from 2.6% of GDP in 1997 to 16.3% by 2022, one of the fastest expansions in the system.⁷
Their credit book roughly doubled from ₹24 lakh crore in 2021 to about ₹48 lakh crore by March 2025, and NBFCs now supply around a fifth of all systemic credit.¹²

But the bypass has a fatal flaw. NBFCs cannot take cheap retail deposits, so they fund long-term loans with short-term market borrowing. When confidence cracks, as it did when IL&FS collapsed in 2018 and DHFL in 2019, the funding freezes and the whole sector seizes. Their stressed-asset ratio had climbed back to 5.9% by March 2025, a reminder that the shadow system trades one problem for another.¹³

So NBFCs solved the availability of credit for some borrowers, but never the cost. Money routed through wholesale markets is dearer than money routed through a savings account, and that premium lands on the same industrial borrower again.

What India can learn from Vietnam

If you want the transferable lesson, do not look at China's firehose or Korea's mature markets. Look at Vietnam.

In 1995, Vietnam's bank-credit-to-GDP ratio was just 19%. By 2025 it had reached about 145%.³
It did this by actively steering mobilised deposits into the real economy, into export zones, infrastructure and private manufacturing, rather than parking them in government paper.
That deep domestic credit, paired with open trade, made Vietnam the prime winner of supply chains shifting out of China.

The honest caveat is that Vietnam is now managing the hangover, with credit concentrated in property and thin bank capital buffers, and its central bank is tightening. So the lesson is not "flood the system". It is narrower and sharper: a developing country that wants to industrialise has to let its banking system finance industry, not the fiscal deficit.

What is portable for India is the direction of travel: deepen the sovereign bond market so pension funds, insurers and foreign investors buy government debt instead of banks, then phase down SLR as they step in. Shift the burden of priority lending onto transparent fiscal subsidies rather than a hidden tax on bank balance sheets. Give well-run NBFCs cheaper, more stable funding. None of it is easy in a federal country with a large deficit, but none of it is impossible either.

This is the sort of structural question, why the plumbing matters more than the pump, that the BharatNama community goes back and forth on every day, and I would love for you to be part of that conversation (t.ly/h2jq1).

What to watch

Whether the RBI signals any further move on SLR or the CRR path in upcoming policy reviews, the clearest sign that repression is easing.
Whether foreign portfolio limits in Indian government bonds keep rising, which would let banks step back as the residual buyer of the deficit.
Whether the new PSL Master Directions, effective from April 2025, change how much genuine industrial credit gets crowded out.
Whether NBFC stress at 5.9% stabilises or forces another round of RBI intervention.

Join the BharatNama community

India has the savings, the demographics and, for the first time in a decade, a genuinely healthy banking system. The money required for its manufacturing push is not missing. It is sitting in the vault, spoken for.

The one line: India's credit problem is not that it cannot raise the money, it is that it has already promised most of it to someone else.

And if you want to understand where China stands and what it means for India, do check out my companion newsletter, Decoding the Dragon: https://t.ly/t7uhs

And, do check out my work on the following platforms as well: Instagram, LinkedIn and Youtube

Best,
Jayant

References

Economic Survey / BIS on India's credit-to-GDP, LEAP Insights, link
World Bank, domestic credit to private sector (% of GDP), China, link
Fitch Ratings via The Investor, Vietnam credit-to-GDP at 145%, link
World Bank, domestic credit to private sector (% of GDP), Korea, link
RBI, deposits cross ₹253 lakh crore as of Dec 2025, ANI, link
Economic Survey 2025-26, credit growth and profitability, PIB, link
Harvard MRCBG / RBI, assets of India's financial system as share of GDP, link
RBI, current CRR and SLR, link
RBI, SLR history and current level, link
RBI, Priority Sector Lending Master Directions 2025, TaxGuru, link
PIB / RBI, GNPA at multi-decade low, link
CRISIL / RBI, NBFC credit book and systemic share, link
RBI Financial Stability Report, NBFC stressed assets, Business Standard, link

#39: Why India's states take 13% of every dam? AND WHERE IT GOES?

Jayant Mundhra — Thu, 20 Aug 2026 01:30:20 GMT

The landlords of the Himalaya

I have come to believe India's Himalayan states no longer see themselves as hosts to the national grid. They see themselves as landlords. And the rent is going up.

On 2 July 2026, Himachal Pradesh did something that tells you exactly where India's hydropower story has gone. Chief Minister Sukhvinder Singh Sukhu directed his officials to start taking back three working dams from SJVN, a central public sector company that spent years and thousands of crores building them.¹

The three projects, Sunni (382 MW), Luhri Stage-I (210 MW) and Dhaulasidh (66 MW), add up to 658 MW.¹ The state's argument is blunt: hydropower is the backbone of its economy, and it wants its rightful share of its own rivers.

The instrument the hill states use is a single line of policy. And once you understand it, you understand the whole shift.

The 13%, and how it actually works

Every hydro project in India must hand the host state 12% of its power, free, for the life of the dam, plus another 1% as cash for a local development fund.² Thirteen per cent, off the top, forever.

The detail is the whole story, so start with the mechanism. From the day a plant synchronises with the grid, its output is sliced up before the developer sells a single unit:

12% of everything generated goes to the host state, free of cost. The developer carries all the capital and running costs, the state carries none, and takes an eighth of the output as pure revenue.²
1% more is converted to cash and paid into a Local Area Development Fund (LADF), the pot meant for the villages the dam displaces.²
The state is asked to match that 1% from its own free-power share, and developers must give each project-affected family 100 free units of electricity a month for ten years.²

This was not designed as leverage. It grew out of an older, decent instinct: share a resource windfall with the region that bears the ecological cost. Dams submerge ancestral land, move communities and re-plumb rivers, so the state that loses the valley should share the gains.

Now the scale, because it explains the power shift. India has about 50 GW of large hydro installed, and just over 55 GW including small hydro, around a tenth of the total power mix as of October 2025.³ Against an assessed potential of roughly 133 GW, only a little over 40% is built.³

Almost all of what remains sits in the Himalaya and the Northeast. Whoever owns those valleys owns India's last big block of dispatchable clean power. And they have worked out what it is worth.

From compensation to a cash register

The clearest sign of the shift is what Himachal did with its free power over the past year.

For decades the state simply fed its 12% into its own board, the HPSEB, at a soft rate, and used it to keep local and farm tariffs low. Free power was a domestic subsidy.

At the end of 2025 the state cabinet tore that up. On 30 December 2025 it stripped the HPSEB of 438 MW of free power from the big Sutlej projects, Nathpa Jhakri and Rampur, and handed it to a new Energy Management Centre inside the Directorate of Energy, whose only job is to sell that power on the open exchanges.⁴

The logic is cold, and honestly hard to argue with as pure finance:

In summer evenings, when solar collapses at dusk and demand peaks, dispatchable hydro fetches premium spot prices.
Selling to the exchange puts unencumbered cash straight into the treasury.
Feeding it to your own board at a soft rate leaves money on the table.

But the state utility and its customers foot the bill. Himachal also ended its old "banking" arrangement, where it lent summer surplus to Punjab, Haryana and Delhi and drew it back in the lean winter.⁴ Punjab alone reckons the switch could cost it around ₹400 crore in a single paddy season.⁵ And the HPSEB now has to buy expensive winter power on the market, which lands back on Himachali consumers as fresh surcharges.

The treasury wins. The board loses. This is the kind of thread I keep pulling on, how a welfare rule mutates into a trading desk, and it is exactly what we go back and forth on in the BharatNama WhatsApp community, where I share a smaller India deep-dive like this with thousands of readers every morning (t.ly/h2jq1).

Join the BharatNama community

Someone always pays for "free"

The word "free" hides who actually pays.

When a developer loses 13% of its saleable power up front, it still has to recover its full cost from the remaining 87%. The Central Electricity Regulatory Commission bakes this in: a hydro plant's annual fixed cost is split, and the free-power share is subtracted from the saleable energy in the tariff formula.⁶ Shrink the base you can sell over, and the per-unit price for everyone else rises.

So the arithmetic is blunt. A Haryana or a Delhi consumer, buying their slice of a Himalayan dam, pays a slightly higher tariff to fund the free power gifted to Himachal or Arunachal. The host state's windfall is a surcharge on the plains.

And the plains have started to push back. Himalayan hydro is front-loaded and dear in its first 10 to 15 years, while the developer services construction debt, exactly when cheap solar is flooding the exchanges:

Haryana's distribution companies are already carrying a mountain of deferred fuel-and-power surcharge, and in early 2026 their own firms asked the regulator to let them push the recovery even further out.⁷
The Haryana regulator has been refusing power deals it judges unaffordable, knocking back a solar PPA in April 2026 on precisely that logic.⁸
Meanwhile NHPC is signing 40-year hydro deals elsewhere, like the 3,576 MW Sawalkot and Kamala block it tied up with Gujarat's GUVNL.⁹

The tension is real. States that host the dams want maximum rent. Developers want firm 40-year buyers. And the states that must actually buy the power increasingly say the price is too high.

Why the Centre keeps subsidising it anyway

You would think this friction would stall the sector. It has not, because Delhi cannot let it.

India crossed 505 GW of total installed capacity by the end of 2025, with non-fossil sources past half the base for the first time, and is chasing 500 GW of non-fossil capacity by 2030.¹⁰ The grid is filling with intermittent solar and wind. When the sun drops at 7pm, something has to ramp up in seconds, and only hydro and pumped storage can do it at scale.

So the Centre pays to keep the peace:

It makes buyers take hydro whether they like it or not, carving a Hydro Purchase Obligation into the renewable purchase mandate, the broader trajectory climbing towards 43% of power procurement by 2029-30.¹¹
It waives inter-state transmission charges for hydro and pumped storage for years after commissioning, quietly suppressing the delivered price so regulators like Haryana's will swallow it.⁶

Both are subsidies that hide the cost of the host state's 12%. The free power is never free. It is smeared across obligations and waivers so that no single bill ever shows it.

Where India stands, against the world

Set India's model beside the country that has done this longest, and the gaps are instructive.

Norway, on hydropower rent:

Host municipalities can buy up to 10% of a plant's output at a regulated cost price, close in spirit to India's 12%, but they pay for it rather than getting it free, and the difference funds local budgets transparently.¹²
Developers pay a resource-rent tax of 45% on top of the 22% corporate tax, a combined marginal rate of about 67%, so the national exchequer, not just the local one, captures the windfall.¹³
Municipalities and the state together collected roughly $90mn in licence fees in a single year.¹²

Then Bhutan, which turned the same resource into a nation's balance sheet. Hydropower is about a fifth of its economy and 63% of its exports, and India is the buyer.¹⁴ In Thimphu, hydro rent built a country. In Shimla, it is starting to build a treasury.

And where India stands, against itself

The domestic spread is the real story, because the same 12% rule is being played three completely different ways:

Himachal Pradesh, the aggressive extractor. It is monetising free power, formally asking the Centre in June 2026 to raise the royalty beyond 12% and to clear ₹7,784 crore in BBMB arrears,¹⁵ insisting dams revert to the state free of cost after 40 years, and moving to seize working projects.¹ It even tried a water cess on generation, until its own High Court struck it down as unconstitutional, since electricity is a central subject.¹⁶
Jammu & Kashmir, the accommodator. Its 2025 hydro policy eases terms to attract developers rather than squeeze them,¹⁷ because in a high-risk, high-cost basin a smaller share of a built dam beats a full share of a stalled one.
Arunachal Pradesh, the sleeping giant. Holding roughly a third of India's potential, in 2023 it handed 11,517 MW across 12 projects to central companies in one stroke,¹⁸ and expects to earn upwards of ₹4,500 crore a year from free power by 2035.¹⁹

Same rule, three different games. That state-versus-state spread is the sort of India-versus-India comparison I find most revealing, and it is what the BharatNama WhatsApp group, where thousands of us dig into a smaller India story each morning, exists for (t.ly/h2jq1).

Join the BharatNama community

What India could take from Norway, and where it cannot

Norway cracked the exact problem India is fumbling: how to share hydro rent fairly without wrecking the sector or the consumer.

What travels:

Charge for the local share, do not gift it. A priced entitlement, like Norway's concession power at cost, still enriches the host region but shows up honestly in a budget line, instead of being smeared across every other state's tariff.¹²
Ring-fence and audit the local fund. Norway's fees flow to local bodies on a transparent formula. India's 1% LADF too often never reaches the valley, with compensation money for displaced families dogged by governance gaps.
Tax the rent nationally, so the windfall is shared beyond the one lucky border district.¹³

What does not travel so easily:

India is a federation of 28 states, not one national utility, so a single state can genuinely hold a central developer hostage in a way a Norwegian municipality cannot.
India's dams sit in far more fragile geology. In October 2023 a glacial lake burst in Sikkim and wiped out the 1,200 MW Teesta-III dam,²⁰ a reminder that no revenue formula compensates for a project built in the wrong place.

The lesson is not "copy Norway". It is that India has built an elaborate machine to move hydro rent around, and forgotten to make it transparent, fair to the consumer, or safe for the people living below the wall.

What to watch

Whether Himachal actually completes its takeover of the 658 MW of SJVN projects, or settles for a higher royalty. It is the test case for how far a state can push.¹
Whether the Centre concedes any rise above the 12% baseline. Himachal is formally asking, and a yes would reset the economics of every future project.¹⁵
Whether more plains regulators follow Haryana and refuse long-dated hydro contracts on cost grounds, which would force the Centre to sweeten terms even further.⁸
Whether a single rupee of the extra spot-market money Himachal is now earning ever reaches the HPSEB's stranded consumers, or stays in the treasury.

The one line

India's hill states have realised their rivers are not just scenery, or even electricity. They are a sovereign wealth fund, and the meter is running.

And if you want to understand where China stands and what it means for India, do check out my companion newsletter, Decoding the Dragon: https://t.ly/t7uhs

And, do check out my work on the following platforms as well: Instagram, LinkedIn and Youtube

Best,
Jayant

References

1. Himachal CM orders takeover of Sunni, Luhri and Dhaulasidh hydro projects from SJVN, ANI News: aninews.in
2. Ministry of Power, National Hydro Power Policy (12% free power and 1% LADF): powermin.gov.in
3. Renewed Momentum: strong revival in hydro capacity addition and generation, Renewable Watch: renewablewatch.in
4. Himachal Pradesh to monetise free hydropower, The Tribune: tribuneindia.com
5. Himachal's open power market shift may cost Punjab Rs 400 cr in paddy season, The Tribune: tribuneindia.com
6. Vital Resource: policy initiatives to promote hydropower, Power Line: powerline.net.in
7. Power firms in Haryana seek rule change to defer fuel surcharge recovery, Business Standard: business-standard.com
8. HERC rejects PPA approval over delay and tariff concerns in Haryana, SolarQuarter: solarquarter.com
9. NHPC signs 40-year PPAs with GUVNL for 3,576 MW hydropower, Energetica India: energetica-india.net
10. India crosses 505 GW installed power capacity with non-fossil fuels exceeding 50%, SolarQuarter: solarquarter.com
11. Renewable Purchase Obligation and Energy Storage Obligation Trajectory till 2029-30, Ministry of Power: powermin.gov.in
12. Taxation of the power sector (concession power, licence fees), Norwegian Energy Facts: energifaktanorge.no
13. Increased resource rent tax on hydropower, Government of Norway: regjeringen.no
14. India-Bhutan hydropower cooperation, MP-IDSA: idsa.in
15. Himachal CM Sukhu urges Centre to raise free-power royalty beyond 12 per cent, The Tribune: tribuneindia.com
16. Himachal High Court strikes down water cess on power generation as unconstitutional, ThePrint: theprint.in
17. Jammu & Kashmir's new hydropower policy drops key restrictions to woo developers, ETV Bharat: etvbharat.com
18. 12 stalled hydropower projects (11,517 MW) in Arunachal handed to hydro PSUs, PIB: pib.gov.in
19. Dams, dreams and a decade that could change India, Policy Mandala: policymandala.theindiahouse.org
20. Glacial lake outburst flood destroys Teesta-III and damages Teesta-V, International Rivers: internationalrivers.org

#38: Why is private equity paying 16x for India's schools?

Jayant Mundhra — Tue, 18 Aug 2026 01:31:29 GMT

For a decade, the smart money was certain the classroom was dead. Software would eat it, the way it ate retail, taxis and travel.

They were wrong. And the correction has been brutal, and instructive.

In July 2026, a European buyout firm, Vitruvian Partners, paid ₹1,159 crore for roughly 16% of K12 Techno Services, the company behind the Orchids chain of schools.¹ That values a business with ₹523 crore of revenue and a net loss of about ₹25 crore at around ₹7,200 crore.²

Call it 15 to 16 times sales. For a school operator that does not yet turn a profit.²

Now let that sit for a second. The same class of investor that spent 2021 funding apps to make schools obsolete is now paying software-style multiples to own the schools themselves.

The disruption that got disrupted

This is a rotation, not a one-off. To see why, rewind to the boom that just collapsed.

Indian EdTech raised money at a pace that looked unstoppable:

~$1.88bn in 2020, then a peak of ~$4.73bn in 2021.³
Then the fall: ~$2.6bn in 2022, and just ~$297mn in 2023.³
Roughly 4,500 startups were fighting for the same shrinking market, and over 14,000 people were laid off in 2022 alone. Lido and Udayy shut entirely; Byju's, once the poster child, imploded.³

The moment physical schools reopened, the demand for a screen substitute evaporated. Without a building to enforce attendance, without the peer group, without the childcare a school quietly provides, the churn was savage. The survivors, Byju's and PhysicsWallah, ended up spending heavily to buy physical coaching centres.³

Here is the interesting bit. Investors did not conclude that education was a bad business. They concluded the opposite. The physical school was never the weakness in the model. It was the moat.

Why a school beats a SaaS app

Apply the two numbers every investor lives by, and you see what they saw.

Customer acquisition cost (CAC): what it costs to win one customer.
Lifetime value (LTV): the total profit that customer throws off before they leave.

In software, an LTV to CAC ratio of 3 to 1 is considered healthy. A premium school makes that ratio look almost embarrassing, for three reasons.

Near-zero churn. Once a family clears the admissions gauntlet and enrols a child, the odds of that child switching schools before Class 12 are tiny. One acquisition buys 12 to 15 years of recurring fees. A SaaS company would kill for that retention.
Inelastic, K-shaped demand. In a downturn, a family cancels the holiday, delays the car and eats out less, long before it touches the child's school. And the segment PE is chasing, the aspirational top, is the one whose incomes have risen fastest.
Brutal barriers to entry. Prime urban land, a maze of approvals, good teachers and, above all, decades of trust. You cannot spin up a credible school in a funding round. That scarcity is pricing power.

This kind of buried unit economics, the thing that quietly turns a school into a 15-year annuity, is exactly what I keep pulling apart in the BharatNama WhatsApp community, where I share a smaller India deep-dive like this one with thousands of readers every morning (t.ly/h2jq1).

Join the BharatNama community

The legal fiction that makes it investable

There is one problem. In India, you are not legally allowed to run a school for profit.

For generations, the law has insisted that schooling be charitable. Schools must sit inside not-for-profit trusts or societies. They cannot pay dividends. Any surplus must be ploughed back. To a buyout fund, whose entire job is to return capital, that is uninvestable.

So the industry engineered a workaround, the same one it built for private hospitals. The school is split into two entities:

The not-for-profit Trust. It holds the licence and the board affiliation, employs the teachers, and technically enrols the students. It collects the fees. It cannot distribute profit.
The for-profit Operating Company, or OpCo. This is where the PE money goes. It owns or leases the land, the technology, the brand and the curriculum, and sells all of it back to the Trust through long-term management and service agreements.

The surplus the school generates is then drained out of the charitable Trust and lands on the OpCo's books as revenue, amplified by the high-margin extras: uniforms, transport, digital subscriptions, catering. The distinction between a philanthropic school and a profit machine becomes, in practice, a legal fiction.

K12 Techno is the OpCo done at scale. It does not own the school trusts. It sells them a full stack, curriculum, software, hiring, admin, marketing, across more than 113 institutes in 17 cities.² That is the business Vitruvian just paid ₹7,200 crore to buy a slice of.¹

And the re-rating is the tell. K12 Techno was worth about ₹4,721 crore in its previous round; the July step-up to ₹7,200 crore let an early backer, Peak XV, the old Sequoia India, book close to a 12-fold return on its original bet.² When an asset class starts minting returns like that, the rest of the capital follows fast.

Where India stands, against the world

Zoom out, and India is early to a game the world has already matured into.

In October 2024, a consortium led by Neuberger Berman with EQT and Canada's CPP Investments agreed to buy Nord Anglia Education for an enterprise value of $14.5bn, one of the largest education deals ever. It closed in March 2025.⁴ ⁵
Nord Anglia runs 80-plus schools across 33 countries for more than 85,000 students, with tie-ups to MIT, Juilliard and UNICEF. EQT first invested back in 2008 and has compounded the asset ever since, bringing in CPP in 2017.⁴
Its mirror image, Inspired Education, backed by TA Associates, Warburg Pincus, Stonepeak and Singapore's GIC, now runs more than 125 schools across five continents.⁶

Against that, India's chains are small. K12 Techno at ₹7,200 crore is roughly $850mn, a fraction of a single Nord Anglia. And that gap is precisely the pitch: the runway is enormous, and the roll-up has barely begun.

Where India stands, against itself

Now break the national number open, because this is where the real story sits.

The UDISE+ 2025-26 data, the government's own school census, is stark:

Government schools have shed enrolment from ~128mn to ~119mn, losing roughly 8.6mn students in two years.⁷
Private unaided schools have gone the other way, from ~90mn to ~99mn, adding roughly 8.8mn students.⁷
Private unaided schools now hold close to 40% of all enrolment.⁷

So the migration is real and it is large. But the slice PE actually wants is tiny. Of India's ~248mn school students, only about 7.9mn sit in schools charging ₹1 lakh a year or more, drawn from perhaps 28.2mn households that can afford it.¹ That is the sliver the capital is racing for.

This split, 8.6mn children walking out of one system while 8.8mn walk into another, is the sort of India-versus-India thread we chew on daily in the BharatNama WhatsApp group (t.ly/h2jq1), where a version of this deep-dive lands every morning.

The other big consolidator shows how far this goes. KKR-backed Lighthouse Learning, the old EuroKids, now runs 1,850-plus preschools and 60 K-12 schools for more than 190,000 students, under EuroSchool, Kangaroo Kids and Billabong High.⁸ KKR bought EuroKids in 2019 for about ₹1,400 crore⁹, brought in Canada's PSP Investments in a fresh round in late 2025⁸, and is now buying the ultra-premium Pathways School in Gurgaon for around ₹1,500 crore.¹ A trophy IB asset, at the very top of the market.

Join the BharatNama community

What India can learn

Here is the uncomfortable part, and it is why this matters beyond the deal pages.

The rich world has already run this experiment, in higher education, and the results are on record. A landmark study by Eaton, Howell and Yannelis found that when private equity bought colleges, enrolment and profits went up, but educational inputs fell, tuition and student debt rose, graduation rates dropped and graduates earned less.¹⁰ The firms were very good at capturing subsidies. They were not good at teaching.

The mechanism to watch is the teacher. Salaries are the biggest cost line in any school, so the incentive is to trim them, raise class sizes and lean on younger staff. But the evidence says that is exactly where the damage lands. Ronfeldt, Loeb and Wyckoff, studying over 600,000 student records in New York, showed that higher teacher turnover measurably lowers achievement in maths and English, and not only by losing good teachers, but by tearing the school's social fabric.¹¹

And the consumer squeeze is already visible in India as "eduflation". Fees at a school like Delhi Public School, Vasant Kunj roughly doubled from ₹90,000 in 2019-20 to ₹1.7 lakh by 2024-25, and many urban families now spend 20% to 30% of household income on their child's education.¹² When a handful of well-capitalised chains set the regional benchmark, that pricing power only hardens.

And underneath all of it sits a mismatch that no amount of financial engineering resolves. A buyout fund needs its exit in three to seven years; a school's reputation compounds over decades. The pressure to inflate value before the exit window is precisely the pressure that pushes against patient, expensive, unglamorous good teaching. That is the unresolved variable in this entire thesis.

So what is portable, and what is not?

Portable: transparency and fee discipline. India already has the tools, state fee-fixation committees and disclosure laws in Delhi, Rajasthan and Chhattisgarh. The gap is enforcement, not statute.
Not portable: a simple "cap the fees" answer. The Supreme Court, in TMA Pai and PA Inamdar, protected private schools' autonomy while barring profiteering. India has to hold both at once, capital that funds good schools, without letting the classroom become pure extraction.

What to watch

Whether K12 Techno or Lighthouse files for an IPO. The exit is the whole point of a buyout, and a listing would price the entire sector.
Whether states tighten scrutiny of the Trust-OpCo structure and ancillary charges, or keep looking the other way.
Whether the Pathways deal actually closes at ₹1,500 crore, and what multiple it sets for trophy schools.
The court cases and protests, from the Orchids affiliation row in Bengaluru to the fee-hike agitations in Gurgaon and Delhi, and whether any regulator's order is actually enforced.

The one line

Private equity did not disrupt the Indian classroom; it discovered that the classroom was the safest annuity in the country, and quietly bought the annuity.

And if you want to understand where China stands and what it means for India, do check out my companion newsletter, Decoding the Dragon: https://t.ly/t7uhs

And, do check out my work on the following platforms as well: Instagram, LinkedIn and Youtube

Best,
Jayant

References

Business Standard, "The business of education: Why PE is lining up for India's premium schools", link
SaaSRise deal brief, "Vitruvian Partners acquires 16% stake in K12 Techno Services for ₹1,159 crore", link
Observer Research Foundation, "Edtech in India: Boom, bust, or bubble?", link
CPP Investments / EQT, consortium to acquire Nord Anglia Education, link
Ropes & Gray, "EQT Consortium Completes $14.5 Billion Acquisition of Nord Anglia Education", link
Inspired Education Group, corporate news, link
Free Press Journal, "UDISE+ 2025-26 Report: Government Schools Lose Nearly 86 Lakh Students As Private Schools Gain Over 88 Lakh Enrolments", link
PSP Investments, "KKR Further Invests in Lighthouse Learning to Support Next Phase of Growth", link
Business Standard, "PE firm KKR enters education sector in India, buys 90% stake in EuroKids", link
Eaton, Howell & Yannelis, "When Investor Incentives and Consumer Interests Diverge: Private Equity in Higher Education", NBER Working Paper 24976, link
Ronfeldt, Loeb & Wyckoff, "How Teacher Turnover Harms Student Achievement", American Educational Research Journal, link
The Print, "Indian parents trapped in a spiral of soaring private school costs. Aspiration fuels eduflation", link

#37: Why EXACTLY are India's agri-exports being REJECTED?

Jayant Mundhra — Sat, 15 Aug 2026 01:30:30 GMT

India is the largest mango producer on earth. We grow roughly 26mn tonnes a year, about 43% of the entire global crop.¹ And we export less than 1% of it.¹

Let that sit for a second. The country that basically defines the fruit sells almost none of it to the world.

That one number tells you most of what you need to know about where Indian agriculture actually stands in 2026. We have never been better at growing food, and we have never negotiated harder to sell it. Farm exports hit $53.1bn in 2025-26, a whisker below the all-time record and up 2.3% in a year when overall merchandise exports barely moved.²

And yet, at inspection counter after inspection counter, our produce keeps getting turned away.

We win the negotiation, then lose the container

Here is the thing. India has quietly become very good at one half of trade, and stayed fragile at the other.

The first half is market access, the diplomacy of tariffs. On that, New Delhi has been on a tear:

New Zealand signed a free trade deal on 27 April 2026 giving India duty-free access across all 8,284 of its tariff lines, plus a $20bn investment pledge over 15 years.³
The India-UK CETA came into force on 15 July 2026, scrapping duties on about 99% of Indian goods and targeting a doubling of two-way trade to $56bn by 2030.⁴

The second half is compliance: the world of sanitary and phytosanitary rules, or SPS, the residue limits, fumigation certificates and heat-treatment protocols that decide whether a container actually clears the border. On that, we are bleeding shipments every week.

Australia has just suspended 44 Indian fumigation agencies.⁵
The EU ranked India first in the world for pesticide alerts last year.⁶
Japan halted our mangoes for the first time in 20 years.⁷
And banned antibiotics drove 43% of our shrimp rejections.⁸

This is the compliance paradox. We have optimised the art of the trade deal, and our fragmented supply chain still cannot meet the zero-tolerance rulebooks on the other side. This split, brilliant at the table and shaky at the counter, is exactly the tension I keep pulling apart, and it is the sort of thread we chew on every morning in my BharatNama WhatsApp community, where I share a smaller India deep-dive like this one with thousands of readers (t.ly/h2jq1).

Join the BharatNama community

The new trade diplomacy: take a lot, give a little

India's trade posture has flipped. For decades we were the reluctant partner who walked out of RCEP. Now we are signing, and fast.

But read the fine print and one pattern jumps out. We take almost everything, and give back carefully. Call it asymmetric liberalisation.

New Zealand:

Wellington opens 100% of its 8,284 lines to us, immediately.³
India liberalises 70.03% of its lines, and only 30% of those go to zero on day one; the rest phase in over three to ten years.³
Crucially, India walls off 29.97% of its tariff lines completely: dairy, onions, chickpeas, corn, sugar and edible oils.³

UK CETA:

Britain drops duties on about 99% of Indian goods.⁴
India opens roughly 89.5% of its lines, but only 24.5% of UK exports get duty-free access immediately.⁴
Scotch whisky, for instance, only falls from 150% to 100% now, reaching 40% after a decade.⁴
Dairy, cereals, millets, pulses and apples stay excluded.⁴

The logic is defensible. About 89% of India's agricultural households farm under two hectares.¹⁴ You cannot expose that base to New Zealand milk powder or Australian chickpeas overnight.

But here is the quiet cost. By winning duty-free access abroad and shielding the farm gate at home, the government has handed the entire burden of execution to the exporter. Tariffs are gone. Non-tariff barriers are not. And non-tariff barriers are technical, unforgiving, and exactly where India's supply chain breaks.

Look at the UK prize. Britain imports about $37.5bn of agriculture a year and takes just $811mn of it from India, around 2.2%.⁹ The headroom is enormous. Shrimp, tuna and fishmeal that carried 4.2% to 8.5% UK tariffs are now duty-free.⁹ Tea, spices and basmati walk in free.⁹ On paper, a windfall. In practice, only if the container clears inspection.

SPS: the barrier that replaced the tariff

When tariffs fall, the friction does not disappear. It just moves to SPS and to technical barriers to trade, or TBT: the standards, certifications and residue limits a country is allowed to set under WTO rules, as long as they rest on science and are not disguised protectionism.

As tariffs vanished worldwide, SPS notifications surged. They have quietly become the modern weapon of trade friction, and they fall hardest on developing exporters who cannot pivot the moment a rule changes.

India's problem is rarely the intrinsic quality of the food. It is the chain of proof. In global trade the absence of contamination is not enough; you need an unbroken chain of accredited paperwork and residue compliance. When that chain snaps, or when a foreign regulator moves faster than our farms can adapt, the shipment dies. Four live crises show exactly how.

1. Australia: the cost of a broken certificate

Australia runs one of the strictest biosecurity regimes on earth. It does not test every container; it trusts accredited Indian fumigators to treat rice with methyl bromide or aluminium phosphide before it sails.

On 13 July 2026, its agriculture department suspended 44 Indian fumigation agencies, roughly 60% of the ones it accredits, most clustered in the basmati belts of Haryana and Punjab.⁵ Audits had found certificates that did not add up, including consignments recorded as packed before they were fumigated.⁵

The fallout is brutal:

More than 100 containers of basmati, worth about ₹200 crore, are stuck in regulatory limbo.⁵
Each now faces mandatory re-fumigation on arrival, at $700 to $1,200 a container.⁵

The rice itself may be spotless. But the paperwork lost its validity, so the shipment became a liability. That is the paradox in a single container.

2. The EU: a chemical chasm

India and the EU farm by different rulebooks. Our food regulator sets residue limits for a hot, humid, pest-heavy country; Brussels pushes maximum residue limits toward near-zero for anything it has banned. Chemicals our farmers lean on are simply illegal in Europe.

The collision shows up in the EU's rapid alert system. In 2025, India ranked first in the world for pesticide alerts, with 124 of them, ahead of Turkey at 105, Egypt at 88, China at 48 and Brazil at 34.⁶ Being number one in that particular league is not a badge anyone wants.

The repeat offenders are familiar: ethylene oxide, a banned fumigant, turning up in sesame; tricyclazole, a rice-blast fungicide, in basmati; chlorpyrifos in rice. Indian sesame now carries a 20% EU border-check rate for ethylene oxide, and cumin a 30% rate for other residues.¹⁰ Every alert triggers destroyed consignments, higher insurance and slower clearance for everyone shipping behind you.

3. Japan and Nepal: blocked by a protocol

For 20 years India shipped premium Alphonso and Kesar mangoes to Japan using vapour heat treatment, a chemical-free process that cooks out fruit-fly larvae. Then, on 29 May 2026, Japan suspended imports carrying certificates issued after 25 March, after inspectors flagged lapses at a treatment facility in Rehmanpur, Uttar Pradesh.⁷

Japan is tiny for us, about $1.54mn a year, mostly Gujarat's Kesar.⁷ But it was our first Japanese restriction in two decades, and the reputational hit travels.⁷

Nepal added to the noise. Reports of a Nepali "ban" flew around, but both governments have since clarified there is no formal ban; Nepal has instead introduced new phytosanitary conditions, including hot water treatment, and India is contesting the measures through the WTO and bilateral channels.¹¹ Notice the pattern across all three: the block is never about the fruit. It is about a protocol.

4. Shrimp: zero tolerance on antibiotics

Our marine sector faces the hardest wall of all. Western regulators permit no trace of antibiotics like chloramphenicol or nitrofurans, because of links to antimicrobial resistance.

Per India's own drug regulator, banned residues drove 43% of shrimp rejections in 2025 across the EU, the US and Japan.⁸ The US FDA alone logged 93 antibiotic-linked shrimp refusals that year, its sixth-highest count in 24 years.⁸

Where India stands, against itself

Now BharatNama's favourite question. Where inside India does the problem actually live? Because the national number hides a very lopsided map.

When our marine authority traced the farms behind the 2026 antibiotic detections, the concentration was staggering:

Andhra Pradesh accounted for 74% of the implicated farms.⁸
Odisha made up 13%, West Bengal 8.7% and Gujarat 4.3%.⁸

One state drives three-quarters of a national reputational risk. The basmati story is just as concentrated: the suspended fumigators sit overwhelmingly in Haryana and Punjab.⁵ These are not "India" problems in the abstract. They are problems of a handful of clusters, which is the bad news, concentration of risk, and the good news, because a handful of nodes is exactly what you can fix. This kind of state-versus-state spread is the sort of number I find genuinely fascinating, and if unpacking them is your thing too, come join the daily India deep-dives in the BharatNama WhatsApp group (t.ly/h2jq1).

Join the BharatNama community

Where India stands, against the world

Set India beside the countries it actually competes with, on current numbers, and the picture sharpens.

On EU pesticide alerts, India leads the whole table: 124 against China's 48 and Brazil's 34.⁶ Yes, we ship more, so we trip more, but the gap is not only about volume.
In seafood, Vietnam out-earns us outright. It exported about $11.3bn of seafood in 2025, shrimp alone at roughly $4.65bn, growing 13% in a year.¹⁶ India's marine exports were about $7.45bn in 2024-25.¹⁷

Vietnam is the tell. It faced the same Western zero-tolerance wall we do. It responded by wiring tighter traceability, cold chains and compliance into the supply chain, and then climbed the value ladder.¹⁶ We are the bigger farm; they are the better exporter.

What India can learn

The countries that cracked this all did one thing we have not. They pushed compliance upstream, to the farm, instead of trying to inject it at the port.

Vietnam built traceability and third-party certification into its seafood chain, and is now an $11.3bn exporter.¹⁶
Kenya and Peru turned smallholder horticulture into EU-grade exports by clustering farmers under GlobalGAP certification, so a single verified block, not a lone farmer, carries the audit.¹⁸

What actually transfers to India, and what does not, matters here, because I have no time for lazy "India should just copy them" takes.

Portable:

Farm-level traceability, geo-tagged plots, residue testing before harvest, and clustering small farmers into certified blocks. We have started: our marine authority runs pre-harvest ELISA testing across a network of 12 labs and a Shaphari antibiotic-free certification scheme, and APEDA is building farm-to-port traceability for crops.^12,13

Not portable wholesale:

We cannot simply adopt EU residue limits. India's roughly 89% smallholder base farms in humid, pest-heavy conditions and depends on affordable agrochemicals to feed 1.4bn people.¹⁴ A blanket ban on tricyclazole would help exporters and hurt domestic yields. The realistic path is phased harmonisation inside designated export zones, not a national copy-paste.

Credibility begins at home

There is a home-front dimension foreign regulators watch closely. In July 2026, our food safety authority suspended the licence of Westend Agro Products, maker of the "Organic Shastra" brand, after finding machinery altering batch numbers and expiry dates, and packaging that falsely claimed the goods were "organic".¹⁵

When your own regulator keeps uncovering date-tampering and fake-organic claims, foreign border agencies quietly mark up your risk profile. The enhanced checks abroad and the adulteration raids at home are two ends of the same credibility problem.

And underneath it all sits the real question: who actually owns export quality? On paper, everyone. In practice, no one. State governments run primary agriculture, the food authority sets domestic safety, APEDA promotes crop exports, the marine authority governs seafood, Plant Quarantine issues phytosanitary certificates, Customs signs the exit. Every silo does its slice; none owns the outcome. The exporter, aggregating produce from hundreds of untraceable micro-farms, ends up a collector of paperwork rather than an architect of quality. When Brussels tightens a residue limit, the signal has to crawl back through that maze to a village pesticide dealer, and by then the container is already destroyed.

What to watch

Whether India wins a one-time Australian exemption for the ₹200 crore of basmati already in transit, using the consultation framework in the India-Australia trade agreement.⁵
Whether Japan restores mango access once the Rehmanpur treatment facility is re-audited.⁷
Whether pre-harvest testing actually pulls Andhra Pradesh's antibiotic rejection share down over the next two shrimp cycles.^8,12
Whether APEDA's farm-to-port traceability moves from pilot to scale before UK CETA volumes ramp up.⁴

India has learned to open the world's doors. It has not yet learned to walk its containers through them.

And if you want to understand where China stands and what it means for India, do check out my companion newsletter, Decoding the Dragon: https://t.ly/t7uhs

And, do check out my work on the following platforms as well: Instagram, LinkedIn and Youtube

Best,
Jayant

References

#36: Why has India built only 217 of the PROMISED 5,000 biogas plants?

Jayant Mundhra — Thu, 13 Aug 2026 01:30:22 GMT

I have been staring at two numbers all week and I cannot get them to sit comfortably together.

In October 2018, India launched SATAT, the Sustainable Alternative Towards Affordable Transportation programme, and promised 5,000 compressed biogas plants producing 15 MMT of gas a year by 2023-24.¹

As of 6 August 2026, two years past that deadline, India has 217 commissioned plants producing 0.4 MMSCMD.²

That is about 4% of the plant target. On volume it is under 1%.

And on that exact same day, the Union Cabinet approved GOBARdhan, the National Circular Bioenergy Scheme, with ₹23,731 crore behind it over ten years.³

So the question I keep coming back to is not whether biogas matters for India. It obviously does. The question is why the last promise did not land, and whether ₹23,731 crore actually fixes the thing that broke.

What was actually approved

Compressed biogas, or CBG, is simple in principle. You put cattle dung, paddy straw, sugar mill press mud or wet municipal waste into a sealed digester, let bacteria eat it, and you get a gas that is 55% to 65% methane. Scrub out the carbon dioxide and the hydrogen sulphide, push methane purity past 95%, and you have a fuel chemically identical to CNG that runs in the same engines and the same pipelines.

The residue left behind, Fermented Organic Manure or FOM, is a fertiliser.

GOBARdhan does not invent any of this. What it does is collapse five separate programmes into one. SATAT, the Market Development Assistance for organic manure, the Biomass Aggregation Machinery scheme, the Development of Pipeline Infrastructure scheme, and the older capital assistance window now sit under a single roof at the Ministry of Petroleum and Natural Gas.⁴

The mechanics that matter:

A legally enforceable blending mandate. City gas distribution companies must blend 3% CBG into CNG for transport and PNG for homes in FY2026-27, 4% in FY2027-28, and 5% from FY2028-29 onward.³
A fixed administered price of ₹2,110 per MMBTU, roughly ₹105 per kg, locked for a minimum of ten years.⁵ That is a jump of about 43% on the ₹1,478 per MMBTU (₹77.4 per kg) that producers were getting in mid-2025.⁶
Capital assistance of up to ₹2 crore per tonne per day of installed capacity for greenfield plants.³
₹1,500 per tonne of Market Development Assistance on Fermented Organic Manure.⁵
An 85% credit guarantee on loans to MSME-scale CBG projects.²

The government's own projection over the decade: ₹40,000 crore of import substitution, 1.5 lakh jobs, ₹75,000 crore added to GDP, 250 MMT of organic fertiliser and a 40 MT cut in carbon dioxide.²

Read that list and it looks like the sector's every complaint has been answered at once. Which is exactly why it is worth asking what went wrong the first time.

Why 3,694 letters of intent became 217 plants

Here is the interesting bit. The bottleneck was never demand for the gas, and it was never enthusiasm from entrepreneurs.

By October 2022 the oil marketing companies had already handed out 3,694 letters of intent.¹ Today 1,908 plants are formally registered. Only 217 are commissioned. Another 339 are under construction. And 1,352 have simply not started.²

The failure happened after the paperwork, at the plant gate. Four things killed it:

Feedstock is not a commodity, it is a logistics problem. Wet organic waste has low bulk density and high moisture, so hauling it beyond about 25 to 30 km destroys the plant's margins. Every CBG plant is therefore trapped inside a tiny catchment, and must aggregate from thousands of scattered smallholders. Press mud is available roughly six months a year. Dung prices in one surveyed plant rose 40%.⁷
So plants ran half empty. Indian CBG facilities operate at 20% to 60% of designed capacity. European plants routinely clear 80%.⁸
The gas had nowhere to go. Only about 15% of functional Indian plants are connected to a city gas distribution network.⁸ The rest truck compressed gas in steel cascades on diesel lorries, which is expensive and quietly undercuts the whole carbon argument.
The manure was a liability, not a revenue line. Every 10 tonnes of gas leaves behind roughly 25 tonnes of solid digestate and 80 tonnes of liquid.⁸ Without a market for it, plants physically clog.

Add to that a developer having to walk eight government departments that rarely coordinate, and officials rotating out every two years, and you get 1,352 registered projects that never broke ground.⁸

This is the pattern I keep running into across Indian infrastructure, where the announcement is federal and the failure is local, and it is the kind of thread we pull apart most mornings in the BharatNama WhatsApp group, where I share a smaller India deep-dive like this with thousands of readers every day (t.ly/h2jq1).

Join the BharatNama community

The cattle problem hiding inside the energy problem

There is a second story buried in this scheme, and it is the one I find genuinely elegant.

India's last published livestock count, the 20th Livestock Census of 2019, put the bovine population at about 303mn animals, of which roughly 5mn were stray.⁹ The 21st Census was enumerated through 2024 and 2025 and its results are still not out, so 2019 remains the newest official number we have.

The cause is not sentiment, it is mechanisation. Tractors made bullocks commercially useless. Slaughter restrictions mean a farmer cannot sell a non-productive animal. Fodder still costs money. So the animal gets released onto the road, where it eats standing crops and plastic and causes accidents.

CBG changes the arithmetic of owning that animal. India generates roughly 3mn tonnes of cattle dung a day, enough in theory for about 180mn m³ of biogas daily.⁷ A gaushala that can sell dung to a digester stops being a charity line item and starts being a feedstock supplier.

That is the part legislation alone has never managed to solve, and a fuel-blending mandate might.

Where India actually stands

Now the comparison that matters, and it is not flattering.

India consumed 197 MMSCMD of natural gas in June 2026, with 56% of that imported as LNG and domestic output flat at 87 MMSCMD.¹⁰ CBG contributes 0.4 MMSCMD. That is about 0.2% of India's gas.

Set that beside the countries that took biomethane seriously:

Denmark: biomethane is about 40% of total gas consumption, up from roughly 10% in 2019, with a target of a fully green gas grid by 2030. It does this with just 173 biogas plants, about 55 to 60 of which upgrade to grid quality.¹¹
United States: the world's largest biomethane producer at roughly 136 PJ, having grown 2.2 times since 2020.¹²
Germany: the largest combined biogas and biomethane market on earth at 329 PJ in 2024, though output has been broadly flat since 2017.¹²
India: the IEA expects combined Indian biogas and biomethane output to grow just 21% between 2024 and 2030.¹²

And the theoretical Indian potential? About 62 MMT of CBG a year. Current output is below 1% of it.¹³

Now break India open against itself, because the national number hides the real story. Of the 217 commissioned plants:

Uttar Pradesh: 40
Gujarat: 25
Maharashtra: 24¹⁴

Three states hold roughly 40% of India's CBG fleet. Until recently only four states, Uttar Pradesh, Haryana, Bihar and Gujarat, had any bioenergy policy supporting CBG at all.¹⁵ Maharashtra only notified its own CBG policy on 5 May 2026, with viability gap funding of up to ₹75 lakh per tonne of capacity, capped at ₹15 crore a project, and ₹500 crore allocated for FY2026-27.¹⁴

So the binding constraint is not central money. It is that most Indian states have no CBG policy, no land norms, no single-window clearance and no feedstock aggregation machinery. A central price guarantee does not build a district-level dung supply chain.

What India can learn from Denmark

Denmark is the right teacher here, and not for the reason people usually assume. It is not that the Danes are greener. It is that they solved precisely India's feedstock problem.

About 75% of Danish biogas feedstock is livestock manure and slurry.¹¹ Denmark then did four specific things:

It mandated the manure. A national target that 50% of livestock manure be processed for biogas by 2025 turned a waste-disposal obligation into guaranteed plant feedstock.¹¹
It guaranteed grid access. A right to inject gives biomethane producers a standardised, non-negotiable connection to the gas grid.¹¹ Compare that with India's 15% CGD connection rate.
It gave 20-year contracts. Since 2024, competitive tenders award two-decade production contracts to the lowest bidder, which is what banks actually lend against.¹¹
It capped energy crops at 4% of feedstock from 2025, deliberately keeping the industry on waste rather than on farmland.¹¹

India has now copied the price certainty and the offtake mandate. It has not copied the manure mandate or the right to inject, and those are the two that solve the exact problems that killed SATAT.

What does not transfer is scale. Denmark runs 173 plants for 6mn people in a compact country with an existing gas grid. India needs thousands of plants across a subcontinent where the gas grid does not reach most of the catchments that have the feedstock. That is why the Development of Pipeline Infrastructure component, at ₹994.5 crore, may matter more per rupee than the headline capital subsidy.⁵

The money has already moved

The corporates are not waiting to find out.

Reliance has put in ₹6,500 crore, has 55 CBG plants in execution of which 41 are operational, and reckons it accounts for about a quarter of India's CBG ecosystem. Its stated long-term target is 500 plants.¹⁶ It broke ground on the first of a planned 500 plants in Andhra Pradesh in April 2025 under a ₹65,000 crore state commitment.¹⁷
Adani Total Gas is building a 500 TPD bio-CNG plant on Ahmedabad's Pirana dumpsite for the municipal corporation, on a 20-year PPP at about ₹130 crore.¹⁸
GAIL acquired a 49% stake in Leafiniti Bioenergy, TruAlt's CBG subsidiary, in March 2026.¹⁹

Two regulatory changes quietly de-risked all of this. In March 2025 the Fertiliser Control Order was amended to create a whole new category, Organic Carbon Enhancer, formally recognising FOM and its liquid form as fertilisers.²⁰ And in April 2025 the CPCB created a Blue Category for industries providing essential environmental services, which lets qualifying waste-to-energy and CBG plants run on a two-year extended consent instead of re-applying constantly.²¹

The fertiliser angle is the one that should interest the finance ministry most. India budgeted ₹1.71 lakh crore for fertiliser subsidy in FY2026-27, and the Department of Fertilisers has said the bill could exceed ₹3 lakh crore if West Asian supply disruption persists, against about ₹2 lakh crore in FY2025-26.²² India imported over 100 lakh tonnes of urea last fiscal year. Every tonne of quality FOM that displaces imported urea is a rupee the exchequer does not spend.

This is the sort of second-order link I think gets badly under-covered in Indian business writing, and it is the kind of thing I unpack daily in the BharatNama community (t.ly/h2jq1).

Join the BharatNama community

Why CBG and not hydrogen

One last piece of context, because the obvious objection is that India should be building for hydrogen instead.

Green hydrogen in India currently clears competitive tenders at about ₹397 per kg for supply to Indian Oil's refineries and ₹387 per kg for BPCL and HPCL, including GST. Against a 2030 target of 5 MMT a year, commissioned capacity stands at 8,000 tonnes, which is 0.16% of the target.²³

CBG sells at roughly ₹105 per kg, needs no new engine, no new pipeline standard and no new refuelling network. For heavy trucking and intercity buses over the next decade, that is not a compromise, it is the only thing that actually works today.

What to watch

Whether the CBO is enforced or quietly relaxed. The 3% obligation bites in FY2026-27. Watch whether city gas distributors are penalised for missing it, or granted a waiver on grounds of unavailable supply.
The conversion rate of the 1,352 not-started projects. That number, not the announcement, is the real scoreboard. Track it against the 339 currently under construction.
Whether more states follow Maharashtra. Watch for CBG policies with viability gap funding from Karnataka, Tamil Nadu, Punjab and Andhra Pradesh over the next two quarters.
Whether FOM actually sells. The ₹1,500 per tonne MDA only works if the Organic Carbon Enhancer category generates real farm demand. Watch FOM offtake volumes, not production.
The 21st Livestock Census release. It will tell us whether the stray cattle population has grown since 2019, and therefore how big the dung feedstock base really is.

India did not fail at biogas because it lacked cattle, straw or ambition. It failed because it wrote a target in Delhi and then asked a district to build the supply chain for free.

And if you want to understand where China stands and what it means for India, do check out my companion newsletter, Decoding the Dragon: https://t.ly/t7uhs

And, do check out my work on the following platforms as well: Instagram, LinkedIn and Youtube

Best,
Jayant

References

1. PIB, Targets under SATAT Scheme, pib.gov.in
2. IANS, GOBARdhan scheme to cut imports by Rs 40,000 crore, create 1.5 lakh jobs, ianslive.in
3. PMIndia, Cabinet approves GOBARdhan, India's National Unified Scheme for Compressed Biogas, pmindia.gov.in
4. Business Standard, Cabinet clears ₹23,731 crore GOBARdhan scheme for compressed biogas, business-standard.com
5. Drishti IAS, GOBARdhan: National Circular Bioenergy Scheme, drishtiias.com
6. iamRenew, Centre raises CBG procurement price to Rs 1,478/mmBtu, iamrenew.com
7. Global Methane Initiative, Methane Mitigation: Lessons Learned from India's CBG sector, globalmethane.org
8. Autocar Professional, Why India's CBG revolution is stuck in neutral, autocarpro.in
9. PIB, Livestock Census, pib.gov.in
10. Asianet Newsable, India's gas consumption recovers, import dependence hits 56%, asianetnews.com
11. IGEM, Denmark's path to 100% green gas, Country Brief, igem.org.uk
12. IEA, Renewables 2025, Biogases, iea.org
13. Prokerala, India's clean fuel transition opens Rs 1 lakh crore biogas opportunity, prokerala.com
14. Down To Earth, Maharashtra State Compressed Biogas Policy 2026, downtoearth.org.in
15. CSE, Developing an effective state-level compressed biogas policy in India, cseindia.org
16. ANI, Reliance at forefront of CBG push with Rs 6,500 crore investment, 55 plants, aninews.in
17. Business Standard, Reliance lays foundation for first of 500 CBG plants in Andhra Pradesh, business-standard.com
18. Free Press Journal, Adani Total Gas receives Bio-CNG plant order from Ahmedabad Municipal Corporation, freepressjournal.in
19. Business Standard, GAIL to acquire 49% stake in Leafiniti Bioenergy, business-standard.com
20. Rural Voice, Organic manure from CBG plants recognised as new fertiliser category, ruralvoice.in
21. Drishti IAS, Blue Category for Industries, drishtiias.com
22. The Pioneer, Fertiliser subsidy to cross Rs 3 lakh crore in FY27, dailypioneer.com
23. Business Today, India's green hydrogen ambitions face a reality check, businesstoday.in

#35: Why does a Gulf war threaten 23% of Kerala's income?

Jayant Mundhra — Tue, 11 Aug 2026 01:43:45 GMT

Here is a number that should not really be possible for an Indian state. The money Keralites working abroad sent home in 2023 came to ₹2.17 lakh crore, which is 23.2% of the state's income.¹

Now put that beside a country instead of a state, and it lands. Nepal, the textbook remittance economy, runs at about 26% of GDP.² Kerala is sitting right next to it. India as a whole is at just 3.5%.²

So on this one metric Kerala does not really behave like a part of India. It behaves like a small nation that exports its people and lives on what they wire back. That model has worked beautifully for half a century. This year, a war is quietly asking whether it still works.

A state that runs on other people's wages

The mechanics are worth stating plainly, because they are unusual.

Kerala does not earn its keep by making things. It manufactures very little, imports most of what it eats, wears and builds, and pays for those imports with foreign wages. Roughly 2.2mn Keralites work overseas,¹ about one in every fifteen people in the state, and their money has quietly underwritten the hospitals, the schools and the property market that the "Kerala model" is celebrated for.

Two features make this fragile.

It is dangerously concentrated. 80.5% of Kerala's emigrants are in the six Gulf states, and 38.6% are in the UAE alone.¹ There is almost no geographic hedge.
It is exposed to precisely one thing going wrong: instability in the Gulf.

That is the thing now going wrong. The 2026 war between Israel, the United States and Iran has flared through this year, with a fragile ceasefire that keeps breaking down into fresh strikes across the region.³ For most of India this is an oil-price story. For Kerala it is an income story, and a far more direct one.

The people who track this closely, led by the migration economist S. Irudaya Rajan, expect remittances into Kerala to fall by around a fifth this financial year if the war drags on.⁴ The state's new Chief Minister has put the same number on record. V.D. Satheesan, who took charge in May 2026 after the Congress-led UDF ended the Left's ten-year run,⁵ told an interviewer in July that he expects a 20% dip in remittances this year, called them "the backbone of the state economy", and confirmed he has gone to Delhi asking for special assistance to ride it out.⁶

A 20% fall does not sound apocalyptic until you remember what those remittances actually do.

How far Kerala has travelled, in the wrong direction

The instinct is to assume Kerala has been slowly weaning itself off the Gulf. The data says the opposite. The dependence has deepened.

Remittances into Kerala went from ₹85 thousand crore in 2018 to ₹2.17 lakh crore in 2023, a rise of about 155% in five years.¹
As a share of the state's income, they climbed from 13.5% to 23.2% over the same period.¹

In other words, in the years when everyone assumed Kerala was diversifying, the state actually became more hooked on foreign wages, not less. The Gulf money flowed in, went first into repaying the debt that funded the migration, and then into land, houses and gold, inflating an asset base that has very little to do with what Keralites at home actually earn.

This is exactly the kind of thread I keep pulling at, and it is the sort of thing we go back and forth on in the BharatNama WhatsApp group, where I share a smaller India deep-dive like this most mornings (t.ly/h2jq1).

Join the BharatNama community

The composition of who leaves has shifted underneath all this, and that shift is the real story of the next decade.

Student emigration from Kerala doubled, from about 130 thousand in 2018 to 250 thousand in 2023, and students are now 11.3% of all emigrants.¹
Among women who leave, 51.6% go as nurses, and 40.5% of them now prefer Europe and the West over the Gulf.¹

The young and the highly qualified are increasingly skipping the Gulf entirely for Canada, the UK, Australia and Germany, where there is a path to residency and a life, not just a contract. That is good for those families. It is quietly terrible for the remittance model, because a nurse who settles in Toronto with her family stops wiring the bulk of her salary back to Kochi in a way a construction supervisor in Dubai never did.

Where India stands: against the world, and against itself

Two comparisons make the position clear, and both are built on current, like-for-like data.

India against the world. On remittance dependence, Kerala is an outlier not just within India but globally:

Tajikistan: about 48% of GDP.²
Nepal: about 26%.²
Kerala: 23.2% of state income.¹
Philippines: about 9%.²
India as a whole: 3.5%.²

Kerala belongs on the list of economies that survive on their diaspora, sitting beside Nepal, far above the Philippines, while the country it is part of barely registers. That is a genuinely strange place for one of India's richest, most literate states to be.

India against itself. Here is the sharper contrast. India's remittance base has already pivoted away from the Gulf. Kerala has not.

For India as a whole, the United States is now the single largest source of remittances at 27.7%, having overtaken the UAE at 19.2%, according to the RBI's latest survey.⁷
Advanced economies, the US, UK, Singapore, Canada and Australia together, now send more than half of all money remitted to India, while the Gulf's share has slipped to 38%.⁷
Kerala, meanwhile, is still 80.5% dependent on those same Gulf states.¹

So the rest of India has, almost accidentally, diversified its way to safety through white-collar migration to the West. Kerala is still carrying the blue-collar Gulf model that India as a whole is walking away from. When a war hits the Gulf, the country shrugs and Kerala shakes.

The fiscal scissor closing on the state

None of this would matter so much if Kerala had room in its budget to absorb a shock. It does not. It is caught in what economists call a scissor effect: revenue is being squeezed from one side while spending obligations are fixed on the other.

Start with the spending blade. Kerala's 2026-27 budget shows a state with almost no financial flexibility left:

Committed expenditure, the money it has no choice but to spend, is 72% of revenue receipts, split across salaries (31%), pensions (21%) and interest payments (20%).⁸
Interest alone eats one in every five rupees the state earns.⁸
Capital spending, the money that actually builds the future, is stuck near 1.2% of GSDP, among the lowest in India.⁸
The revenue deficit is 2.2% of GSDP and outstanding debt is 34% of GSDP.⁸

Now the revenue blade, and this is the fresh cut. When a state consumes less, its tax take falls, because Kerala's revenue is overwhelmingly consumption-based (it makes little to tax at the factory gate). A 20% drop in remittances feeds straight through into weaker spending on cars, gold and construction, and therefore into weaker GST. The income that funds the state and the income that gets taxed are the same Gulf income.

That is the trap. The shock that cuts Kerala's revenue is the same shock that would force it to spend more, on returning, jobless migrants who need rehabilitating.

The federal squeeze, and a case that could redraw the map

Faced with a revenue hole, any government would normally borrow. Kerala's problem is that it cannot, and the reason has become a landmark constitutional fight.

Two things happened at once. First, the Sixteenth Finance Commission, whose award runs from 2026, raised Kerala's share of central taxes slightly, from 1.925% to 2.382%, but scrapped the revenue deficit grants that had cushioned the state.⁹ Kerala had drawn ₹37,814 crore in such grants over the previous five-year cycle; the new figure is zero.⁹ The Chief Minister called it "a huge shock to the state exchequer".⁶ Kerala gains on paper and loses in cash.

Second, the Centre tightened the borrowing rules. States can normally borrow up to 3% of GSDP under the fiscal-responsibility framework. But the Union began counting the off-budget borrowings of state entities, most importantly the Kerala Infrastructure Investment Fund Board (KIIFB), against that same ceiling, shrinking the room Kerala actually has to raise money.

Kerala's response was extraordinary. In December 2023 it sued the Union of India in the Supreme Court, arguing that borrowing is a state subject and that Delhi was strangling its fiscal autonomy. In April 2024 the court refused Kerala interim relief of ₹26,226 crore in borrowing space, but it also referred the deeper questions to a five-judge Constitution Bench, because Article 293, which governs how states borrow, has never been properly interpreted in the country's history.¹⁰ That reference is still pending, and whenever it is decided it will settle, for every Indian state, exactly how much fiscal freedom the Constitution actually grants.

This is the part I find genuinely gripping, and it is the kind of structural, second-order story we dig into daily in the BharatNama community rather than just the headline (t.ly/h2jq1).

Join the BharatNama community

What India can learn: the Philippines figured this out

India does not have to invent an answer here. The Philippines, the other great labour-exporting economy, faced the same dependence decades ago and built its way out of the worst of it.

It diversified geography relentlessly. Filipino workers are spread across more than a hundred countries, so no single war or oil crash can take down the whole flow. Kerala's 80.5% concentration in one region is the opposite of this.
It built institutions around migration, culminating in a full Department of Migrant Workers, with mandatory pre-departure training, contract vetting and reintegration support for returnees.
It kept remittances at a manageable share of GDP, about 9%,² precisely because the domestic economy, especially business-process outsourcing, grew fast enough to give people an option at home.

The transferable lesson for Kerala and for India is not "copy the Philippines". It is narrower and harder: diversify the destinations, formalise and protect the skilled migration that is already happening (the ILO and Kerala's own NORKA agency began exactly this work on nurses and care workers in 2026), and above all build enough of a domestic economy that a returning migrant has somewhere to land. Kerala has the human capital; what it has never built is the local job.

What to watch

Whether the 2026 Gulf ceasefire holds, and whether Gulf construction and hiring restart. This is the single biggest swing factor for Kerala's FY27.
Kerala's own remittance and bank-deposit data through the year, to see if the projected 20% fall actually shows up.
Whether the Centre grants Chief Minister Satheesan the special assistance he has asked for, and on what conditions.
Whether the Supreme Court's Constitution Bench finally takes up the Article 293 reference, the outcome of which binds every state.

Kerala spent fifty years proving that you can build a rich, healthy, literate society without building an economy. The bill for that shortcut is now arriving in the same envelope as a war.

And if you want to understand where China stands and what it means for India, do check out my companion newsletter, Decoding the Dragon: https://t.ly/t7uhs

And, do check out my work on the following platforms as well: Instagram, LinkedIn and Youtube

Best,
Jayant

References

Kerala Migration Survey 2023, International Institute of Migration and Development (IIMAD) / GIFT: iimad.org
Remittances as % of GDP (World Bank data), TheGlobalEconomy: theglobaleconomy.com
2026 Iran war ceasefire and July 2026 escalation, Al Jazeera: aljazeera.com
Kerala likely to see 20% drop in remittances as West Asia war drags on (S. Irudaya Rajan), FLAME University / GIFT: flame.edu.in
Kerala Assembly election results 2026, UDF returns to power, Business Standard: business-standard.com
West Asia war a blow to Kerala's remittance-backed economy: CM Satheesan (interview, July 2026), Business Standard: business-standard.com
Advanced economies overtake Gulf nations in share of remittances to India (RBI 6th Remittances Survey, 2023-24), Business Standard: business-standard.com
Kerala Budget Analysis 2026-27, PRS Legislative Research: prsindia.org
Why Kerala gains on paper but loses fiscally under the 16th Finance Commission, The Federal / CSES: thefederal.com
Extent of state's power to borrow funds from Union (Article 293 case status), Supreme Court Observer: scobserver.in

#34: Why are 9 in 10 Indian workers still informal?

Jayant Mundhra — Thu, 06 Aug 2026 01:31:28 GMT

Here is a number that should stop any celebration of India's rise cold. After three decades of world-beating growth, roughly 9 in 10 Indians who work still do so informally.¹ No written contract. No paid leave. No provident fund. Nothing between them and a bad month.

I keep coming back to this because it quietly breaks the story we tell ourselves. We have the GDP. We have the unicorns. We have UPI rails clearing more digital payments than any system on earth. And yet the typical Indian worker is about as exposed today as their parent was a generation ago.

The thing is, this is not an accident of bad luck. It is the direct result of the particular road India took to get richer. We grew fast, but we skipped the one phase that usually turns growth into secure work.

First, what "informal" actually means

Two ideas get muddled here, so let me separate them cleanly.

The informal (or unorganised) sector is about the enterprise: tiny, unregistered firms that sit outside the tax, legal, and regulatory net. About 82% of India's workforce sits in it.¹
Informal employment is about the worker: no written contract, no social security, no paid leave, whoever the employer is. Nearly 90% of Indians are informally employed.¹

That second gap is the important one, because informal employment exists even inside formal, registered companies. The contract cleaner in a gleaming IT park and the delivery rider working for a listed, deep-pocketed platform are both informally employed, working for very formal firms.

And this is not a fringe of the economy. The unorganised sector still produces close to 45% of India's GDP.² This is the water most Indians swim in, not a puddle at the edge.

Look at the shape of who works and how, from the latest Periodic Labour Force Survey, the government's official employment survey:

Self-employed: 56.2% of all workers in 2025, the largest slice by far.¹
Regular wage or salaried: 23.6%.¹
Casual labour: 20.2%.¹

Even that "regular" fifth is misleading. Most of them still have no contract and no benefits. Genuine formal, protected jobs are a rounding error, not the norm.

We took the services lift and skipped the factory floor

To see why we are stuck, go back to the baseline. Around 1991, agriculture employed close to 60% of Indians. By 2025 that is down to 43%.¹ Vast numbers left the farm. The question is where they went.

The textbook path, the one every East Asian success story followed, runs farm to factory to formal. Surplus farm labour moves into labour-intensive manufacturing, and the factory itself does the formalising: it needs payrolls, contracts, shift rosters, and eventually provident funds and unions.

India did something unusual. It jumped straight from farms to services, largely bypassing the factory phase. Economists call it "premature de-industrialisation", ageing out of manufacturing before ever fully industrialising.

The tell is in one stubborn number. Manufacturing's share of employment has barely moved in a decade, sitting at just 12.1% in 2025.¹ It simply never became the sponge that soaked up the people leaving agriculture.

Services grew instead, and grew spectacularly. But modern services, IT, finance, professional work, are capital-heavy and skill-heavy. They cannot absorb the enormous numbers of workers with modest schooling. So the labour leaving farms did not walk into air-conditioned offices. It walked into the low-productivity end of services: petty retail, security, domestic work, construction, and now app-based delivery.

This is the thread I find myself pulling on again and again, and it is exactly the kind of structural puzzle we chew over in the BharatNama WhatsApp community, where I share a smaller India deep-dive like this with thousands of readers every morning (t.ly/h2jq1).

Join the BharatNama community

The missing middle, and why firms refuse to grow up

If services could not absorb low-skill labour, why did manufacturing not step up? Because the incentives actively punish scale.

India's factory landscape is barbell-shaped. A vast ocean of tiny firms at one end, a handful of giants at the other, and almost nothing in the middle. Roughly three-quarters of enterprises employ fewer than ten people. These micro-firms run on thin margins and have neither the money nor the reason to put workers on formal payrolls.

Why stay small on purpose? Because growing up is expensive and frightening:

A growing enterprise can face over 1,450 compliance obligations a year, costing around ₹13 lakh annually, with a chunk carrying the risk of jail for procedural slips.³
Formal credit is scarce. Only about 14% of MSMEs can access formal, institutional credit at all.⁴
The sector's credit gap is roughly ₹25 lakh crore, so the rest borrow from moneylenders at punishing rates or stay stunted.⁴

Faced with that, thousands of firms make a rational choice: stay below the radar. Stay informal. Which means their workers stay informal too.

When even the formal sector goes informal

Here is where it gets genuinely uncomfortable. Even when the organised sector does grow, it increasingly refuses to create secure jobs. It hires contract workers instead of permanent ones.

The number is striking, and it is current. Contract workers are now 42% of the organised manufacturing workforce, the highest share since 1997-98.⁵ Go back to 1999-2000 and it was around 20%. It has doubled.⁵

The usual explanation is that rigid labour laws make it hard to fire permanent staff, so firms hire contract labour to stay flexible. But deeper plant-level work by economists Radhicka Kapoor and P.P. Krishnapriya finds a sharper motive: contract labour is used to dilute the bargaining power of permanent workers and keep overall wages in check.⁶

Tellingly, contractualisation kept rising even as the wage gap between contract and permanent workers narrowed. If it were only about cheap labour, that should have slowed it. It did not. This is about control, not just cost. The formal sector has learned to reproduce informal conditions inside its own gates.

And you can see the result in pay. The real monthly wage of a regular salaried worker actually fell to about ₹10,925 in 2022, from ₹12,100 in 2012.⁷ A decade of headline growth, and the typical salaried Indian was earning less in real terms at the end of it.

Where India stands, against the world and against itself

This is the part that matters most to me, because a number for India alone is just a fact. Set beside others, it becomes an insight.

India versus the world, on the metric that drives formalisation, factory jobs:

India: manufacturing employs ~12% of workers.¹
Vietnam: closer to a fifth, around 21%, and rising every year.⁸
South Korea, at its industrial peak, took manufacturing to about 25% of employment while farm work collapsed from 50% to under 20%.⁹

Same starting point, roughly. Poor, agrarian, labour-rich. Wildly different destinations. Vietnam and Korea pulled their farm workers onto factory floors that formalised them by design. India did not build enough of those floors.

India versus India is just as revealing, and here the disparities and the commonalities both bite:

Even in the richest western and southern states, three in four workers still sit outside the formal sector.¹⁰ Formality is thin everywhere, not just in poor states.
Unemployment ranges from 0.9% in Gujarat to 8.3% in Goa.¹ One national average, many different countries inside it.
Earnings vary most in self-employment across states, far more than in regular salaried work.¹⁰ The people with the least protection also face the widest spread in what they take home.

What India can learn from the ones who cracked it

I am wary of lazy "India should just copy X" takes, so let me be specific about what actually transfers and what does not.

Vietnam is the closest parallel. It was agrarian and poor in the 1980s. Then, from the 1986 Doi Moi reforms, it slashed tariffs on manufactured goods from about 16.6% to 1.1%, threw the doors open to foreign investment, and became a factory to the world. South Korea alone has poured in over $92bn and is now Vietnam's largest foreign investor.⁸ Those big, compliance-bound multinationals hire on formal contracts by default.

Then Vietnam went further and legislated. Its new Social Insurance Law took effect on 1 July 2025, pulling part-time workers, unsalaried managers, and long-contract foreigners into compulsory social insurance, and reclassifying deliberate non-payment as "evasion" that can trigger criminal proceedings.¹¹

South Korea did it a generation earlier, with the 1973 Heavy and Chemical Industry drive, channelling capital into large industrial groups that scaled fast and formalised as they grew.⁹

What is portable for India: the sequence of scale first, enforcement second. Build firms big enough to formalise, then legislate to lock it in. India has laid the groundwork with e-Shram, its registry of unorganised workers, now carrying over 30.9 crore registrations, 53.68% of them women.¹² Registration is real progress. But a name in a database is not a provident fund. The next step is portable social security that a migrant carries across state lines and between gig apps without losing cover.

What does not transfer neatly: India's federal structure, its tighter fiscal room, and a state that is brilliant at big-bang digital rollouts but patchy at the daily grind of enforcing labour law across a sea of tiny firms. This is exactly the gap between formalising the economy and formalising the worker, and it is the thread I most enjoy arguing out with readers in the BharatNama community every day (t.ly/h2jq1).

Join the BharatNama community

What to watch

Whether the four new labour codes are actually implemented, not just notified, and whether they widen social security or simply ease hiring and firing.
Whether e-Shram registrations convert into real, portable benefits, or stay a database.
The manufacturing employment share in the next PLFS. If it does not break clearly above 12%, the structural problem is unchanged.
The contract-labour share in the next Annual Survey of Industries. A move past 42% means the formal sector is informalising further, not less.

The puzzle of India is not that it failed to grow. It is that it grew rich in output while staying poor in the one thing that makes growth felt in a household: a secure job.

And if you want to understand where China stands and what it means for India, do check out my companion newsletter, Decoding the Dragon: https://t.ly/t7uhs

And, do check out my work on the following platforms as well: Instagram, LinkedIn and Youtube

Best,
Jayant

References

Periodic Labour Force Survey (PLFS) Annual Report 2025, PIB / MoSPI: https://www.pib.gov.in/PressReleasePage.aspx?PRID=2246009
Informal sector's share of India's GDP, Data For India: https://www.dataforindia.com/informal-sector-sizing/
MSME compliance burden, TeamLease RegTech 2025: https://www.teamleaseregtech.com/msme-compliance-report/
MSME credit gap and formal credit access, Deloitte via YourStory: https://yourstory.com/2025/05/msme-credit-gap-india-financing-challenges-2025
Contract labour share at 42%, highest since 1997-98 (ASI 2023-24), Business Standard: https://www.business-standard.com/economy/news/on-the-rise-share-of-contract-labour-at-42-highest-in-over-25-years-125090801010_1.html
Explaining the contractualisation of India's workforce, Kapoor & Krishnapriya (ICRIER): https://icrier.org/pdf/Working_Paper_369.pdf
Real wages of salaried workers declined 2012-2022, ILO India Employment Report 2024 via Business Standard: https://www.business-standard.com/economy/news/real-wage-of-salaried-workers-dipped-in-2012-2022-period-ilo-report-124040100999_1.html
Vietnam manufacturing employment and South Korea's $92bn investment: https://en.baochinhphu.vn/south-koreas-investment-in-viet-nam-hits-us92-bln-111250515092839514.htm
South Korea's industrialisation, Korea Economic Institute: https://keia.org/the-peninsula/south-koreas-important-role-in-vietnams-rapid-development/
State-level employment disparities, PLFS 2025 analysis, SBI Research: https://sbi.bank.in/documents/13958/14472/08052026_PLFS_SBI+RESEARCH.pdf
Vietnam's 2024 Social Insurance Law, effective 1 July 2025, Vietnam Social Security: https://vss.gov.vn/
e-Shram registrations over 30.68 crore, 53.68% women, PIB: https://www.pib.gov.in/PressReleasePage.aspx?PRID=2109828

#33: Why does 41% of Mumbai still live in slums?

Jayant Mundhra — Tue, 04 Aug 2026 01:31:34 GMT

I want to start with a number that should not be possible. Mumbai generates roughly a third of India's income tax and around 40% of its foreign trade¹, and yet about 41% of the city lives in slums.² That is close to eight million people, in the richest city in the country, packed into settlements with no proper roads, no sanitation, and barely any light.

The easy explanation is poverty, or migration, or corruption. I believed some version of that for years. But the deeper I go into how Indian cities are actually zoned, the more I think we did this to ourselves, deliberately, with a single number written into a rulebook.

That number is the Floor Space Index. And the whole tragedy of Indian urbanism sits inside one misunderstanding of it: capping floor space does not cap people. It only caps housing.

What a single number controls

The Floor Space Index (FSI), known globally as the Floor Area Ratio, is simply the ratio of the total built-up floor area you are allowed to construct to the size of the plot.

An FSI of 1 on a 1,000 sq m plot means 1,000 sq m of floor, full stop. One tall tower on a small footprint, or one flat building sprawled across the whole plot, but never more total floor than that.
Mumbai wrote FSI into its very first development plan in 1964, and for decades held the base at 1.33 for the island city and 1.0 for the suburbs.³
The logic was Malthusian: cap the floor, and you cap how many people and firms can crowd onto the existing water pipes, drains, and streets.

Here is the flaw, and it is fatal. People do not move to Mumbai because floor space is available. They move because the jobs, the capital, and the opportunity are there. The migration is driven by wages, not by building rules.

So when you cap the legal floor, the people arrive anyway. The demand does not vanish. It simply gets pushed out of the formal, legal market, because in any market where you freeze supply, the rich outbid everyone else for the little that exists. The clerk, the delivery rider, the nurse, the construction worker: none of them can win that auction. So they build their own housing, informally, off the books, where no FSI rule can reach them.

That is what a slum is. Not a failure of the poor. A workaround for a supply cap.

A city that fought its own geography

Mumbai is the sharpest case in the world because geography already dealt it a terrible hand, and then policy made it worse.

Within a 25 km radius of the old city centre, roughly 66% of the surface is water. For Jakarta that figure is 22%, for Seoul just 5%.⁴ Mumbai has less than half the developable land of a normal coastal megacity before a single rule is applied.

Every other land-starved city on earth responded to that constraint by building up.

Manhattan zones its prime districts for a floor-area ratio as high as 15.⁵
Central Tokyo permits ratios up to 13 in its densest commercial core around Marunouchi.⁶
Singapore, an island as cramped as Mumbai, allows floor ratios climbing into the 20s in its central nodes, and still ends up with more greenery, because taller towers on smaller footprints free the ground.⁷

Mumbai, facing the worst land shortage of the lot, did the opposite. It held its base FSI at 1.33. A city that should have been the densest vertical metropolis in Asia was legally required to sprawl.

The human cost shows up in one brutal statistic. The urbanist Alain Bertaud calculated that the average Mumbaikar consumed about 4.5 sq m of residential floor space, a level you would expect from a city in wartime rationing, not the financial capital of a $4trn economy.⁴ Meanwhile the slums, which escape FSI entirely by being informal, cover nearly all their ground and leave no room for a road, let alone a park.

This is exactly the kind of thread I keep pulling at, and it is the sort of thing we go back and forth on in the BharatNama WhatsApp community, where I share a smaller India deep-dive like this with thousands of readers every morning (t.ly/h2jq1).

Join the BharatNama community

The metros we build, then starve

The second cost of capping floor is sprawl, and sprawl quietly sabotages the most expensive thing Indian cities are building right now: mass transit.

When a city cannot grow up, it grows out. The edge pushes further from the centre, commutes lengthen, and everyone is forced onto two-wheelers and cars. To fight the resulting gridlock, India has gone on a metro-building spree that is genuinely world class.

India now runs over 1,000 km of operational metro, across 29 cities, the third-largest network on the planet after China and the US.⁸
Delhi alone operates nearly 400 km; Bengaluru and Mumbai are expanding fast.⁸

But a metro only pays for itself if enough people live and work within walking distance of its stations. That is the entire logic of Transit-Oriented Development (TOD): concentrate dense, mixed-use towers around the station so the train is the natural choice and the fares actually cover the cost.

And here is where the old FSI mindset defeats the new infrastructure. Bengaluru is the cautionary tale.

The state allowed extra building rights only in a thin 150 m band around metro stations, shrinking the prize before anyone could claim it.
Even inside that band, developers often could not use the higher floor allowance, because the old setback and ground-coverage rules still applied to the small, fragmented plots typical of the inner city.
To build dense you had to merge plots, but the state offered no incentive to do so, so the amalgamation rarely happened.⁹

The result was a multi-billion-rupee metro running past land the rules would not let anyone develop. The transit got built; the density to feed it did not.

Where India stands, against the world and against itself

Set India beside the leaders and the gap is stark. But the more useful comparison, as always, is India against itself, because within the same country one city has quietly cracked the problem.

India versus the world, on the floor cap:

Mumbai base FSI: 1.33.³
Manhattan: up to 15.⁵
Central Tokyo: up to 13.⁶
Singapore: into the 20s.⁷

India versus the world, on what happens next. Tokyo is the control experiment. Its permissive, supply-first system means the city has held more homes than households for three decades, with vacancy near 11% and housing stock growing about twice as fast as its population.¹⁰ The pay-off: even after recent rises in its central wards, Tokyo rent remains a fraction of London's or New York's for a comparable flat.¹⁰ Shanghai makes the same point from the other direction: it lifted average floor space per person from 3.6 sq m in 1984 to 34 sq m by 2010, simply by raising FSI and letting redevelopment rip.¹¹

India versus India. This is the part that should give planners pause.

Mumbai clings to a base FSI of 1.33 and a 41% slum share.²
Ahmedabad, using its Town Planning Schemes, pools and re-plots land, funds its own trunk infrastructure, and sells premium FSI along its transit corridors. Revenue from that premium FSI grew from about 10% of the city's land-based finance in 2009 to roughly 60% by 2016.¹²

Same country, same constitution, opposite philosophy. One city treats density as a threat to be rationed. The other treats it as the asset that pays for the city.

What India can learn from Japan

The deepest lesson is not really about numbers. It is about who gets to say no.

In India, and much of the West, zoning is local and discretionary. A local body can delay, dilute, or block a dense project, and wealthy neighbourhoods use that veto to keep apartments out. Supply stays frozen because the people who benefit from scarcity hold the pen.

Japan took the pen away from them. Under the City Planning Law of 1968, zoning is national. The whole country runs on just 12 standardised land-use zones, applied uniformly, with local plans needing central sign-off.¹³ Three features make it work:

Nuisance-based, not exclusionary. Even the strictest residential zone allows a clinic, a small shop, a home office. Mixed use is the default, which is what makes walkable, transit-friendly neighbourhoods form on their own.¹³
By-right approval. If your design meets the objective national rules on height and light, the permit must be issued. No discretionary gauntlet, no room for the quiet extraction that plagues Indian approvals.¹³
Continuous deregulation. After its 1990s crash, Japan kept loosening the codes to pump out supply, and prices obeyed.

Not all of it transfers. India is a federation, land is a state subject, and state capacity varies wildly. But the portable idea is precise: standardise the rules, make density a right rather than a favour, and take the veto away from whoever profits from scarcity.

Delhi is inching this way. Its freshly notified TOD policy opens roughly 207 sq km along metro and rapid-rail corridors, allows a floor-area ratio up to 500 (that is, 5.0) on payment of charges, and reserves at least 65% of that floor for smaller homes.¹⁴ Whether it works will depend entirely on whether the old setback and coverage rules are relaxed to let that bulk actually get built. Bengaluru already showed what happens when they are not.

That distinction, between a rule that looks generous on paper and one that a developer can actually use, is exactly the sort of thing worth chewing over, and it is the kind of thread we keep pulling in the BharatNama community (t.ly/h2jq1).

Join the BharatNama community

What to watch

Whether Delhi's TOD notification is followed by matching relief on setbacks, ground coverage, and parking minimums, or stalls the way Bengaluru's did.
Mumbai's DCPR 2034 in practice: how much of the higher transit-corridor FSI (up to 5, and up to 7 near some metro stations) is genuinely usable versus trapped behind narrow roads.³
Whether any large state copies Ahmedabad's Town Planning Scheme and premium-FSI model to self-fund infrastructure, rather than waiting on central grants.
The next PLFS and urban housing data for signs that formal supply is finally outpacing household formation in the big metros. It has not, yet.

The one line

India's cities did not run out of land. They ran out of permission to build on it.

And if you want to understand where China stands and what it means for India, do check out my companion newsletter, Decoding the Dragon: https://t.ly/t7uhs

And, do check out my work on the following platforms as well: Instagram, LinkedIn and Youtube

Best,
Jayant

References

1. Economy of Mumbai, Wikipedia: https://en.wikipedia.org/wiki/Economy_of_Mumbai
2. Mumbai population and slum share (2025 estimate), World Population Review / Census of India data: https://worldpopulationreview.com/cities/india/mumbai
3. FSI in Mumbai under DCPR 2034, base 1.33 and transit-corridor limits: godrejproperties.com
4. Alain Bertaud, The Mumbai FAR/FSI conundrum (floor space per person, 66% water within 25 km): alainbertaud.com
5. Maximum FAR in Manhattan, NYC Zoning Resolution / UpCodes: up.codes
6. Floor-area ratio in central Tokyo (Marunouchi 1,300%), HouseRep Tokyo: houserep-tokyo.com
7. Floor area ratio, international comparison, Wikipedia: en.wikipedia.org/wiki/Floor_area_ratio
8. India's operational metro network, third largest globally, Metro Rail Today / Urban rail transit in India: en.wikipedia.org/wiki/Urban_rail_transit_in_India
9. WRI India, Synergizing Land Value Capture and Transit-Oriented Development: A Study of Bengaluru Metro: wri-india.org
10. Tokyo housing supply, vacancy and affordability vs London and New York, Foreign Affairs Review / UCLA Lewis Center: foreignaffairsreview.com
11. Shanghai floor space per capita, urban economic analysis of FSI reform, Alain Bertaud: alainbertaud.com
12. Ahmedabad Town Planning Schemes and premium FSI revenue, World Bank case study: World Bank
13. Japan's City Planning Law 1968 and 12 national zoning categories: theurbanist.org
14. Delhi TOD policy notified (207 sq km, FAR up to 500, 65% residential), The Tribune: tribuneindia.com

#32: Why does 25% forest cover hide India's shrinking forests?

Jayant Mundhra — Sat, 01 Aug 2026 01:30:17 GMT

Here is a number the government is proud of, and it deserves a second look. India's forest and tree cover now stands at ~8.3 lakh sq km, or 25.17% of the country's land, the highest it has ever been recorded.¹

That is the headline of the India State of Forest Report 2023, the biennial audit that is still the country's latest word on its own forests.¹ It is a good news story. Forest and tree cover rose by 1,445 sq km in two years. India's forests hold a carbon stock of 30.43bn tonnes of CO2 equivalent, and have already added 2.29bn tonnes of the 2.5 to 3.0bn-tonne sink India promised the world by 2030.²

Now here is the number nobody puts on a press release. Over the decade to 2021, roughly 93,000 sq km of India's forests, an area larger than Andhra Pradesh, structurally degraded: dense forest thinning to open, forest and scrub collapsing into bare land.³⁴

Both numbers are true at once. That is the paradox this edition is about.

The definition doing all the work

Start with what "forest cover" actually means in India, because the definition is doing almost all the work.

The Forest Survey of India counts as forest any patch of at least one hectare with a tree canopy of 10% or more.¹ That is it. Not who owns it, not whether it is a legal forest, and crucially, not what is growing there. As the FSI's own methodology says, the count includes commercial orchards, bamboo, and palm and rubber plantations.⁴

So the picture is stitched together from satellite imagery, mostly the indigenous LISS-III sensor at 23.5-metre resolution.⁴ A satellite is superb at seeing green. It cannot tell an ancient, multi-layered rainforest from a monoculture rubber estate. Both read as "forest".

Layer on the classification the FSI uses within that green:

Very Dense Forest (VDF): canopy above 70%.
Moderately Dense Forest (MDF): canopy 40 to 70%.
Open Forest (OF): canopy 10 to 40%.
Below 10%, land is scrub or non-forest.

The trouble is that a number can climb even as the thing it measures falls apart. A rubber plantation crossing 10% canopy adds to the tally exactly as a thinning old-growth forest slips down through the classes. The aggregate goes up. The ecology goes down. This is a story I keep coming back to in the BharatNama WhatsApp group, where I share a smaller India deep-dive like this with thousands of readers every morning, and forest data is a near-perfect case of a metric that flatters the thing it is meant to police (t.ly/h2jq1).

→ Join the BharatNama community

The mechanics of a comforting number

Look under the aggregate and the direction of travel reverses.

Between 2011 and 2021, inside recorded forest areas, the degradation ran in a clear downward staircase:³⁴

40,709 sq km of Very Dense and Moderately Dense Forest, India's richest carbon and biodiversity stores, thinned into Open Forest.
5,573 sq km of forest slipped further, into scrub.
46,707 sq km of forest and scrub lost their canopy entirely and became non-forest.

That is close to 93,000 sq km of structural loss. The FSI itself concedes that restoring this degraded land to its former density would lock up an additional ~637mn tonnes of carbon.³ That is the sequestration the current trajectory is bleeding away.

So how does a report built on that reality announce a net gain? Through a step that has spatial ecologists genuinely puzzled. A straight comparison of the 2021 and 2023 raw data suggests nearly 1,93,000 sq km of dense and open forest was lost or altered.⁴ The published report applies an unexplained "normalisation factor" that turns that gross loss into a net increase of just 297 sq km for those categories.⁴ The method is opaque to independent researchers, which is a polite way of saying nobody outside the FSI can reproduce it.

Here is the tell that makes the whole thing click. Organised plantation activity in India is estimated at nearly 18,000 sq km every year.⁴ Yet total forest and tree cover rose by only 1,445 sq km over two years.¹ Either vast tracts of natural forest are being cut at a pace that almost cancels all that planting, or most of those plantations are simply dying. Neither reading supports the triumphant headline.

Not aesthetics, chemistry

The instinct is to shrug: a tree is a tree, green is green. The soil says otherwise.

Most of a forest's durable carbon is not in the trunks and leaves you can see. It sits underground, as soil organic carbon, held there by the tangle of diverse roots, continuous leaf litter, and stable microbes that only a natural forest builds up over centuries.

A study by the Kerala Forest Research Institute and collaborators in the southern Western Ghats put paired plots side by side: pristine natural forest and adjacent teak plantations of 40 to 50 years.⁵ Over the top 50 cm of soil, the natural forest held roughly 225 tonnes of organic carbon per hectare. The teak plantation held about 195.⁵

That gap is the buried cost of every "offset" plantation. When a complex forest is cleared for a monoculture, the plantation may hit the 10% canopy bar within a few years and register as recovered cover, but the soil carbon, the acidity, the buffering capacity, all degrade over successive harvest rotations. On the satellite it looks like a win. In the ground it is a slow, uncounted haemorrhage.

Where India stands: against the world, and against itself

Strip away canopy area and measure forests the way foresters actually do, by "growing stock", the volume of living wood per hectare, and India's position is far more sobering.

Growing stock captures height, girth, age and density. It is the honest proxy for how much forest is really there. On this measure:

India: roughly 56 m³ of wood per hectare.⁶
World average: around 110 m³ per hectare.⁶
OECD average: about 131 m³ per hectare.⁷
South Korea: 165 m³ per hectare, as of 2020.⁷

India's forests, in other words, hold barely half the wood per hectare of an average forest anywhere, and roughly a third of South Korea's. India's total growing stock is nudging up in aggregate, to ~6.4bn cubic metres, but that is being propped up by trees outside forests, orchards and farm trees, even as the dense natural forests that carry the real volume thin out.¹

The picture from abroad is not all grim, and honesty cuts both ways. The FAO's Global Forest Resources Assessment 2025 ranks India 9th in the world by total forest area, at 72.7mn hectares, up a place, and 3rd globally in annual forest gain.⁸ India's forests remove ~150mn tonnes of CO2 a year, the 5th-largest forest carbon sink on earth.⁸ India is planting furiously. The question is what it is planting, and what it is losing to plant it.

Now turn the lens inward, because the national average of ~56 m³ per hectare hides a striking spread:⁶

Kerala: ~139 m³ per hectare, more than twice the national figure.
Uttarakhand: ~106 m³ per hectare.
Goa: ~101 m³ per hectare.

And the loss is concentrated in exactly the places India can least afford to lose. Between 2021 and 2023, almost every northeastern state, the country's richest biodiversity belt, lost forest inside its recorded areas: Tripura shed 116 sq km, Assam 86, Manipur 47, Arunachal Pradesh 45, with Mizoram the lone exception.⁹ In the Western Ghats, forest cover outside protected areas fell by 58 sq km against 2013 levels.³ Mangroves, India's coastal shield, slipped to 4,991 sq km, a net loss of 7.43 sq km, with Gujarat alone down 36 sq km.³

Even fire tells a local story the national average erases. Overall fire hotspots fell, yet Uttarakhand's incidents exploded from 5,351 in 2022-23 to 21,033 in 2023-24, and Himachal Pradesh's rose more than 1,300%.¹⁰¹¹ A single dry season in the hills can undo years of slow canopy gain.

What India can learn: the South Korea playbook

If one country has earned the right to lecture on forests, it is South Korea, and its lesson is precisely about the metric.

By 1953, war and colonial extraction had stripped Korea's forests to roughly 10 m³ of growing stock per hectare, near-total devastation.⁷ Seoul launched a national rehabilitation drive in 1973, wove it into the Saemaul Undong village movement, and planted billions of trees. So far, so familiar; India plants at scale too.

The difference was what Korea chose to measure. Not hectares of canopy, but growing stock, the three-dimensional volume of the forest. And it climbed relentlessly: to 50 m³ per hectare by 1990, and to 165 by 2020.⁷ Korea did not get there by watching saplings grow. It actively thinned overcrowded stands, paid landowners to convert commercial plantations back to natural forest, and stretched timber rotation ages by decades.⁷

This is the thread worth pulling, and it is exactly the kind of "what actually works abroad" comparison we chew over in the BharatNama community every day (t.ly/h2jq1). The portable lesson is not "plant more". India already does. It is to change the yardstick: reward volume, soil carbon and biodiversity, not the bare fact of green. What does not transfer easily is Korea's state capacity and its smaller, more uniform geography. India's federal, forest-dwelling, wildly diverse landscape makes the accounting harder, which is all the more reason to get the accounting honest.

→ Join the BharatNama community

The money that never moves

None of this is accident. It is incentive.

When forest is diverted for a mine or a highway, the user agency pays into CAMPA, the compensatory afforestation fund, which states are meant to spend replanting an equivalent area. In practice the money sits still. Between 2019-20 and 2023-24, National CAMPA approved ~₹3.85 lakh crore for state plans, of which only ~₹2.6 lakh crore, about 68%, was actually spent.¹²

Andhra Pradesh is the parable. It holds a CAMPA corpus of ~₹2,257 crore, yet spent just ₹658 crore between 2019-20 and November 2025, leaving ₹1,947 crore idle in government accounts.¹³ Planting is seasonal and monsoon-bound; when funds arrive months late, the window is missed and saplings die. National assessments put the survival rate of compensatory saplings at ~33.5%, roughly half the 60 to 65% the Forest Research Institute considers viable.¹³

Then there is the newer instrument, the Green Credit Programme, launched in October 2023, which lets companies buy credits by funding plantations on "degraded" land.¹⁴ Two problems have surfaced fast:

There is no firm definition of "degraded", so naturally open ecosystems, savannas, grasslands, scrub that support unique species and pastoralist livelihoods, get swept in. Roughly 13,240 hectares selected under the scheme overlap with likely open natural ecosystems.¹⁴
The human cost is real. In Assam's Goalpara, reporting found five sites covering 214 hectares cleared of homes, farmland and even schools in the name of restoration.¹⁴ Plantation survival on these sites runs between 6 and 30%.¹⁴

And in August 2025 the government widened the programme so those credits can now count towards compensatory afforestation itself, letting industry, as one former forest officer put it, buy an easy route out of land it would otherwise have to replace.¹⁵ The scheme now faces a Supreme Court challenge.¹⁴

What to watch

The Supreme Court and the definition of "forest". In interim orders on 3 February and 4 March 2025, the Court told every state to keep applying the wide 1996 Godavarman definition of forest, and barred any reduction of forest land without compensatory land first, effectively freezing the dilution in the Forest (Conservation) Amendment Act 2023.¹⁶¹⁷ Watch for the final ruling.
ISFR 2025. The next State of Forest Report is due. Watch whether the FSI finally separates natural forest from plantations, or repeats the normalisation.
CAMPA drawdown. Track whether idle state balances, Andhra's ₹1,947 crore above all, actually get spent on the ground before the next planting season.
Green Credit rules. Watch the litigation and whether the "degraded land" definition is tightened to protect open natural ecosystems.

India's forest map is greening on a satellite screen while its natural forests thin out underneath, and until the country measures wood, soil and biodiversity instead of bare canopy, it will keep mistaking a plantation for a forest.

And if you want to understand where China stands and what it means for India, do check out my companion newsletter, Decoding the Dragon: https://t.ly/t7uhs

And, do check out my work on the following platforms as well: Instagram, LinkedIn and Youtube

Best,
Jayant

References

Press Information Bureau, India State of Forest Report 2023 release: pib.gov.in
DD News, India's forest cover increases by 1,445 sq km, ISFR 2023: ddnews.gov.in
The Wire, ISFR 2023's grey areas: forest degradation, decline and FRA: thewire.in
Mongabay India, Forest survey reveals India's shift towards plantations: india.mongabay.com
Springer, Soil organic carbon pool under tree plantations in the Southern Western Ghats of Kerala: link.springer.com
Down To Earth, Forest survey report: India's growing stock and per-hectare volume: downtoearth.org.in
Korea Forest Service, Korean Forests at a Glance (growing stock 165 m³/ha, 2020): english.forest.go.kr
Press Information Bureau, GFRA 2025 ranks India 9th globally in total forest area: pib.gov.in
The Sentinel, Arunachal loses 549 sq km of forest cover, ISFR 2021-2023: sentinelassam.com
Business Standard, Uttarakhand, Himachal record sharp rise in forest fires: ISFR: business-standard.com
Down To Earth, Forest fires increased 1,339% in Himachal Pradesh: downtoearth.org.in
Khan Global Studies / CEC, Compensatory afforestation fund utilisation in India: khanglobalstudies.com
Counterview, Andhra Pradesh's unspent CAMPA funds raise ecological concerns: counterview.net
IndiaSpend, What's ailing India's Green Credit Programme?: indiaspend.com
Mongabay India, Green credit rules tweaked to favour canopy cover, remove trade provision: india.mongabay.com
Down To Earth, FCAA 2023: states must act as per Godavarman, says SC: downtoearth.org.in
LiveLaw, States must follow Godavarman definition of forest till 2023 rules applied: Supreme Court: livelaw.in

#31: Why are over 10 crore Indians denied their ration!?

Jayant Mundhra — Thu, 30 Jul 2026 01:30:27 GMT

Right now, until 13 July, the government has the National Food Security Act open on its desk for amendment.¹ Public comments are being invited on a draft bill. And here is what it proposes to change: the fixed 35kg monthly grain quota for the very poorest households would become 7kg per person, capped at 35kg per family.¹

That is the reform on the table. A tweak to how grams are split inside a household.

What the draft does not touch is the hole that actually matters. India's food law is legally supposed to cover two-thirds of the country. It currently rations that coverage using population figures from the 2011 Census.² More than a decade later, that single design choice has quietly locked out somewhere between 10 and 14 crore of the poorest Indians.³ ⁴

This is not corruption. It is not a leaky ration shop or a stolen sack of wheat. It is an exclusion written into the arithmetic of the law itself. And that, to me, is the more unnerving kind of failure, because you cannot fix it by catching a thief.

What the law actually promises, and how it counts

The NFSA, passed in 2013, was a genuine landmark. It turned subsidised food from a discretionary scheme into a legal right, and it made the eldest woman of the household the ration-card holder by law.² It entitles up to 75% of rural India and 50% of urban India to cheap grain, which works out to roughly two-thirds of the population.²

The catch sits in Section 9. The Act says the number of people to be covered "shall be calculated on the basis of the population estimates as per the census of which the relevant figures have been published."²

In plain terms: the size of the safety net is pinned to the last published census.

In 2013, the last published census was 2011, when India had ~1.21bn people.²
Apply 75% rural and 50% urban to that, and you get a national ceiling of 81.35 crore beneficiaries.²
That ceiling has not moved since.

Every state gets a slice of that 81.35 crore and cannot legally exceed it. As of July 2025, states had filled 80.56 crore of those slots, leaving room for just 0.79 crore more names nationwide before the cap is breached.² A country of 1.46bn people has legal headroom to add fewer than 80 lakh ration cards.

The frozen headcount versus the growing country

Here is the interesting bit. The census froze. The population did not.

India crossed China to become the world's most populous nation, and the UN now puts the count at ~1.46bn for 2025.⁵ The denominator kept climbing. The numerator stayed bolted to 2011.

Economists Jean Drèze, Reetika Khera and Meghana Mungikar did the arithmetic that made this concrete. On 2020 population projections of ~1.37bn, the two-thirds mandate should have covered ~922mn people. Actual coverage was ~809mn. The gap was already ~113mn.³
By late 2024, reporting using the government's own projections put the shortfall at ~13 crore.⁴
Run the same sum on today's 1.46bn and the number only grows. Two-thirds of 1.46bn is close to 98 crore. India covers ~80.6 crore. On paper, the law owes food to millions it has never counted.

Now let that sink in for a second. These are not people who failed a means test. They are people the state has simply refused to see, because seeing them would mean admitting the 2011 baseline is dead.

And here is where it gets strange. The system is perfectly capable of subtracting names. It just cannot legally add them.

In the past few months alone, states weeded out ~2.25 crore beneficiaries flagged as ineligible, and 2.49 crore cards have been cancelled since 2020 for death, duplication or migration.¹⁴
That machinery works because removing a name never breaches the cap. Adding one does.
So the pool churns furiously at the edges while its total stays frozen. A poor family can only get in if another family drops out, regardless of how many genuinely need it.

The result is a welfare net that is energetically policed for who should leave and legally blind to who should enter. That asymmetry is the whole tragedy in miniature.

This kind of buried, structural number is exactly what I keep pulling apart, and it is the sort of thread we chase most mornings in the BharatNama WhatsApp community, where I share a smaller India deep-dive like this with thousands of readers every day (t.ly/h2jq1).

Join the BharatNama community

The court could see them, even when the government would not

The proof that these people exist did not come from the census. It came from a different government database.

After the 2020 lockdown exodus, the Supreme Court took up the plight of migrant workers and pushed the Centre to build the e-Shram portal, a register of unorganised-sector workers. It has now crossed 30.9 crore registrations.⁶ When the court cross-checked it against ration records, the gap was stark: of the workers registered, around 8 crore had no ration card at all.⁷

So in March 2024, the court did something remarkable. It ordered every state to issue ration cards to those 8 crore workers, and to do it irrespective of the Section 9 quotas, later setting a final deadline of 19 November 2024.⁷ It effectively told the executive to bypass its own frozen formula because, in its words, the hungry cannot wait.

The executive did not agree. And this is where the story stops being administrative and becomes a genuine clash of worldviews:

The judiciary reads the right to food as flowing from Article 21, the right to life, and treats the 2011 freeze as a violation to be overridden.⁷
The executive, backed by NITI Aayog, reads Section 9 literally: coverage cannot change until a new census is published, full stop.⁸

And NITI Aayog wants to go the other way entirely. Rather than expand, it has floated cutting coverage from 75/50 down to 60/40, which would drop the covered pool to ~71.62 crore and save the exchequer roughly ₹47k crore a year.⁸ One arm of the state wants to add 8 crore names. Another wants to remove nearly 10 crore. The poor sit in the middle of that argument.

Where India stands: the world updates, India waits

This is the part that should sting, because the fix is not a mystery. Other countries stopped tying welfare to a decadal snapshot a long time ago.

The cleanest benchmark is Brazil's Cadastro Único, the single social registry behind Bolsa Família and dozens of other schemes.⁹

It is continuous, not decadal. A family that falls into poverty can walk into a municipal centre and register any day of the year, and benefits follow.⁹
It is huge and live: over 40mn families sit in the registry, with ~19.9mn families drawing Bolsa Família today.⁹
Because everyone vulnerable is already pre-registered, Brazil could switch on emergency cash to millions within weeks during COVID, simply by activating dormant records.⁹

The gap is really a gap in clocks. Brazil's headcount updates the day a family's circumstances change. India's updates once a census, which has now meant once every 16 years and counting.⁵ One system is designed to notice poverty as it happens. The other is designed to wait.

India, by contrast, waits for a census. When COVID hit and the Centre gave free grain over and above the NFSA quota, every one of those extra kilos went only to people who already held a card. The 10-crore-plus excluded got nothing, because to the system they did not exist.

The lesson from Brazil is not "copy Bolsa Família." It is narrower and sharper: delink the count from the census, and let a live registry do the counting.

Where India stands: against itself

Now break the national number open, because the intra-India picture is where the real inequity lives.

With the Centre frozen, richer states have quietly built their own parallel safety nets out of their own budgets:²

NEEM cards (entitlements equal to or better than NFSA) cover ~90mn people, funded by states, not Delhi.¹⁰
LEAN cards (lesser entitlements) cover ~51mn more.¹⁰
In the southern states, NEEM cardholders make up 31% to 45% of everyone the PDS reaches. Tamil Nadu runs a universal system and gives rice free to all.¹⁰ ¹¹

Combine these and states are feeding ~14 crore people the Centre refuses to count. That is a quiet act of federal generosity, but it is also a scandal.

Join the BharatNama community

What India can learn, and already knows

Here is the honest bit: India does not need Brazil to prove this works. It has already built the pieces at home.

Haryana's Parivar Pehchan Patra gives each family an 8-digit ID and pushes pensions and certificates to people automatically, no re-application needed.¹²
Karnataka's Kutumba and Rajasthan's Jan Aadhaar do the same, and Uttar Pradesh has copied the model.¹² When Karnataka linked scholarships to its family registry, uptake nearly doubled overnight because the system found eligible students instead of waiting for them to apply.¹²
And the e-Shram database already holds 30.9 crore unorganised workers, live and Aadhaar-linked.⁶

So the technology, the data and the state capacity all exist. What is missing is the legal will to let a dynamic registry override a dead census. That requires Parliament to amend Section 9, which is precisely the sort of change the draft bill on the table this week could have made, and did not.¹

There is a real prize here too, not just a fix. A live family registry lets the state track intergenerational poverty, target nutrition, and respond to shocks in weeks instead of a decade. That is the opportunity India keeps leaving on the table.

What to watch

The amendment bill. Comments close on 13 July.¹ Watch whether the final draft touches the coverage cap at all, or stays limited to the 7kg-per-person tweak.
Census 2027. House-listing began in April 2026, enumeration is set for February 2027, and the Centre has put ₹6k crore behind it.¹³ ⁵ But results will not land until ~2028, so the freeze holds for at least two more years.
The 8-crore compliance. Track whether states actually issue the e-Shram ration cards the court ordered, and whether the Centre supplies the extra grain to back them.⁷
State fiscal strain. Watch how long poorer states can keep funding NEEM and LEAN cards without central tide-over grain.¹⁰

The one line

India has built a food-security net wide enough to be the envy of the world, then chosen to measure it with a ruler that stopped growing in 2011.

And if you want to understand where China stands and what it means for India, do check out my companion newsletter, Decoding the Dragon: https://t.ly/t7uhs

And, do check out my work on the following platforms as well: Instagram, LinkedIn and Youtube

Best,
Jayant

References

1. Down To Earth, draft National Food Security (Amendment) Bill: link
2. NextIAS / The Wire, NFSA coverage 81.35 crore vs 80.56 crore identified and Section 9: link
3. IndiaSpend, Drèze-Khera-Mungikar estimate of 100mn+ excluded and 922mn mandate: link
4. Newslaundry, 13 crore Indians denied rations: link
5. TASS / UNFPA, India population surpasses 1.46bn in 2025: link
6. IBEF / PIB, over 30.9 crore unorganised workers registered on e-Shram: link
7. LiveLaw, Supreme Court directs states to issue ration cards to 8 crore e-Shram workers: link
8. Scroll, NITI Aayog recommends cutting NFSA coverage to 60/40: link
9. Global Alliance against Hunger and Poverty, Brazil Cadastro Único and Bolsa Família: link
10. Ideas for India, NEEM and LEAN cards and state-funded expanded PDS: link
11. Wikipedia, Tamil Nadu universal PDS and free rice: link
12. ORF / Bajaj Finserv, Haryana PPP, Karnataka Kutumba, Rajasthan Jan Aadhaar family IDs: link
13. Outlook / PIB, Census 2027 schedule and ₹6,000 crore budget: link
14. The Pioneer / Outlook, 2.25 crore ineligible removed and 2.49 crore cards cancelled since 2020: link

#30: Why does toxic air cost India 9.5% of GDP?

Jayant Mundhra — Tue, 28 Jul 2026 01:31:14 GMT

Here is a number that stopped me cold. India's own accounting says air pollution costs the economy 1.36% of GDP. An equally credible method says the real figure is 9.5%.¹²

Same country. Same toxic air. Same year, near enough. The two numbers are seven times apart.

That is not a rounding error. It is a fault line running straight through how the Indian state decides what to spend on clean air, and which side of that line you sit on quietly decides whether pollution control looks like a cost or the highest-return investment India could make.

The health story is well told by now. In 2022, fine particulate matter (PM2.5) was linked to about 1.72mn premature deaths in India, roughly a third of all such deaths on the planet.¹ What almost nobody talks about is the accounting. And the accounting is where the money, and the policy paralysis, actually lives.

The two numbers, and why they diverge

Both numbers are real. They just answer different questions.

The World Bank used what economists call the human capital, or lost-output, approach. It asks a narrow question: if a person dies early or falls sick, how much economic production is lost? Add up forgone wages and medical bills, discount to today, and you get a figure.

The result: air pollution cost India about $36.8bn in 2019, or 1.36% of GDP.²
Of that, ~$28.8bn was lost output from premature death, and ~$8bn was the direct cost of illness.²

The Lancet Countdown, the health-and-climate assessment run with the WHO, used a completely different lens: the Value of a Statistical Life (VSL). It does not price a named person. It measures how much a whole society is collectively willing to pay to shave down the risk of dying early, and multiplies that across the deaths.

The result: the monetised cost of premature mortality from outdoor air pollution in India in 2022 was $339.4bn, or 9.5% of GDP.¹
That is nearly ₹30 lakh crore, a sum that dwarfs India's entire annual health budget.¹

Now, the honest caveat first, because BharatNama does not paper over these. The World Bank figure is 2019 data and the Lancet figure is 2022 data. But the three-year gap explains almost none of the divergence. India's GDP did not grow seven-fold in between. The gap is about method, not timing.

The lost-output approach has a brutal blind spot: it values you only for your wages.

A retiree who dies from pollution-triggered heart failure has zero forgone earnings, so the model records almost nothing.
A child, an unpaid homemaker, an elderly parent: all near-invisible on the ledger, precisely the people most likely to die from bad air.
It captures the pay cheque you lose, and nothing of the life.

The VSL method captures the welfare loss the lost-output method throws away. Neither is "wrong". But when a finance ministry runs a cost-benefit test on a clean-air investment using the lost-output number, it is comparing the full cost of scrubbers and grid upgrades against a benefit figure that has been deflated by a factor of roughly seven. Cleaning the air looks like a luxury India cannot afford. That is the fiction the accounting builds.

The bill that never reaches the health ministry

Mortality is only the headline. The quieter drain is what pollution does to people who are very much alive and at work.

The landmark study here, by Dalberg with the Clean Air Fund and the Confederation of Indian Industry (CII), put the cost to Indian business at about $95bn a year, close to 3% of GDP.³ Break it open and it stops being abstract:

Presenteeism, ~$24bn. Employees show up but underperform, because ultrafine particles trigger inflammation that dulls stamina and cognition. This is the single biggest line, and it is roughly the size of India's entire pharma output.³
Consumer pull-back, ~$22bn. On filthy-air days people stay home, and offline, footfall-dependent businesses (restaurants, apparel retail) eat the loss.³
Absenteeism, ~$6bn. India loses about 1.3bn working days a year to pollution-related sick days and caring for sick dependents.³

Sit with the presenteeism number for a second. It is the most invisible of all, because attendance looks fine on the register. The person is at the desk. They are just thinking a little slower, all day, every high-pollution day. In a services economy that sells cognition (software, analytics, finance) that is the worst possible tax to levy, and it lands hardest on exactly the sectors India is betting its future on.

This is the sort of buried, second-order number I keep pulling on, and it is the kind of thread we go back and forth on in the BharatNama WhatsApp community, where I share a smaller India deep-dive like this with thousands of readers every morning (t.ly/h2jq1).

Join the BharatNama community

Where India stands: against the world, and against itself

Air pollution is not evenly spread, and this is where BharatNama's two comparisons earn their keep.

India vs the world. India carries a share of the global air-pollution death toll wildly out of proportion to its share of people.

India accounted for roughly a third of all global PM2.5 deaths in 2022, on about 18% of the world's population.¹
The Indo-Gangetic Plain, stretching across India's north, is one of the single most polluted large regions on Earth. Counting it with the Himalayan foothills, close to ~1bn people breathe hazardous air there, with exposure often more than twenty times the WHO safe limit.⁸
China, once the poster child for the "airpocalypse", cut its national PM2.5 from 72 to 29 µg/m³ between 2013 and 2022.⁶ India has not managed anything close to that pace.

India vs India. The national average hides a spread that is itself the story. On the World Bank's own state numbers, the economic drag is heavily concentrated in the north and east:

Uttar Pradesh: 2.15% of state GDP. Bihar: 1.95%. Madhya Pradesh and Rajasthan: 1.70% each. Chhattisgarh: 1.55%.²
These are multiples of what cleaner southern states lose. A single national figure of 1.36% flattens a landscape where the worst-hit states bleed nearly twice as much.

That north-heavy pattern is not a coincidence, and it points straight at why India's whole approach is built on the wrong map.

The governance gap: a city-sized answer to a regional problem

India's flagship response is the National Clean Air Programme (NCAP), launched in 2019. Its target has since been sharpened: up to a 40% cut in particulate concentration by 2025-26, against a 2019-20 baseline, across 131 "non-attainment" cities, up from the original 20-30% goal.⁴ Progress is tracked on a dashboard called PRANA.

The intent is right. The architecture is where it breaks, on two counts.

First, it is city-shaped, and pollution is not. NCAP hands each of 131 cities a local action plan for road dust, local traffic, and waste burning. But PM2.5 does not respect a municipal boundary.

The north's winter smog is a regional airshed problem: stubble burning in Punjab and Haryana, industrial clusters in the hinterland, weather inversions that trap it all.
A single city has no legal power to regulate an upwind farm or a power plant a hundred kilometres away. Isolating the fix to the city guarantees the fix falls short. The World Bank is now pushing India toward exactly this: its first regional airshed plan for the Indo-Gangetic Plain, spanning seven states.⁸

Second, the targets are advisory, and the enforcement void is glaring. Nowhere is this clearer than coal power, which is a major source of the sulphates and nitrates that become PM2.5.

Emission norms mandating Flue Gas Desulphurisation (FGD) to capture sulphur dioxide were first set in 2015, with a 2017 deadline. Deadlines were then extended, and extended again, for a decade.
Then in July 2025, the environment ministry did not just delay, it retreated. A new notification fully exempted the roughly 78% of coal units categorised "Category C" from installing FGD at all. Only ~11% ("Category A", near Delhi and big cities) must comply, by December 2027; another ~11% ("Category B") by 2028, subject to review.⁵
The stated reason: FGD costs about ₹1.2 crore per MW, and the ministry argued sulphur dioxide's role in PM2.5 was limited. Ten years after the first mandate, the primary source of baseload pollution got a pass.⁵

When regulatory deadlines are this soft, no heavy industry has any reason to internalise the cost of the air it fouls. The externality just stays external, and the 9.5% keeps compounding.

This state-versus-state, city-versus-airshed mismatch is the kind of structural knot I dig into daily, and it is one of the threads readers and I keep chasing in the BharatNama community (t.ly/h2jq1).

Join the BharatNama community

What India can learn from China

China faced a worse "airpocalypse" than India has and pulled out of it fast. The lesson is not "copy China", because much of it does not transfer. But two parts genuinely might.

What China actually did in its $270bn 2013 action plan:⁶

It set legally binding, region-specific targets, not city ones. Whole airsheds (Beijing-Tianjin-Hebei, the Yangtze and Pearl River Deltas) got hard numerical mandates. Beijing was told to cut PM2.5 34% from 2013 levels, full stop.⁶
It rewired the incentives of officials. Environmental performance was written into the "cadre evaluation" system that decides promotions. Suddenly a mayor's career depended on clean air, not just local GDP. That single move converted paper targets into aggressive local enforcement, and Beijing-Tianjin-Hebei saw particulate exposure fall by more than half.⁶

The result was one of the fastest air-quality turnarounds in modern history. What is portable to India: the shift from city plans to binding airshed authorities, and hard targets with real consequences. What is not portable: China's top-down control over promotions has no clean analogue in India's federal, multi-party democracy, where states guard their turf and coal towns vote.

And here is the caveat India must not miss. In 2023, China's national PM2.5 rose for the first time in a decade, up about 3.6%, with 80% of provincial capitals worsening.⁷ The moment Beijing pivoted back to post-Covid growth, industry roared and the gains reversed.

The lesson inside the lesson: end-of-pipe fixes and temporary bans do not hold. Only a structural move off fossil fuels does. If political pressure swings back to pure GDP maximisation, dirty air comes straight back.

What to watch

Whether the Indo-Gangetic Plain airshed plan actually gets statutory teeth, or stays advisory like NCAP.
Whether any Category A coal units genuinely hit the December 2027 FGD deadline, or win yet another extension.
Whether NITI Aayog or the finance ministry start using VSL-based valuation in cost-benefit tests. That single accounting change would reprice every clean-air investment overnight.
The next Lancet Countdown, for whether India's mortality and cost figures move at all from the 2022 baseline.

The one line

India is not failing to fix its air because it cannot afford to; it is failing because its own books value the damage at a seventh of what it costs, and you do not fund a fight you have decided is cheap.

And if you want to understand where China stands and what it means for India, do check out my companion newsletter, Decoding the Dragon: https://t.ly/t7uhs

And, do check out my work on the following platforms as well: Instagram, LinkedIn and Youtube

Best,
Jayant

References

Down To Earth / Business Standard, on the 2025 Lancet Countdown on Health and Climate Change (9th edition): link
World Bank cost estimate, via Forbes; state-level shares from GBD 2019 (Lancet Planetary Health): link
Dalberg Advisors with Clean Air Fund and CII, "The Business Case for Clean Air": link
Business Standard / PIB, NCAP revised 40% target: link
Health Policy Watch, on the July 2025 FGD exemption notification: link
AQLI (EPIC, University of Chicago) and Earth.org, China's 2013 National Air Quality Action Plan: link
Centre for Research on Energy and Clean Air (CREA), China 2023 PM2.5 rebound: link
World Bank, "A Breath of Change", Indo-Gangetic Plain airshed: link

#29: Why does India lose 80% of its elite AI talent?

Jayant Mundhra — Sat, 25 Jul 2026 01:31:17 GMT

I have been staring at one number all week, and I cannot make peace with it.

India now trains the second-largest pool of top AI talent in the world, roughly 50,000 elite researchers and inventors, behind only the United States.¹ And yet, at the very frontier of the field, the people publishing at NeurIPS and ICML, only about one in five of those trained in India actually work in India. Four in five have left.³

So we are the world's second-biggest factory for AI genius. And its single largest exporter of it.

That is the paradox I want to sit with today. The comfortable story, that India simply needs to produce more engineers, is wrong. We have solved the making. We have not solved the keeping.

Let me lay the numbers out, because they are stark.

Stanford's 2026 AI Index ranks India second in the world for elite AI authors and inventors, a pool of about 50,000, beaten only by the US.¹
The same report finds India had the worst net outflow of AI talent of any country on earth in 2025, a net-flow score of -16.9. Germany was -2.4 and Canada -7.1. India is in a league of its own.^{1, 2}
India's AI workforce grew 120% between 2019 and 2025, among the fastest anywhere, and India ranks first globally in AI skill penetration.¹
But at the frontier, the MacroPolo Global AI Talent Tracker, built on the latest NeurIPS and ICML papers, finds India keeps only about 20% of its top-tier researchers. The rest work abroad, overwhelmingly in America.³

Here is the thing. This is not a pipeline problem. India is minting the talent. It is a demand-and-retention problem. We have built the nursery and forgotten to build the home.

And the loss compounds. When a frontier researcher leaves, India does not just lose their output. It loses the students they would have mentored, the intellectual property they would have generated, and the deep-tech company they would have founded. Run that for a decade and the top of the pyramid stays permanently hollow.

So why do they go? Not for a nicer lifestyle. They go because the modern AI frontier is gated by two things India is still short of: compute and capital.

It starts with the machines

For a frontier researcher, compute is not an IT line item. It is the medium of the work itself. Your ability to test a new architecture, train a large model, and publish at a top venue is capped directly by how many GPUs you can command.

India has understood this and moved. The IndiaAI Mission, approved in March 2024 with an outlay of over ₹10,000 crore, is building shared national compute.⁴ Through its compute portal, India has now onboarded more than 38,000 GPUs, plus about 1,050 Google TPUs, offered to startups and academics at heavily subsidised rates.⁵

That is a serious national effort. Next to the frontier, it is also tiny.

The big US hyperscalers spent roughly $410bn on capital expenditure in 2025, and are guiding towards $600bn and more in 2026.⁶
A single US project, Stargate, is budgeted at around $500bn.⁶
India's entire sovereign compute pool is a fraction of what one American lab can rent on demand.

An elite researcher choosing between a subsidised queue for a slice of 38,000 shared GPUs and instant, dedicated access to tens of thousands of the newest chips will pick the second every time. Queuing for compute is fatal to the fast iteration that research runs on.

You can see the gap in the output. Take two labs.

Sarvam AI, one of India's flagship sovereign-AI startups, is now valued at about $1.5bn after a Series B targeting $300mn, backed by HCLTech, Bessemer, Khosla and Peak XV.⁷ It has a large team and premier access to government compute.
Arcee AI, a roughly 30-person lab in San Francisco, has raised under $50mn in total.⁸

Yet in early 2026 Arcee spent about $20mn, nearly half its lifetime funding, on a single 33-day training run across 2,048 of NVIDIA's newest Blackwell chips, and shipped Trinity, a fully open 400-billion-parameter model that briefly became the most-used open model in the US.⁸

The lesson is not that Sarvam is weak. It is that a lean Western lab with unconstrained compute can outrun a better-funded, better-staffed Indian one, because it never has to queue.

The private sector is trying to close this. Yotta Data Services has committed over $2bn to build one of Asia's largest AI clusters, deploying 20,736 of NVIDIA's Blackwell Ultra GPUs by August 2026, with more than 10,000 of them committed to the IndiaAI Mission and campuses scalable towards 2GW.⁹

That is exactly the right direction. But racking the hardware is necessary, not sufficient. Someone still has to pay the people who will use it.

This gap between the talent we train and the compute we can actually give it is exactly the kind of thread I keep pulling at, and it is what the BharatNama WhatsApp community and I go back and forth on most mornings, where I share a smaller India deep-dive like this one with thousands of readers (t.ly/h2jq1).

Join the BharatNama community

The salary chasm, and the missing middle

India's pay for AI researchers has genuinely exploded. In 2026:

Entry-level research roles at funded startups and GCCs, the in-house "global capability centres" that foreign firms run in India, pay roughly ₹15 to 35 lakh a year.¹²
Senior scientists at the Indian labs of Google DeepMind, Microsoft and Meta reach ₹80 lakh to ₹1.5 crore.¹²
Principal researchers can cross ₹2 crore.¹²

For the Indian market, that is a revolution. Against the frontier, it is still no contest. A researcher publishing seminal work at NeurIPS can command $300,000 to $500,000, roughly ₹2.5 to ₹4 crore, in the Bay Area, with far richer equity in a far more liquid market.

Indian startups like Sarvam are fighting back with equity rather than cash, and for some senior engineers it is working. But equity is only worth what the market underneath it is worth. And here the Indian venture market has a specific, dangerous shape.

Money is flowing. Indian startups raised $6.9bn in the first half of 2026, up 21% on the year,¹⁰ and $4.37bn in the second quarter alone.¹¹ But look at where it sits.

Heavy at the seed stage, where investors spray small early bets.
Heavy at the very top, in a few late-stage megarounds.
Thin in the middle, at Series A and B.¹¹

That missing middle is a killing field for AI startups. A young lab that raised a seed round cannot raise the Series A it needs to buy dedicated GPU clusters and match foreign pay. So its best engineers get absorbed by a US firm or its Indian GCC, and the startup quietly dies as an independent force. The talent may stay in India. The frontier ambition does not.

Why bringing them back has not worked

India has known about this drain for years and has tried to reverse it. VAJRA, GIAN, the Ramanujan and Ramalingaswami fellowships, all designed to pull the diaspora back.

They have barely moved the needle. And the reason, as India's own policy reviews concede, is not a lack of interest from the diaspora. It is institutional inertia at home.

Frontier AI runs at a velocity that mid-twentieth-century bureaucracy cannot match. A returning star hits heavy teaching loads, tenure-based promotion, and procurement rules where buying a dataset or a GPU cluster takes months. In a field where models go stale in weeks, a six-month procurement cycle is a career-ender.

The newest attempt is the Prime Minister Research Chair scheme, opened in 2026, which offers Indian-origin researchers grants of up to ₹14 crore to return and work in priority areas including AI and semiconductors, with applications closing this July.¹³ It is the most generous version yet.

But money was never the binding constraint. Autonomy was. Unless the host institutions are freed from procurement rules, hiring quotas and rigid HR, a red carpet into a slow building still leads into a slow building. A scheme is not a strategy.

Where India stands: against the world, and against itself

Set India beside itself and you see two countries.

The base is extraordinary. India ranks first in the world for AI skill penetration, its AI workforce is among the fastest-growing anywhere, and the vast majority of Indian firms now use AI in some form.¹
The apex is hollow. India holds just about 0.4% of the world's AI patents, against roughly 70% for China and 14% for the US.^{1, 14}

That is the real intra-India divide in AI. Not state against state, but a broad, world-class application layer sitting on top of an almost non-existent frontier-IP layer. We are superb at using the technology and negligible at owning it.

There is one hopeful counterweight. India now accounts for about 5% of all open-source AI projects on GitHub, a far better showing than its patent share.^{1, 16} The talent is clearly there. It is just building in the open, and often on someone else's stack.

China's answer to the very same brain drain is the kind of comparison I find myself making constantly, one India number set honestly against the world, and it is what we dig into every morning in the BharatNama community (t.ly/h2jq1).

Join the BharatNama community

What India can learn from China

Here is the part that should sting, and instruct.

China loses its apex too. On the same MacroPolo tracker, China retains only about 11% of its very top researchers, down from 16%, and 72% of its elite work in the US.³ On the absolute frontier, America drains China even harder than it drains India.

And yet China is a genuine AI superpower. How?

Because China does not try to keep everyone. It accepts it will lose some of the top 1%, and instead retains roughly 90% of the enormous tier just below, the engineers who implement, optimise and deploy. It does this by manufacturing domestic demand.

Its new 15th Five-Year Plan, running 2026 to 2030, makes AI the organising logic of the whole economy. The plan names AI 52 times, against six in the last one, and carries a standalone "AI Plus" action plan.¹⁵
It is backed by a state venture fund of around 1 trillion yuan, roughly $140bn, for AI and related technology.¹⁵
Beijing actively subsidises the use of home-grown compute, tilting the whole ecosystem towards domestic chips and models.

The lesson for India is precise. India will not out-agglomerate Silicon Valley for the top 1%, and it is fine to admit that battle is lost. But India is losing a battle it could win, which is building a domestic frontier-adjacent layer deep enough and rich enough to keep the other 99%. China keeps its workforce by making the country thirsty for AI. India lets its talent leak because there is not yet enough at home for it to do.

What to watch

Whether Yotta's Blackwell Ultra cluster actually goes live on schedule in August 2026, and how much of it reaches startups rather than sitting in a queue.⁹
Whether the Tata-PSMC fab at Dholera hits its first-silicon target of December 2026, India's first real step towards making its own chips rather than importing all of them.^{17, 18}
Whether the Prime Minister Research Chair scheme attracts genuine frontier names, and whether their host institutions are actually freed from procurement and hiring rules.¹³
Whether Series A and B funding for deep-tech deepens, or the missing middle stays missing.

India has proven it can forge the world's best AI minds. The question of the decade is whether it can build a home worth keeping them in.

And if you want to understand where China stands and what it means for India, do check out my companion newsletter, Decoding the Dragon: https://t.ly/t7uhs

And, do check out my work on the following platforms as well: Instagram, LinkedIn and Youtube

Best,
Jayant

References

Stanford HAI, 2026 AI Index Report: hai.stanford.edu/ai-index/2026-ai-index-report
The Print, India leads in AI talent but also brain drain: theprint.in
MacroPolo, Global AI Talent Tracker 3.0: archivemacropolo.org
PIB, Cabinet approves over Rs 10,300 crore for IndiaAI Mission: pib.gov.in
PIB, India's common compute crosses 34,000 GPUs: pib.gov.in
Futurum, AI Capex 2026: futurumgroup.com
Business Standard, Sarvam raises $234mn from HCLTech, Bessemer: business-standard.com
VentureBeat, Arcee's open-source Trinity-Large: venturebeat.com
Yotta, to deploy 20,736 NVIDIA Blackwell Ultra GPUs: yotta.com
YourStory, Indian startups raise $6.9B in H1 2026: yourstory.com
Analytics Insight / ANI, Indian Tech Funding Q2 2026: aninews.in
Recrew, AI Research Engineer Salary India 2026: recrew.ai
InsightsonIndia, Prime Minister Research Chair Scheme 2026: insightsonindia.com
Business Standard, India's AI patent record lags far behind global leaders: business-standard.com
Strider Intel, The PRC's 15th Five-Year Plan and the AI+ agenda: striderintel.com
Open Source For You, India's open source AI boom outpaces patents: opensourceforu.com
Ananta IAS, India Semiconductor Mission and approved fabs: anantamias.com
ABHS, India Semiconductor Mission 2.0, Tata first silicon late 2026: abhs.in

#28: Why does Vietnam out-export India 8 to 1 per person?

Jayant Mundhra — Thu, 23 Jul 2026 01:30:25 GMT

I want to start with a number that stopped me cold this week.

India exported a record ~$825bn of goods and services in FY25.¹ Vietnam, a country you could fold into Uttar Pradesh and still have room, exported ~$471bn in 2025.² On the headline, India wins by a mile.

Now divide by people. India is ~1.47bn strong.³ Vietnam is ~102mn.³ That works out to roughly $560 of exports per Indian, against ~$4,600 per Vietnamese.

India: ~$825bn ÷ ~1.47bn people = ~$560 per head.¹ ³
Vietnam: ~$471bn ÷ ~102mn people = ~$4,600 per head.² ³

So the average Vietnamese sells about eight times more to the world than the average Indian does. That is not a rounding error. That is the difference between a country that has plugged its people into global trade and one that has not.

The metric that actually predicts jobs

Here is the thing. Aggregate exports tell you how big your economy is. Exports per person tell you how deep trade reaches into ordinary lives. And it is the second one that predicts mass jobs and the exit from poverty.

The logic is old and unglamorous. Development economists call it structural transformation: moving people off low-productivity farms and onto higher-productivity factory floors and service desks. The 1954 Lewis model made the point simply. Poor countries sit on a vast pool of surplus farm labour, and growth means draining that pool into wage work.

But a poor country's own citizens are too poor to buy all the shirts and phones its factories could make. So exports become the release valve. You employ tens of millions by selling to the rest of the world, not to yourself.

By that test, Vietnam did the classic thing and India did something very strange.

Vietnam ran the textbook. India skipped a chapter.

Vietnam took the well-trodden path: cheap labour, low-skill assembly, sell it abroad. Japan did it, then South Korea, then China, then Bangladesh, and now Vietnam.

You can even date Vietnam's takeoff. When the US-Vietnam trade agreement came into force in the early 2000s, average American tariffs on Vietnamese manufactures fell from over 30% to under 3%.¹⁵ Multinationals piled in almost overnight, and formal factory employment climbed and stayed high. Open the door to the market, and the jobs follow. That is the whole mechanism, in one natural experiment.

India, uniquely, bypassed the low-skill factory stage almost entirely. It leapt straight into high-skill exports: software, back-office services, complex pharma. Economists like Dani Rodrik and Arvind Subramanian call this India's "endowment-defying" growth, a country rich in unskilled labour specialising in skill-heavy sectors.

And on its own terms, that leap has been spectacular:

India's IT and business-services industry earned ~$283bn in revenue in FY25, with exports of ~$233bn.⁷
It is a genuine world-beater and a huge earner of foreign exchange.

But look at who it employs. The whole IT-BPM sector directly employs ~5.8mn people.⁷ In a country of ~1.47bn, with millions joining the workforce every year, 5.8mn is a rounding error. It draws almost entirely from the English-speaking, college-educated top slice, the Indians furthest from the farm.

And here is the cost of skipping the low-skill chapter. Economists who have studied India's trade pattern estimate the country is "missing" well over $100bn a year of the unskilled, labour-intensive manufacturing it should, on paper, dominate, most of it in textiles, clothing and simple goods.¹⁵ That missing activity is precisely the kind that hires people with a school certificate, not an engineering degree. It is the jobs India most needs, in the sectors India most obviously suited itself for, and it never showed up.

This is exactly the kind of thread I keep pulling in the BharatNama WhatsApp community, where I share a smaller India deep-dive like this with thousands of readers every morning (t.ly/h2jq1). The IT miracle is real. It is just not a jobs machine for the masses, and it was never going to be.

Join the BharatNama community

What "export depth" looks like on the ground

Vietnam's export engine is not really Vietnamese. Foreign firms drive roughly three-quarters of its exports, and in electronics, foreign-invested firms account for ~98% of the total.¹⁴ It is, in effect, a duty-free assembly floor rented out to the world's biggest manufacturers.

But that floor employs a nation. Take the anchor tenant:

Samsung has poured in a cumulative ~$24bn and now runs some of its largest plants on earth in Bac Ninh and Thai Nguyen.⁵
In January 2025 alone, Bac Ninh cleared a further ~$1.2bn top-up for Samsung Display, taking that unit past ~$7.7bn.⁵
Samsung's Vietnam revenue is worth around 13% of the country's entire GDP, and it employs close to 90,000 people directly, before you count the ~80,000 in its supplier cluster.⁵

Electronics is now the spine of the whole economy:

Computers, electronics and components alone brought in ~$108bn in 2025, up an eye-watering ~48% in a single year.⁴
Add phones and the electronics cluster crossed ~$165bn, about 35% of all Vietnamese exports.⁴
Crucially, a lot of these assembly-line jobs go to women and to workers who a decade ago were in the paddy fields.

That is structural transformation you can photograph.

Where India stands, against the world and against itself

Set the two countries side by side on the metric that matters, and the picture is stark. But the intra-India split is just as revealing, because India's own success is wildly concentrated.

India vs the world (exports per person):

Vietnam: ~$4,600.² ³
India: ~$560.¹ ³
The gap is ~8x, and it is almost entirely a manufacturing-depth gap, not a services gap.

India vs itself (who actually exports):

India's merchandise and services exports lean heavily on a handful of clusters: IT out of Bengaluru, Hyderabad and Pune; petroleum out of Jamnagar; gems out of Surat; and, increasingly, phones out of a few plants in Tamil Nadu and Uttar Pradesh.
A short list of states, led by Gujarat, Maharashtra and Tamil Nadu, accounts for the bulk of goods exports, while the poorest and most populous states in the Gangetic belt barely feature. India's export map is a few bright dots on a dark canvas.
Smartphones became India's single largest export commodity in 2025, and India's electronics output hit ~₹11.3 lakh crore in FY25.⁸
Yet that same electronics push has created only ~2.5mn jobs over a decade.⁸ Good, but a fraction of what a Vietnam-scale assembly base would throw off.

Here is the honest part. India is finally winning the very game Vietnam plays. In smartphone shipments to the US in Q2 2025, India took the top spot with a 44% share, ahead of Vietnam's 30% and China's collapsing 25%.⁹ The direction of travel is genuinely good.

The problem is depth. One flagship product in one or two states is not the same as an export base wide enough to soak up the ~7-8mn young Indians entering the labour force every year.

Why India's factories cost more before they even start

If India has the cheaper labour, why is it not eating Vietnam's lunch? Because cheap wages get cancelled out by two self-inflicted costs.

One, India taxes the parts it needs to build things. The India Cellular and Electronics Association found India runs the highest input tariffs on electronics among all its rivals.¹⁰

India's average import tariff on these components is ~9.7%, against ~3.2% in China and an effective ~1% in Vietnam, because over 80% of Vietnam's component imports come in duty-free under trade deals.¹⁰
The net result is a cost disability of ~7% versus Vietnam and ~7.5% versus China, before a single phone is boxed.¹⁰
In an industry that runs on wafer-thin margins, a 7% penalty decides where the factory gets built.

Two, India is simply further from the parts. This is geography, and it does not care about policy.

Vietnam shares a ~1,450km land border with China. Components leave Shenzhen on a truck and reach a factory in Bac Ninh in 2 to 3 days.
The same parts from Shenzhen to Chennai mean 2 to 3 weeks on a ship.
That is weeks of extra working capital tied up in inventory, on every order.

One myth worth killing here, because the note I started from repeated it and the latest official data contradicts it. India's logistics cost is often quoted at 13-14% of GDP. The first systematic government study, by DPIIT and NCAER, actually pegged it at ~8% of GDP.¹¹ Vietnam's own logistics costs, by its government's own admission, run far higher, closer to 16-18%. So India's disadvantage in electronics is not some blanket logistics tax. It is the specific, brutal fact of sitting three weeks by sea from the world's component belt.

This kind of buried, counter-intuitive number is exactly what we go back and forth on in the BharatNama WhatsApp group (t.ly/h2jq1), and it is why I never take a widely repeated statistic at face value.

Join the BharatNama community

What India can learn from Vietnam

Vietnam's lesson is not "have cheaper labour". India already does. The lesson is about friction.

Trade agreements. Vietnam has signed ~18 free-trade agreements, and in July 2026 it wrapped up one more with the European free-trade bloc.⁶ That network lets it import a Chinese chip and a Korean screen tariff-free, assemble them, and ship the finished phone into the EU or Canada duty-free.
India spent years outside this game. It walked out of the RCEP bloc in 2019, and its big Western deals sat unsigned.¹²
But this is where the note's pessimism is now out of date, and the update is genuinely encouraging. India signed a trade deal with the UK in July 2025 and concluded one with the European Union in January 2026.¹² The scaffolding is finally going up.

The other lesson is sequencing. Vietnam built the export factory first and let domestic consumption follow. India did the reverse.

Private consumption is around 60% of India's GDP, far higher than in China or Vietnam.¹³
So Indian manufacturing grew mostly to serve 1.4bn domestic buyers, behind a high tariff wall, rather than to win foreign markets.
India's flagship ₹1.97 lakh crore Production-Linked Incentive scheme has pulled in investment and jobs, but much of it rewards assembly for the Indian market.¹⁶

That is a walled garden. It creates jobs for the local shop, but it does not, by itself, force the ruthless global cost-competitiveness that export depth demands. The tariff cuts and the new FTAs are the way out, and India has finally started walking through the door.

What to watch

Whether India's new FTAs actually lower input costs. Watch if the UK and EU deals, plus any tariff rationalisation in the next Budget, start narrowing that ~7% cost disability the ICEA measured.
The next PLI iteration. Track whether it is tied to hard export targets rather than domestic-sales milestones.
Component ecosystems, not just assembly. Watch whether India's new electronics-component scheme actually pulls display and PCB makers onshore, the thing that would shorten those 3-week supply lines.
Vietnam's own squeeze. Its domestic value-add is still only ~5-10% per device, and US tariff pressure is rising. If Vietnam gets caught, India's window widens.

The one line: India already sells more to the world than Vietnam; it just has not yet found a way to let more than a sliver of its own people do the selling.

And if you want to understand where China stands and what it means for India, do check out my companion newsletter, Decoding the Dragon: https://t.ly/t7uhs

And, do check out my work on the following platforms as well: Instagram, LinkedIn and Youtube

Best,
Jayant

References

1. Business Standard / PIB, India total exports FY25: link
2. VnEconomy, Vietnam 2025 trade and export record: link
3. United Nations / Worldometer population data (India and Vietnam, 2025): link
4. Vietnam Investment Review, electronics exports 2025: link
5. The Investor / TNGlobal, Samsung Vietnam investment and employment: link
6. European Free Trade Association, EFTA-Vietnam FTA concluded July 2026: link
7. NASSCOM, tech industry Strategic Review FY25: link
8. PIB, electronics manufacturing and production FY25: link
9. VnEconomy (Canalys data), Vietnam second-largest smartphone exporter to US: link
10. ICEA, tariff disability study across competing economies: link
11. Business Standard, India logistics cost 7.97% of GDP (DPIIT-NCAER): link
12. India Briefing, India FTA updates (UK CETA, EU FTA): link
13. World Bank, final consumption expenditure (% of GDP), India: link
14. RMIT University / ASEAN Briefing, Vietnam electronics FDI (98% of exports): link
15. Rodrik, Subramanian, Chatterjee and related trade research on India's endowment-defying specialisation and the "missing" manufacturing: link
16. IBEF / PIB, Production-Linked Incentive scheme outlay: link