The Efficiency Gap: India, the World's Fastest Growing Poor Economy

In 1990, the average Indian earned more than the average Chinese. In 2025, the average Chinese earns five times more. That single fact is the question this guide exists to answer, because everything else about the Indian economy, the growth headlines, the fifth largest economy ranking, the 170 million lifted out of extreme poverty, is real and still does not explain it away.

The growth headlines describe the size and speed of the economy. They say nothing about its efficiency: how much lasting income, wealth and opportunity it manufactures out of a given quantity of people, capital and years. On efficiency, the honest reading is uncomfortable. For a country that started level with China, sits on the largest young workforce in history, and has grown respectably for three decades, why is the average Indian still so poor, and getting richer so slowly?

The answer is not one villain. It is a chain of linked inefficiencies: too little investment, spent on the wrong things, moving through infrastructure that leaks, over a workforce the state under equips. This guide takes each link in turn, compares it against economies that made the other choice, and does not flinch at what the data says, including where India has genuinely done well.

Chapter 1 · The Great Divergence: two countries, one starting line

Begin with the comparison every Indian economist reaches for, because it is the cleanest natural experiment in modern development. In 1990, India and China were the same kind of poor. India's income per person was $371; China's was $348. India was, if anything, marginally ahead (IMF). Two enormous, agrarian, formerly closed Asian economies, roughly level, both about to open up.

Then the lines split, and never came back together.

By 2025 China's income per person, about $14,000, was more than five times India's $2,675, and its total economy roughly 4.6 times larger (IMF; StatisticsTimes). Even after stripping out exchange rates and comparing on purchasing power, China is about 2.4 times richer per head. Two economies that shared a starting line now sit a generation apart. The interesting question is not that it happened but through which specific choices, because those choices are the template India keeps declining to copy.

What did China actually do differently?

China's rise is often waved away with "they are authoritarian, we are a democracy." That flatters India by dodging the mechanics. Three concrete, copyable decisions did most of the work.

One: it built manufacturing, deliberately and brutally. China rode a youthful workforce into factories, absorbing tens of millions of low and medium skill workers into globally integrated production, the phenomenon economists call the China shock (CaixaBank Research). The state was not a neutral referee. It offered land below market price, subsidised credit, sometimes took an equity stake, and protected favoured firms behind average industrial tariffs of 40 to 55% while they scaled (Aeon; academic estimates). This is the manufacturing job ladder that lifted close to 800 million Chinese out of extreme poverty. China's share of global poverty fell from 41% in 1990 to under 1% by 2020 (Our World in Data; World Bank).

Two: it invested at a rate India never approached. For two decades China ploughed 40 to 45% of GDP back into fixed capital: roads, ports, grids, plant, machinery. India's investment rate has hovered around 30 to 33% of GDP, and the private share of it fell to a decade low of 33% in FY24 (World Bank; RBI). Investment is the engine of future output. India has simply run a smaller engine.

Three: it chose factories over services as the entry point. India did the opposite. It leapfrogged straight to services, IT, back office work, finance, without ever building a mass manufacturing base. That is why manufacturing sits near 17% of India's GDP against roughly 27% for China, and why India's factories never became the great absorber of its workforce (Ideas for India; Goldar). Services generate enormous output per graduate. They cannot employ a hundred million people with a Class 10 education. China's model made ordinary workers productive; India's made a minority of skilled workers rich.

The demographic dividend was supposed to be India's China moment: a young workforce meeting a manufacturing boom. India got the workforce and skipped the boom.

Chapter 2 · The per capita mirage: why "fifth largest" flatters us

"Fifth largest economy" is a fact about size, and size is mostly a fact about population. India is a $4 trillion economy because it has 1.4 billion people, not because those people are individually well off. Divide the economy by its population and the triumph deflates. As of the World Bank's July 2026 classification, India remains a lower middle income country, a bracket it has occupied since 2007 (World Bank; Business Today).

Here it is worth correcting a common framing, because the sharper point is more damning than the usual one. It is not true that "Bangladesh and Sri Lanka reached middle income and India didn't." India is middle income. So is Bangladesh. The real story is which rung of the ladder each reached. India sits on the lower middle rung, unchanged since 2007, with a GNI per capita of roughly $2,760 in 2025. Sri Lanka reached the upper middle rung in 2019, lost it in the 2022 crash, and regained it in July 2026. Bangladesh sits on the same lower rung as India, except its income per person is projected to edge past India's this year.

Sri Lanka, a country that literally defaulted on its sovereign debt and ran out of fuel in 2022, has now crossed into upper middle income twice, a rung India has never touched (World Bank, July 2026). And Bangladesh, which India used to cite as the basket case it had left behind, is the more startling case.

Did Bangladesh really overtake India?

Yes, on one narrow measure, and probably briefly. The IMF's April 2026 outlook projects Bangladesh's income per person at $2,911 for 2026, edging past India's $2,812, a moment the former World Bank chief economist Kaushik Basu called shocking (The Daily Star; Scroll). Bangladesh had already been ahead of India for stretches around 2020. This is genuinely embarrassing for a country eight times its neighbour's economic size. But intellectual honesty requires two caveats that make it both less alarming and, in a deeper way, more alarming.

First, the crossover is largely an exchange rate artefact. Nominal income per person is a country's output converted into dollars at the market rate. When the rupee depreciates faster than the taka, India's dollar income shrinks on paper regardless of how much real output its people produced. On the purchasing power measure, which strips exchange rates out and asks what incomes actually buy, India remains comfortably ahead, roughly $11,800 against $10,300 in 2025. The IMF expects India to retake the nominal lead by 2027 (The Daily Star). So the overtaking is thin and probably temporary.

Second, and this is the part that should sting: the fact that it is even close is the indictment. A far smaller, more crisis prone economy has pulled level on income per head. You do not get within touching distance of a neighbour you are genuinely outperforming. The crossover is a symptom of exactly the efficiency problem this guide is about.

The flaw in "GDP per capita growth"

You will hear that India's income per person has multiplied several times since 1991, and it has: about 7.2 times over, from $371 in 1990 to $2,675 in 2025 (IMF). By the percentage yardstick, that beats almost every rich country on earth. America's income multiplied only 3.8 times over the same window. India's growth rate is the good news headline, and it is not fake.

But growth rates hide a scale problem, because percentages compound off the base you start from. When the base is $371, spectacular percentage growth adds a small absolute amount of income, and it is the absolute dollars and rupees that determine whether a family can afford a doctor, a scooter, a decent school. Put every major economy on the same yardstick, actual dollars of annual income added per person between 1990 and 2025, and the growth rate triumph inverts.

Three rows carry the argument.

The United States was already the richest large economy in 1990, grew at unremarkable single digit nominal rates, and ran lower inflation than India throughout. It still added roughly $66,000 of annual income per person. India, the perennial growth champion, added about $2,300. The US added nearly 29 times more absolute income per head while growing slower every single year. That is what a large base does: 4% of $24,000 is $960 a year, while 6% of $371 is $22. India never had a chance in absolute terms unless it grew at Chinese rates, which it never did.

China is the proof that the low base is an excuse, not a sentence. It started 1990 below India, $348 against $371, and added $13,620 per person, six times India's addition, from a smaller starting point. Its income multiplied 40 times over; India's multiplied 7. Vietnam started at a third of India's level and still added twice as much.

Even the unremarkable rows indict. Poland, emerging from communism, added more per person than China did. Indonesia and Thailand, nobody's idea of miracle economies, out added India roughly two and three fold. The only major economy in the table India out added is Bangladesh, by $66 per person over thirty five years.

This is the precise sense in which "India's GDP per capita growth" is a flattering statistic. The rate was world class; the income delivered was near the bottom of the table. And because the yardstick here is nominal dollars, the rupee's depreciation (Chapter 9) is quietly embedded in India's row: a slice of every year's rupee growth was clawed back before it ever showed up as dollar income.

Chapter 3 · Is PPP a cheat code?

Whenever the nominal numbers look bad, someone invokes purchasing power parity: in PPP terms India is already the third largest economy. PPP is a legitimate and useful adjustment. It accounts for the fact that a haircut or a plate of rice costs less in Patna than in Paris, so a dollar of income buys more in India. For comparing the living standards of the poor, PPP is arguably the fairer lens.

But PPP quietly does two things people rarely acknowledge. It cannot buy imports. Oil, semiconductors, machinery, defence equipment and dollar denominated debt are all paid at the nominal rate, so PPP wealth is useless at the border. And it can flatter a country into complacency: it makes India look like a giant precisely because things are cheap, which is another way of saying incomes are low. A cheap economy and a rich one are not the same thing.

The cleanest way to see through both nominal and PPP spin is to stop asking how big the economy is and start asking how many people are actually not poor. Here the World Bank's own poverty lines are blunt.

At the extreme poverty line, India has done something remarkable, and Chapter 10 both credits and audits it. But raise the bar to the standard of an ordinary upper middle income country, the China standard, and roughly four in five Indians count as poor, against about one in five Chinese or Vietnamese (Data For India; World Bank). India has not built a mass middle class so much as a large class living just above destitution. PPP adjusted bragging rights do not change what that money can buy.

Chapter 4 · The plumbing: an economy that leaks

Even good output is wasted if it cannot move. Two systems, the movement of goods and the movement of electricity, are where India's inefficiency is most physically visible, and most quantifiable.

What does it really cost to move goods in India?

For years the standard claim was that logistics cost India 13 to 14% of GDP, against a world average nearer 8%: a permanent tax on every manufacturer, quietly pricing Indian products out before they leave the warehouse (Ministry of Commerce; Georgetown Journal of International Affairs). That gap, across the whole economy, ran to lakhs of crores a year.

Here the sources genuinely disagree, and the disagreement belongs in the story. In September 2025 a DPIIT study prepared by NCAER re estimated India's logistics cost at 7.97% of GDP for FY24, below the long cited 13 to 14%, and below China's own estimate of 14.4% (Business Standard; DPIIT). It is the first rigorous Indian estimate built on nationwide surveys, and it may well be closer to the truth. It is also a number produced by the department that runs the logistics improvement programmes it flatters. Both things can be true: the old figure was probably too high, and the new one deserves independent scrutiny before it becomes gospel. The safest reading is that India's logistics have improved materially. Its World Bank Logistics Performance Index rank rose from 54th in 2014 to 38th in 2023, while still trailing China and Vietnam on the underlying index.

Underneath the headline number sits the real inefficiency: mode. Rail freight costs about ₹1.96 per tonne kilometre; road costs ₹11.03, nearly six times more. Yet roughly two thirds of Indian freight still travels by road (DPIIT; Georgetown JIA). Decades of underbuilt rail freight pushed cargo onto the most expensive mode. This is improving, and credit is due: rail freight tonnage rose from 1.08 billion tonnes in FY14 to about 1.6 billion in FY25, and 31,180 km of track were added in a decade. But the country spent thirty years paying the road freight premium first.

How much power simply vanishes?

Electricity is the starker case, because here India has long held an unwanted record. A large share of the power India generates never gets paid for, lost to technical leakage in ageing grids and to outright theft. These Aggregate Technical and Commercial (AT&C) losses have fallen sharply thanks to reform. The level remains far above the world's.

India's losses fell from 22.6% in FY14 to 15.4% in FY23, real progress (Ministry of Power). But near a sixth of the electricity India produces still generates no revenue, against a global norm of about 7%. And the consumer pays for this twice: once in tariffs inflated to cover the losses and the cross subsidy that props up free farm power, and again in the unreliability that forces factories and hospitals onto expensive diesel back up. High cost and unreliable supply together are the worst of both worlds, and a direct drag on the manufacturing India says it wants.

Chapter 5 · The workaround state: rules that manage failure instead of fixing it

There is a category of Indian inefficiency that no infrastructure chart captures, because it is not the absence of an asset. It is the presence of a rule, written to manage the absence, that ends up costing more than the absence did. Once you learn to see the pattern, it is everywhere.

The truck that is banned from the city it feeds

Start with the specimen every Indian has driven past. Most large Indian cities bar heavy trucks from entering during daytime hours, because the road network cannot carry commuters and freight at once. The ban does not make the freight disappear. It makes thousands of loaded trucks queue at city borders for the night window, burning driver hours and diesel while parked. The arithmetic of that queue shows up in national fleet statistics: a truck in India works about 20 days a month against 25 in developed markets, and covers roughly 250 to 400 km a day against 700 to 800 abroad (TCI and IIM Ahmedabad study; NITI Aayog).

Think through what a fleet at half utilisation means. To move the same goods, India needs nearly two trucks, two drivers, two loans and two sets of tyres for every one a developed economy needs. That is capital, labour and fuel conscripted into standing still, and every rupee of it lands in the price of whatever the truck carries. The logistics firms themselves have said what would fix it: outer ring roads and bypasses around the tier two and tier three cities that trucks are currently forced to either enter or wait outside (industry statements via PTI). But here is the trap that gives this chapter its name. The daytime ban manages the congestion well enough that the bypass is never anyone’s emergency. The workaround removes the pressure that would have forced the fix. The rule written for one summer of bad traffic quietly turns twenty.

The law that protects workers by preventing jobs

The largest specimen of the genre is not on the roads. Under India’s industrial disputes framework, a manufacturing firm that crosses roughly 100 workers has historically needed government permission to retrench anyone. The rule was written to protect workers. Its actual effect, documented in the government’s own Economic Survey, is that firms refuse to grow past the threshold at all. The Survey named them dwarfs: firms more than ten years old that still employ fewer than 100 people.

Read the two halves of the chart against each other. The firms the policy regime incentivises to exist, small and permanently small, are over half of all organised manufacturers and deliver 8% of its productivity. The firms it discourages, those past 100 workers, are 15% by number and deliver three quarters of the jobs and nearly 90% of the output. When Rajasthan loosened these rules in 2014, the Survey’s own analysis found the number of large factories and total employment grew faster there than in the rest of India. A law drafted to protect the worker ensures that most Indian workers never meet a formal employer at all. It protects the ten percent inside the gate by keeping the ninety percent outside it.

The same pattern, wherever you look

Once named, the pattern repeats across the economy. Power: farm electricity is free, so the distribution companies recover the loss by loading industrial tariffs with a cross subsidy, so factories, the one customer class that pays reliably, defect to captive diesel and solar, so the grid’s finances worsen and tariffs rise again (RBI; PRS). A subsidy meant to help the poor makes the factories that would employ them uncompetitive. Food exports: whenever onion, wheat or rice prices rise, exports are banned overnight to protect urban consumers, denying the farmer income at the exact moment the market finally pays, after which the state compensates his poverty with the subsidies of Chapter 8. The same government that promises to double farm incomes forbids farmers from selling at the top. Cities: building height and floor space are rationed by some of the world’s most restrictive rules, then the resulting sprawl and unaffordability are managed with rent controls and housing schemes rather than by letting the city build upward.

The structure is identical every time. The state inherits or creates a scarcity. Instead of removing the scarcity, it writes a rule to ration it. The rule shifts the cost onto the productive economy, quietly, where no budget line records it. And the rule acquires a constituency, the protected insiders, the compensated farmers, the commuters spared the trucks, which makes it politically permanent long after the original scarcity could have been fixed. Economists call this a second best trap. In India the second best has a way of becoming the only best on offer, which is why the truck is still parked at the border a generation after the ring road was first proposed. These are not corruption stories, and that is what makes them expensive: each rule is sincere, defensible in isolation, and collectively they are a tax on every productive hour in the country that no finance minister ever has to announce.

Chapter 6 · The double invoice: what does the taxpayer get back?

Every audit of a business ends at the same question: what does the customer get for the price? Run that question on the Indian state and the answer is the strangest entry in this whole ledger. The popular complaint says India is overtaxed. The data says something more precise and more damning: India is moderately taxed in aggregate, brutally narrowly taxed in practice, and its citizens then pay for most of what the tax was supposed to buy a second time, privately, out of their own pockets. The true price of the Indian state is not the tax. It is the tax plus the second bill.

Who actually pays, and how

Start with the honest correction. India's combined tax take, centre and states together, runs around 18% of GDP, against an OECD average near 34%. By that headline, India is a low tax country. The burden feels heavy for two structural reasons, and both are choices.

First, the base is a sliver. On the Finance Ministry's own statements to Parliament, fewer than two in every hundred Indians pay any income tax at all. The formal salaried class carries direct taxation for a nation of 1.4 billion, with tax deducted at source before the salary ever arrives, while agricultural income of any size is exempt by law and most of the informal economy sits outside the net.

The second bill: paying again for what the tax should have bought

Now the delivery side, which is where the vendor analogy stops being a metaphor. In a functioning state, the tax buys a bundle: schooling, healthcare, clean water, reliable power, safe streets, usable roads. In India, the household that has already paid its taxes then goes out and buys nearly the whole bundle again.

Health is the measurable core of it. Indians pay 48.2% of all health expenditure out of pocket, at the point of illness, down from 64.2% a decade earlier but still around two and a half times the world norm (Economic Survey; PIB; WHO). The consequences are not abstract: roughly nine crore Indians cross the catastrophic threshold each year, spending over a tenth of household consumption on healthcare, and studies estimate medical bills alone push five to six crore people below the poverty line annually (Frontiers in Public Health; PMC). The state, as Chapter 7 details, spends under 2% of GDP on public health. This is the other half of that entry: the gap is not unspent. It is spent by patients, at private counters, at the worst possible moment, with no risk pooling at all.

And health is only the largest line on the second invoice. The middle class household that pays income tax also pays private school fees and coaching classes because the government school failed, buys an inverter and pays diesel genset charges because the grid of Chapter 4 fails, installs a water purifier and books tanker deliveries because the tap cannot be trusted, pays the housing society for guards because policing is thin, and pays a toll on the highway that a fuel cess already financed. None of this appears in any tax to GDP ratio, but all of it is a payment for state failure, priced into every family budget. Add the second bill to the first and the effective rate an urban Indian family pays for governance is far closer to the European number than the Indian one. The difference is what arrives in return.

Where the first bill actually goes

If the money is not coming back as services, where does it go? Mostly to the state's own past. Roughly a fifth of every rupee in the union budget goes straight out again as interest on old borrowing, before a school or clinic sees anything. In the states the pattern is harsher: committed expenditure, meaning salaries, pensions and interest, consumes the majority of revenue in many of them before a single public service is delivered (PRS), and the escalating transfer schemes of Chapter 8 compete for what remains. Punjab is the endpoint, already met in this piece: interest alone at two and a half times the state's entire capital budget. The Indian taxpayer is not primarily buying services. She is servicing debt raised by earlier governments, paying the wage bill of the state itself, and funding the next election's welfare promise, in roughly that order.

Any private vendor that charged a premium, delivered a fraction of the order, and billed the customer again for the shortfall would lose the customer. The Indian state's customers cannot leave, except the richest few thousand who emigrate each year, and the informal majority who were never inside the formal bargain to begin with. What remains is the narrow band in the middle: formally employed, fully taxed at source, fully self provisioned, and the only party in the country paying retail for a state it does not use.

Chapter 7 · The wasted dividend: a young country the state under equips

India's most hyped asset is its youth: a median age around 28, a working age share rising towards 65% near 2041, a demographic window open until roughly the 2050s (UNFPA). This is real, and it is finite. A dividend is only paid if the young are healthy, educated and employed. On the state's two great tools for making that happen, health and knowledge, India has chronically underspent.

Health: under 2% of GDP, and it shows

Public health spending has been stuck around 1.8 to 2% of GDP for years, below India's own 2017 policy target of 2.5%, and far below peers at a similar income level (Economic Survey; Business Standard). Because the public system is thin, Indians pay for health largely out of pocket, one illness away from poverty.

The consequences show up in the body. India's 2024 Global Hunger Index score of 27.3 and its child wasting rate of 18.7%, among the highest in the world, sit awkwardly beside the growth story (GHI; Aeon). You cannot build a productive workforce out of stunted children. This is not a moral aside; it is an economic one. Under investment in health today is a productivity tax collected for the next forty years.

R&D: the innovation gap that compounds

If health is the workforce's floor, research is its ceiling. India spends about 0.64% of GDP on research and development, a figure essentially frozen for a decade, against China's 2.4% (DST; Economic Survey 2025 to 2026). The innovation led economies, the US, Israel, South Korea, spend between 2.5 and 5%.

The composition is worse than the level. In China, Japan, Korea and the US, business enterprises fund more than 70% of national R&D. In India, government and universities fund about 59%, and private industry barely 40% (DST). India produces world class early stage research, it is now the third largest producer of scientific papers, but consistently fails at the stage between a lab result and a shippable product, the valley of death (Economic Survey 2025 to 2026). And the raw human input is thin: about 262 researchers per million people, against several thousand in Korea or Israel. A young country that under funds both the health and the ingenuity of its young is quietly spending its dividend before it arrives.

And half the workforce is missing

The dividend also assumes people actually work for pay. India's female labour force participation has long been among the lowest in the G20 and well under the world average of around 51% (World Bank; ADB). It has risen recently in the official surveys, but much of that rise is rural self employment and unpaid family labour rather than new paid jobs, which is a weaker thing than the headline suggests. Half of India's potential workforce remains largely outside the paid economy.

Chapter 8 · Freebies: welfare, or the price of a vote?

Governments spend money in two broad ways. Capital expenditure builds a lasting asset: a road, a grid, a port. Revenue expenditure is recurring: salaries, interest, and subsidies, including the ever expanding menu of election freebies. The difference between them is not ideological. It is arithmetical, and India's own public finance economists have measured it.

The landmark NIPFP study by Bose and Bhanumurthy estimated India's fiscal multipliers directly. Every ₹1 the government spends on capital raises GDP by about ₹2.45, because it crowds in private investment and compounds. Every ₹1 spent on subsidies and transfers raises GDP by about ₹0.98 to ₹0.99: you get back roughly what you put in, and nothing durable is left standing. The tax multiplier is around minus 1. Later studies find the cumulative capital expenditure multiplier is 2.4 to 6.5 times the revenue one (NIPFP; Journal of Quantitative Economics).

Twenty years of running the experiment both ways

India has actually tested the multiplier table above, at national scale, in both directions. The last twenty years of union budgets split cleanly into two eras, and the pivot year is 2014.

Era one, FY05 to FY14: the subsidy decade. Major subsidies on food, fertiliser and fuel exploded from roughly ₹46,000 crore to ₹2.55 lakh crore, a five and a half fold rise, climbing from around 1.3% of GDP in the early 2000s to about 2.5% at the FY13 peak (Union Budget documents; Springer long run study). Capital expenditure, meanwhile, crawled: stuck near 1.6 to 1.7% of GDP for the entire decade. By FY13 the arithmetic had fully inverted priorities: the centre spent ₹2.57 lakh crore on subsidies against only ₹1.66 lakh crore building anything, sixty five paise of capex for every subsidy rupee.

Era two, FY15 to FY26: the capex decade. The ratio flipped. Capital expenditure rose from ₹1.88 lakh crore in FY14 to a budgeted ₹11.2 lakh crore in FY26, roughly six fold, taking it from 1.7% to about 3.1% of GDP, while major subsidies were rationalised from that 2.5% peak to 1.1% of GDP by FY26, interrupted only by the pandemic spike to 3.6% in FY21 when free grain was doubled (PIB; Economic Survey 2025 to 2026; Rau's compilation of budget data). Petrol and diesel subsidies were abolished outright, LPG was targeted, and leakages were squeezed through direct transfer. By FY26, the centre spends about ₹2.60 on capex for every subsidy rupee, four times the FY13 ratio.

Did the outcomes track the multipliers? Broadly, and visibly, in this piece's own earlier numbers. The subsidy decade ended in textbook fashion for an economy starving its capital account: double digit consumer inflation from 2010 to 2013, a current account deficit near 4.8% of GDP, and the August 2013 taper tantrum that crashed the rupee and put India in the Fragile Five, while the infrastructure it had not built showed up as the 22.6% power losses and the punishing logistics costs of Chapter 4, both measured at almost exactly the moment era one closed. The capex decade then rebuilt the plumbing this guide audited: logistics cost estimates falling from the old 13 to 14% claims towards 8%, the LPI rank climbing from 54th to 38th, AT&C losses down to 15.4%, expressways from 93 km to 2,474 km, trucks covering more ground after GST. The multiplier table is not a theory. India ran both settings of the dial and got both predicted outcomes.

Three caveats keep the story honest, because era two has its own spin. First, the capex number itself is flattered: independent analysis finds reported capital expenditure includes loans to states and misclassified items, and that adjusted central capex, measured strictly as acquisition of assets, has actually declined in recent years, with total public sector investment peaking as far back as FY18 (CSEP). Second, the theoretical payoff of capex, crowding in private investment, has not yet arrived: the private share of investment fell to a decade low of 33% in FY24 (Chapter 1), which is why critics dispute how much of the multiplier actually materialised. And third, the pivot was only half a pivot, because the federation ran it in reverse: as the centre swapped subsidies for capex, the states swapped capex for subsidies, with the ₹1.68 lakh crore of cash transfers, the free power, and a Punjab whose interest bill is two and a half times its capital budget, all documented in this chapter. One level of government spent twenty years learning the lesson of the multiplier table. The other spent the same twenty years unlearning it.

But is India different? Answering the strongest defence

The standard defence of freebies deserves to be stated at full strength before it is answered. India is not Denmark. Four in five Indians live below the upper middle income poverty line (Chapter 3). There is no unemployment insurance worth the name, health costs are paid out of pocket, and a single illness or failed monsoon can wipe out a family. In that world, the argument goes, transfers are not a luxury or a bribe. They are the safety net a rich country would have built through institutions, delivered through the only channel a poor state has: cash and free things. And the argument has evidence on its side, because the World Bank explicitly credited free and subsidised food transfers for part of India's collapse in extreme poverty, from 16.2% to 2.3% in a decade (Chapter 11).

So the honest answer is not that welfare is waste. It is that India's own data now lets us separate two things that the word freebie lazily bundles together. Transfers that are targeted at the destitute, tied to a real deprivation, and small relative to the budget demonstrably worked: the public distribution system helped pull 171 million people over the extreme poverty line. What has exploded since is a different animal: untargeted, universal, permanent giveaways, priced not by need but by the electoral calendar. The question is not whether India needs a safety net. It is whether what India is actually buying is a safety net.

What is India actually buying? The escalation, measured

Look at the fastest growing item. Unconditional cash transfers to women existed in two states in FY23. By FY26, twelve states run them, and they are budgeted to cost ₹1.68 lakh crore in a single year, 0.5% of India's entire GDP, more than double the share of two years earlier (PRS Legislative Research). Every scheme was launched in the run up to an election, and the pattern is now so reliable that analysts have a name for it: competitive welfarism. Overall state subsidy budgets, which include farm loan waivers, free power and free transport, have risen roughly 2.5 times since FY19 to over ₹4.7 lakh crore (RBI). And the free power sits on top of a second, hidden bill: the state owned distribution companies forced to carry it owed ₹6.8 lakh crore by FY23, about 2.5% of GDP, debt that is a contingent liability of the very states writing the subsidy cheques (RBI).

The fiscal damage is no longer hypothetical. Six of the twelve implementing states are budgeting revenue deficits in FY26, meaning they borrow to fund running costs. Strip the cash schemes out and the picture flips: Karnataka alone moves from a 0.3% revenue surplus to a 0.6% deficit of GSDP because of its scheme (PRS). Karnataka was in revenue surplus the year before its Gruha Lakshmi scheme launched and has run a deficit every year since (Business Standard). And the retreat has already begun, which tells you the promises were never fiscally real. Maharashtra's Ladki Bahin scheme, credited with winning the 2024 state election, has had its allocation cut 42% in two years, its beneficiary list shrunk from 24 million to 16.6 million after identity verification, and a parallel transfer to women farmers cut by 67% (Business Standard). The money runs out. The promise, and the expectation it created, does not.

Punjab: the 28 year experiment, and its results

If you want to know where the freebie road ends, India has already run the experiment, with a control group of 27 other states. Free electricity for Punjab's farmers was born in a game of political one upmanship before the 1997 assembly election (The Tribune). It has never been withdrawn since, by any party. Twenty eight years later, the results are in, and they are measurable.

Start with the resource. Because pumping water costs a farmer nothing, Punjab extracts groundwater at 149% of the rate nature recharges it, the highest in India (Abdul Latif Jameel Poverty Action Lab, from 2013 government estimates). Of 150 assessed blocks, 114, over three quarters, are officially overexploited, and the water table falls by roughly a metre a year (Central Ground Water Board; The Tribune). Power officials describe locked motor rooms whose tubewells start pouring water into empty fields the moment the free eight hour supply switches on.

Now the productivity, which is where this chapter's argument lands. Electricity use per hectare in Punjab has risen for decades while agricultural output per unit of electricity has been falling (Springer, Performance of Agriculture in Punjab). Read that again: the subsidy is buying less and less production per unit every year. Free power did not make Punjab's farms more productive. It locked them into water hungry paddy on a depleting aquifer, so each year needs deeper borewells, bigger pumps and more subsidised electricity to stand still. This is the proven productivity damage from freebies, and it runs through distorted prices, not lazy farmers. It is also regressive: free power flows in proportion to land owned and pump capacity, so the largest landholders capture the most subsidy while the landless labourer gets nothing.

And the fisc. Punjab's debt has reached ₹3.78 lakh crore, around 47% of GSDP, the worst ratio among major states (NITI Aayog Fiscal Health Index; CAG). Interest payments of ₹26,500 crore a year consume nearly a quarter of revenue receipts and are more than two and a half times the state's entire capital expenditure of ₹10,244 crore (Punjab Budget FY26). The CAG found that only 26% of Punjab's net borrowings went into capital spending; the rest funded consumption. A state that entered the 1990s among India's richest per person now borrows to pay interest on the money it borrowed to give things away. Punjab is not an aberration. It is every freebie state's balance sheet, run forward 28 years.

Do freebies make people stop working? The evidence, honestly

Here the popular argument needs correcting, because the piece is only as strong as its weakest claim. The intuition that cash handouts make the poor withdraw from work is the one part of the freebie critique the research does not support. The most rigorous global review, by Banerjee, Hanna, Kreindler and Olken, examined randomised evaluations of cash transfer programmes across developing countries and found no systematic evidence that they reduce adult work (World Bank; VoxDev). Poor households mostly cannot afford leisure; small transfers get spent on food, debt and school fees, and sometimes fund the tools and stock that let people work more.

The real participation story is subtler and, in its way, worse. India's female labour force numbers have risen in recent surveys, but the rise is dominated by unpaid family work and marginal rural self employment, while the share of rural women in regular salaried jobs has actually fallen, from 10.5% in FY18 to 7.8% in FY24 (PLFS, via ORF). A ₹1,500 monthly transfer does not stop a woman working. What it does is let the state book a welfare achievement while the thing that would actually transform her earnings, a formal job near where she lives, goes unbuilt, partly because the money that would have built it went out as the transfer. The freebie does not buy idleness. It buys the appearance of progress at the price of the substance.

Eighteen years, one rung

Put the two clocks side by side. India entered the lower middle income band in 2007 and has not left it in eighteen years (Chapter 2). The freebie escalation is concentrated in the last seven: subsidies up 2.5 times since FY19, women's cash schemes from two states to twelve in three years, every expansion timed to a poll. The pie has refused to grow into the next band, and the political system's response has been to compete ever harder over how the existing pie is handed out. Redistribution replaced growth as the product Indian politics sells, because a road pays off in ten years and a transfer pays off in the next election, and the multiplier table at the top of this chapter says precisely what that trade costs: a rupee that would have become two and a half, spent to remain one.

Chapter 9 · The rupee and the growth illusion

One thread ties the per capita mirage, the Bangladesh crossover and the flawed growth statistic together: the currency. At Independence a dollar bought a little over three rupees; today it buys nearly ninety (see our guide on the rupee's journey). The great devaluations, 36.5% in 1966 and roughly 19 to 20% in the 1991 crisis, and the managed slide since have shaped how India's growth reads in a way worth understanding.

Depreciation cuts both ways. Measured in rupees, a weakening currency mechanically inflates nominal GDP and can flatter the growth story. Measured in dollars, the same depreciation compresses income, which is exactly why India's dollar income per person lags even when real output grows, and why a taka that depreciated slightly slower let Bangladesh nose ahead. When people say India's growth statistics are flawed, this is the honest core of it: a meaningful slice of measured expansion is nominal and currency driven, not a real rise in what a family can buy or convert abroad.

The deeper contrast with China is about intent. China ran a deliberately undervalued, tightly managed currency for years as an export weapon: cheap yuan, cheap exports, a manufacturing juggernaut, reserves stockpiled. India's rupee slid too, but its export response was muted, because a currency edge is worthless if your logistics leak, your power is unreliable, and you never built the factories to sell (Chapters 1 and 4). China depreciated on purpose, into a manufacturing base it had built. India depreciated by gravity, into one it hadn't. Same monetary phenomenon, opposite strategic result. That, in one line, is the efficiency gap.

Chapter 10 · The pyramid: who is actually carrying this economy?

Aggregate statistics describe an India that does not exist at any address. To see the machine, break the economy into its bricks: who produces, who pays, who consumes, and who depends. The most useful segmentation comes from Blume Ventures' Indus Valley reports, and its numbers deserve to be famous.

India1 is about 140 million people, 10% of the population, with income per head near $15,000: were it a country, it would be the world's tenth most populous and comfortably upper middle income. It drives two thirds of all discretionary spending in India. Blume's Rule of 30 finds that every serious estimate of the true consuming class converges on roughly 30 million households, and that this is the ceiling, not the floor: air passenger traffic, the cleanest proxy for the class's size, has barely grown since the pandemic. India1 is deepening, spending more per head, not widening. India2 is roughly 300 million aspirants at Indonesia like incomes near $3,000: salaried juniors, small traders, skilled workers, heavy users of everything digital and reluctant payers for any of it, facing stagnant wages with no social security. India3 is about a billion people at around $1,000, comparable to sub Saharan Africa, with essentially no discretionary spending, dependent on state transfers and informal work.

Now overlay the earlier chapters onto the bricks and the wiring becomes visible. India1 and the top of India2 are the fewer than two in a hundred who pay income tax (Chapter 6). They are the formal workforce whose salaries fund, through the toll booth tax system, the free grain that feeds the 81 crore of India3 and the cash transfers of Chapter 8. They also pay the second invoice, the private school, the hospital, the genset, because they cannot use the state their taxes bought. And they are the customers whose spending is two thirds of the discretionary economy. One brick, four loads: taxpayer, transfer funder, self provisioner, and consumer of last resort. India3 depends on India1 through the fisc. India2 aspires to become India1 and increasingly borrows to imitate it. Nothing in the structure depends on India3 producing more, which is the quiet tragedy of the whole arrangement.

The squeeze: India1's own economics are breaking

The pyramid would be merely unjust if the top brick were thriving. The data says it is being squeezed from three directions at once, and this is where the growth model starts eating its own base.

Start with inflation, measured honestly. Official CPI runs at 5 to 6%, but half the CPI basket is food, and food is a small share of what India1 actually buys. Reweight for the middle class basket, school fees, healthcare, housing, services, domestic help, and analyses find the urban middle class cost of living doubling roughly every eight years, an effective inflation rate near 9% (ThePrint). The class that funds everything experiences the country's highest inflation, and the official index is not built to see it.

Second, that basket is increasingly dollar indexed, which quietly breaks the oldest comfort in Indian macroeconomics. Fuel is priced off dollar crude. Phones, laptops and their import content are dollar goods. The foreign university, the overseas holiday, the streaming bundle: outward remittances under the liberalised scheme are dominated by travel at $17 billion, over half the total (Indus Valley 2025). The textbook says rupee depreciation helps a country by making exports cheap; that logic works for an economy that manufactures and exports, which Chapters 1 and 9 showed India never became. For a class that earns in rupees and consumes in dollars, the slide from 83 to nearly 90 is not an export subsidy. It is a pay cut, applied annually, to precisely the people carrying the pyramid. Devaluation without a manufacturing base does not create winners; it just picks different losers.

Third, the balance sheet is already showing it. Net household financial savings crashed to about 5.1% of GDP in FY23, a 50 year low, from 11.5% just two years earlier; household liabilities climbed to 6.4% of GDP, near a 17 year high; and retail credit excluding mortgages has trebled since 2019, with a rising share funding consumption rather than assets (RBI, Financial Stability Report; Drishti; ThePrint).

A household that saves less and borrows more to maintain consumption is a household whose real disposable income is falling, whatever its nominal salary says. And the arithmetic of a consumption driven economy is unforgiving: private consumption is about 60% of India's GDP. When the class that supplies two thirds of the discretionary part of it starts diverting income to debt service, its marginal propensity to consume falls, and the demand engine sputters exactly where it is narrowest.

The K, visible in the showroom

The evidence is already on the street. Vehicles above ₹10 lakh are 48% of car sales and growing at twice the market rate, while entry level cars decline and two wheeler sales sat below their 2018 level for years (Indus Valley; The Wire). The flattering spin is that Indians are premiumising, leaping from scooters to SUVs. Maruti's own chairman R C Bhargava dismissed it: consumers do not leap like that; the rich are buying more while the middle stalls on stagnant incomes (The Wire). This is what a K shaped economy looks like at ground level: the top of India1 accelerating, the bottom of India1 and all of India2 cutting back, and India3 spending its transfer the day it arrives.

Put the chapter's pieces together and the structure of the problem is stark. India built a consumption led growth model powered by 10% of its people; taxed that same 10% narrowly and twice over; let their true cost of living compound at 9% while their currency of consumption quietly became the dollar; and left the other 90% dependent on transfers those 10% fund. The pyramid is not resting on its base. It is balancing on its tip.

Chapter 11 · The honest ledger, audited: what did India really get right?

A critique that only accuses is propaganda. India has real achievements, and they are not consolation prizes. But an honest ledger cannot stop at booking the wins; it has to audit them, because each headline number turns out to be narrower than it looks. Take the four biggest, one by one.

Did 171 million people really escape poverty?

The claim: India lifted about 171 million people out of extreme poverty between FY12 and FY23, with the extreme poverty rate collapsing from 16.2% to 2.3% at the $2.15 line, and to 5.3% at the newer $3 line (World Bank). Something real sits inside that number: destitution level deprivation, going hungry, owning nothing, has genuinely and sharply declined, and multiple independent measures agree on the direction. But the size and meaning of the claim deserve four audits, and the same state supplies the strongest counter evidence.

First, look at the line. Three dollars a day at purchasing power parity is roughly ₹90 of daily consumption, about ₹5,200 a month for a family at the old $2.15 line, sums independent economists have called absurdly low (EastPost survey of expert criticism). Crossing it does not mean entering the middle class. It means no longer being destitute. Chapter 3 already showed what happens when the bar moves to an ordinary upper middle income standard: four in five Indians are still under it.

Second, the yardstick changed mid measurement. India's 2022 to 2023 consumption survey adopted a new methodology (shorter recall periods, more items, different sampling) that is not comparable with 2011 to 2012. The effect is not small: applying the newer recall method to the old 2011 to 2012 data cuts measured poverty from 21.8% to 12.2%, an eleven point difference generated by questionnaire design alone. The World Bank itself cautions that this transition, by itself, shifts the poverty trend significantly, and that national trends should be judged on national poverty lines, which India has not officially updated since 2011 to 2012 (Down To Earth; EastPost). In one year the Bank's own estimate moved from 12.9% (using a private survey) to 2.3% (using the new official one), a fall statisticians called implausible without an economic transformation nobody observed.

Third, and most fundamentally: a meaningful part of the measured consumption that lifts households over the line is not income they earned. It is the imputed value of free and subsidised foodgrain from the state, monetised and added into their consumption by the estimating economists (The India Forum; The Federal). A household counted as not poor because the government feeds it is one budget decision away from being poor again. The poverty exit, for tens of millions, is conditional, not earned.

Which leads to the fourth audit, the one that requires no statistics degree: the government's own behaviour. The same state that celebrates 5% poverty provides free monthly foodgrain to over 81 crore people, 57% of the population, and has extended the programme for years at a cost in lakh crores. Roughly 30% of rural households still seek MGNREGA work at below market wages (The Wire). If only one Indian in twenty were poor, neither programme would need to exist at a tenth of its scale. Revealed preference is a form of data, and the state's revealed estimate of vulnerability is 57%, not 5%. The honest reconciliation of the two numbers is this: India genuinely crushed destitution, and it did so substantially by transfer rather than by transformation. The safety net worked. The ladder, the formal job that makes the net unnecessary, is what 800 million people are still waiting for. Alternative estimates using India's own Rangarajan methodology put poverty at 26%, and the truth most likely lives between the celebration and the critique.

What did UPI actually solve?

The claim: UPI processed 21.7 billion transactions worth ₹28.3 lakh crore in January 2026 alone, roughly 49% of the world's entire real time payment volume, live in eight countries (NPCI; IMF; ACI Worldwide). All true, and genuinely world leading. Now the audit, and it rhymes uncomfortably with Chapter 8.

UPI is free because the government mandated a zero merchant discount rate in January 2020. Zero price drove the miraculous adoption, and zero price means the system has no revenue model. The costs, infrastructure, fraud, settlement, support, estimated by industry participants in the thousands of crores a year, are absorbed by banks and payment firms, partially offset by a government incentive that has been shrinking year on year even as volumes explode (industry and budget reporting). The result is a national utility run at a structural loss, with the market consolidated into effectively two foreign owned apps handling over four fifths of volume, because nobody else can afford to compete for unprofitable transactions. India built the world's best payment rail and priced it like a freebie, and Chapter 8 explained what zero pricing does to any system's long run health.

And the deeper test: did it replace cash? No. Currency in circulation has kept climbing to record absolute highs, and as a share of GDP sits roughly where it did before demonetisation (RBI data). UPI did not digitise India's cash economy; it added a digital layer on top of it, while the informal economy, the part that generates no tax and no credit history, kept running on notes. What UPI genuinely solved is real: friction, transaction cost, inclusion at the margin, and, most valuably, a credit trail that lets a street vendor be underwritten for the first time. That is a true public good. But it is a payments achievement, not a formalisation achievement, and its own economics remain an unsolved subsidy.

The external fortress: strong walls, wrong bricks

The claim: record remittances of $135.4 billion in FY25, over $700 billion in reserves, a current account deficit under 1% of GDP. All true. Now look at what kind of money is arriving. In FY25 India's net foreign direct investment, gross inflows minus what foreign investors took out and what Indian firms sent abroad, collapsed to $0.4 billion, down 96% from $10.1 billion the year before and $28 billion in FY23 (RBI State of the Economy, May 2025). Gross inflows stayed healthy at $81 billion, but repatriation surged to $51.5 billion and Indian companies themselves invested $29.2 billion overseas, with financial services and trading, not factories, dominating both directions (Business Standard).

Set the two flows side by side. Remittances, the wages of exported labour, brought in roughly three hundred times more net money than foreign direct investment, the capital that builds plants and creates jobs. The RBI framed the FDI collapse as the sign of a mature market that investors can exit smoothly. Perhaps. But an external account this shape describes a country that exports its workers and its services, imports their salaries, and attracts almost no net productive capital, while its own corporates increasingly deploy theirs elsewhere. The fortress is real. It is built out of consumption money, not construction money, and it guards an economy that foreign factory builders, on net, are declining to enter.

Services and pharma: the ladder AI is sawing

The claim: IT and business services exports and a world supplying pharmaceutical industry are genuine, unsubsidised excellences. True, and they remain so. But the audit here is about the future tense. The IT services model is a pyramid: mass recruitment of junior engineers doing standardised coding, testing and support, the exact layer generative AI automates first. The results are already in the filings: the top Indian IT firms' combined headcount declined in FY24 for the first time in over two decades, entry level hiring has collapsed from its post pandemic peak, and in 2025 TCS, the industry's bellwether, announced roughly 12,000 job cuts, about 2% of its workforce, even while revenues grew (company statements; industry reporting). Industry growth has slowed to low single digits from a historic double digit norm.

This matters beyond one sector, because of who climbs this ladder. IT services is the single largest creator of formal, well paid jobs for India's educated youth, the engine that built much of the India1 consuming class of Chapter 10. The one escalator India constructed for its graduates happens to be the most AI exposed job category in the world economy, at exactly the moment the demographic bulge (Chapter 7) needs it most. Pharma and high end engineering research are more defensible, and India may yet capture AI era work. But a country whose plan for its educated young was "services will absorb them" now needs a second plan, and the manufacturing base that should have been the second plan is the one Chapter 1 showed was never built.

Chapter 12 · The verdict: efficient at the top, inefficient at scale

Line the findings up and a single pattern emerges. India is highly efficient at the frontier: software, digital payments, pharma, space, a world beating elite of engineers and founders, though Chapter 10 shows even these frontier wins carry asterisks. It is chronically inefficient at scale, at the unglamorous business of making a hundred million ordinary workers a little more productive each year.

Every chapter is a variation on that theme. India invests too little, roughly a third of GDP against China's two fifths and more. It spends too much of what it does invest on the low multiplier things, subsidies that return a rupee per rupee instead of capital that returns two and a half. It pushes the output through infrastructure that leaks a sixth of its power and spent decades paying a road freight premium. It wraps every unfixed failure in a workaround rule, the truck ban, the 100 worker cliff, the export ban, that costs more than the failure and outlives it. It bills its narrow base of formal taxpayers once at source and then again at the hospital counter, the school gate and the diesel pump, delivering interest payments and election promises in return. It under equips its workforce, with public health below 2% of GDP, research below 0.7%, and half its women outside paid work. It powers its consumption model with a 10% sliver whose savings are at a 50 year low and whose cost of living compounds far above the official index. Then it measures the result in a depreciating currency that makes even the real gains look thin in dollars, while that same depreciation raises the price of the dollar indexed basket its consuming class actually buys. None of these is fatal alone. Together they compound into the defining fact: an economy that grows fast and enriches its people slowly.

The World Bank's own arithmetic frames the stakes. To become a high income country by 2047, India needs to sustain roughly 7.8% real growth for two decades and lift its investment rate from about 33.5% towards 40% of GDP. Its income per person would have to rise roughly eightfold, a feat few nations have managed (World Bank Country Economic Memorandum, 2024). That is not impossible. But it is a different, harder project than the one the growth headlines describe, and it runs directly through every inefficiency audited above. The fixes are known and mostly domestic: build, invest, spend better, keep the lights on, school and heal the young, and let the currency reflect a real manufacturing base rather than substitute for one.

India's tragedy is not that it failed. It is that it succeeded narrowly, and called it success broadly.

The good, the bad and the ugly

The good. A frontier that competes with anyone: UPI, IT, pharma, space. A fortress balance sheet, with $700 billion in reserves and the world's largest remittance inflows. And a genuine, transfer assisted collapse in destitution. India's ceiling is world class.

The bad. An investment rate stuck well below China's, a manufacturing base that never absorbed the workforce, a thicket of workaround rules that keep trucks parked and firms at 99 workers, public health under 2% of GDP, research frozen at 0.64%, and a rising subsidy bill that crowds out the capital spending that actually compounds. India's floor is far too low.

The ugly. Four in five Indians are poor by an upper middle income standard, and the few who are not pay twice for a state that delivers once, if at all. A Sri Lanka that defaulted in 2022 outranks India on income. Bangladesh drew level. Net FDI has collapsed to near zero while the AI wave arrives at the bottom rungs of the one job ladder the educated young were promised. The dividend window is finite, and a young country that under equips its young risks growing old before it grows rich.

What it means for you

The macro story has a personal corollary. In an economy that grows in nominal rupees while the currency gently depreciates, cash quietly loses value, and dollar denominated goals such as overseas education or imported healthcare get more expensive over time. The individual answer to a country that compounds slowly is to make sure your own money compounds: long term, diversified investing anchored in equities, which outruns both inflation and the rupee's slide.

Nora AI, the Norafi web app, helps you plan inside exactly this reality. It models how inflation and rupee depreciation bite into long term and dollar linked goals, and shows what to invest today so that a slowly enriching macro backdrop never derails your own trajectory. The nation has until the 2050s to use its window. You have your working life.

App · coming soon

Where sources disagree, notably on logistics cost, poverty measurement and female labour force participation, the text flags the dispute rather than silently picking the flattering number. Nominal figures move with exchange rates and data revisions; treat single year per capita comparisons as indicative, not precise. The full source list sits in this article's metadata.