A family in a district town spends ten years getting somewhere. The father moves from daily wage work to a shop job with a fixed monthly salary. The mother joins a self-help group and starts a small tailoring business. They stop borrowing for food. They put both children in a private school, because the government one down the road has one teacher for five classes. By the tenth year they are not rich. They are, for the first time, not poor.
In year eleven the father is diagnosed with something that needs surgery.
Eighteen months later the shop job is gone, the tailoring machine is sold, the elder child has left school to work, and the family is borrowing for food again. Ten years of climbing, undone by a single event that nobody in the family did anything to cause.
This is the most important thing about poverty in India, and it is almost never the story we tell. We measure how many people crossed a line. We rarely measure how many crossed it and then crossed back, or how many are standing just above it with nothing underneath them.
The climbing is real. So is the falling. This piece is about the second one.
The good news is real, and it is not the whole picture
India has moved an extraordinary number of people out of absolute poverty in a generation. That is true, it is measurable, and nothing here disputes it.
But “moved out of poverty” almost always means “crossed a threshold at the moment we measured.” It does not mean the household is secure. A family one rupee above a poverty line and a family ten thousand above it are counted the same way, and only one of them survives a bad year.
The question worth asking is not how many climbed. It is how many can stay.
And the answer, in India, depends less on income than on how many uninsured risks a household is carrying. Most families that fall do not fall because they stopped earning. They fall because something arrived that they had to pay for in one go, and the only way to pay for it in one go was to sell the things that were generating the income.
There are four of these events that recur in Indian household life. None of them is exotic. All of them are structural, which is to say they are choices somebody made about how the system works.
One: the hospital bill
Health is the most common single cause of a household falling back, and the numbers explain why.
We looked at this recently in detail. India’s out-of-pocket share of health spending fell from 64.2 per cent to 43.4 per cent over roughly a decade, according to National Health Accounts data. That is a genuine improvement and it deserves to be said plainly.
It also means that after all that progress, more than four rupees in every ten spent on health in India still comes directly out of a family’s pocket at the moment they are least able to pay. Not from insurance, not from the state, from them. And the lowest point in that series was 2021-22, which was distorted by one-time COVID spending rather than a permanent shift.
Here is what makes a medical event different from other expenses. It is large, it is sudden, it is not optional, and it arrives at the exact moment the household’s earning capacity drops, because the person who is ill is usually a person who works. You lose the income and take on the cost in the same week.
Families meet it the only way they can. They borrow from a moneylender at rates that are not survivable. They sell the productive asset, the auto, the machine, the livestock, the gold that was collateral for everything else. They take the child out of school, which converts a short-term cash problem into a permanent reduction in what that child will earn for forty years.
Every one of those responses solves the immediate crisis and removes a rung from the ladder.
Two: the land that is yours until somebody says it is not
The second trap is slower and it is the one outsiders underestimate.
Indian land titles are presumptive rather than guaranteed. What a family holds is not proof of ownership, it is evidence of it, and evidence can be contested. The consequence shows up in the courts: roughly two-thirds of civil disputes in India are about land, on the estimate PRS has documented.
For a household, this matters in two ways.
The obvious one is that a disputed plot can be lost. The less obvious one is that even an undisputed plot is a weaker asset than it looks. Land you cannot prove you own is land you cannot reliably borrow against, sell quickly, or use as security. For most rural families the plot is the single largest thing they have, and its usefulness is throttled by a records system that never finished being modernised.
Then there is time. Land disputes in India run for years, sometimes across generations. The cost is not only legal fees. It is the decade in which the family cannot invest in the asset because ownership is uncertain, and the perfectly rational decision not to improve something that might be taken away.
A family can be asset-rich on paper and one adverse mutation entry away from having nothing.
Three: the education bet, paid up front
The third is the one families choose, which makes it the hardest to write about.
Education is the main mechanism Indian households use to change their position, and they are not wrong to believe in it. But the way the system is built, the family pays first and finds out later.
The paying starts before the exam. Coaching is the largest hidden education expense most families carry, and it exists because there is a gap between what a school syllabus teaches and what a competitive rank list demands. The exam is formally free to sit. Competing in it is not.
So a family commits years of savings to a child’s preparation. And then the risk sits entirely with them.
The exam can be cancelled. India cancelled its national medical entrance exam in May 2026 after a circulated paper matched the real one, interrupting about 22 lakh lives in a single decision. Not one of those families caused the leak. All of them paid for the extra year.
The paper can leak. Across fifteen states, 41 recruitment exam leak cases affected roughly 1.4 crore applicants competing for just over one lakh posts.
And even when everything works, the wait can consume the gain. The years that pass between a recruitment notification and an actual joining letter are years in which a young person is not earning, is often not employable elsewhere because they are studying, and is being supported by a household that has already spent its savings on the attempt.
The clearest picture of where that leads is the PhD holder applying for a peon’s job. That is not a curiosity. It is a family’s entire education investment arriving at a return it was never told to expect.
Four: the move that costs you your entitlements
The fourth is migration, and it is close to universal. Nearly three in ten Indians are migrants by official measures, though the last full count came from a pandemic-year survey, which tells you something about how closely we watch the largest workforce movement in the country.
Moving for work is usually the right decision. It is often the single highest-return thing a household can do. The problem is what the move costs on arrival.
A family’s entitlements are largely tied to where they are registered rather than where they are. Ration access, school admission, health scheme enrolment, the documentation that proves who you are to a local official. Cross a state line and much of that thins out or stops. The household earns more and is simultaneously less protected, which is a strange trade to be forced into.
Housing in the destination city is informal, insecure and expensive relative to what it provides. And when a shock arrives, the migrant household is the one furthest from the support systems that were supposed to catch it, which is precisely what the country watched happen on its highways in 2020.
The pattern underneath all four
Put those four side by side and the common structure is obvious.
In each case, a risk that could be pooled across millions of people has instead been left sitting on one household. A medical event. A title dispute. A cancelled exam. A move for work. Every one of these is a normal, foreseeable thing that happens in a large economy. None of them is a personal failing. And in each case the family absorbs the full cost alone.
That is what an insurance failure looks like in practice. Not an absent policy document, but a household selling a rickshaw to pay for surgery.
The other thing they share is timing. Each shock lands hardest on households that have just started to accumulate. A family with nothing has nothing to sell, and a family with plenty absorbs the hit. The one in the middle, the one that has spent ten years building exactly one productive asset, is the one for whom a single event is catastrophic. Our policy attention is mostly on the poorest, understandably. The falling happens one rung up.
The moneylender is the real safety net
There is a piece of machinery sitting underneath all four shocks that deserves naming, because it is where the falling actually happens.
When a family needs a large sum immediately, the formal system is mostly unavailable to them. A bank loan needs documentation, collateral that can be proven, a credit history, and time. The family has none of the first three and cannot afford the fourth, because the surgery is scheduled for Thursday.
So they go to the local lender, who requires no paperwork, asks no questions, and can hand over cash the same day. That speed is a real service and it is why the arrangement persists. The price of it is an interest rate that no household income can outrun.
What follows is predictable. The loan is taken to cover one event. The repayments then consume the surplus that was funding school fees, or stock for the shop, or the savings that would have covered the next emergency. When the next emergency arrives, and it will, there is no surplus left, so it is met with another loan. The family is now servicing debt rather than accumulating, and the ladder has quietly become a treadmill.
This is why the fall is so much faster than the climb. Ten years of accumulation happens at the speed of saving. The unwinding happens at the speed of compound interest on an informal loan.
Any serious attempt to make gains stick has to answer one question: where does a household get a large sum quickly, at a survivable price, without documentation it does not have? Until there is an answer, the moneylender is the safety net, and he is charging for it.
What families already do, and what it costs them
Indian households are not passive about any of this. They protect themselves, rationally, and the protection itself is expensive.
They hold gold, because it is the only asset that is liquid, portable, socially acceptable to own, and does not require a title deed. Economists sometimes describe this as unproductive saving. It is nothing of the sort. It is a family buying the only insurance product actually available to them.
They keep children geographically close, which limits how far a young person can go for education or work.
They avoid specialising. A household that could earn more by committing fully to one trade often keeps three income sources instead, because diversification survives a bad year and specialisation does not. The result is lower average income, deliberately chosen, in exchange for lower variance.
They under-invest in the productive asset, because an asset that might have to be sold in an emergency is not worth improving.
And they save for emergencies rather than for opportunities, which means the capital that could have started a business sits idle against a hospital bill that may never come.
Add those together and you get something worth sitting with. A large share of the gap between what Indian households earn and what they could earn is not a skills problem or an ambition problem. It is the price they are paying, every day, to self-insure against risks that a functioning system would have pooled.
Where crowding comes into it
India is a crowded country. That is not an insult, it is arithmetic, and it changes the mechanics of everything above.
When a good is scarce and the queue for it is enormous, it has to be rationed somehow. There are only three ways to do it: by price, by queue, or by chance. India uses all three, and each one is a place where households lose their gains.
By price. Government provision exists but is thin, so those who can pay exit to the private version. Private school because the state school has one teacher. Private hospital because the district one has no specialist. Coaching because the syllabus is not enough. Every exit is a household converting income into a service the state was meant to provide, which is money that cannot become savings.
By queue. Where you cannot pay, you wait. Court cases run past five crore pending. Justice posts sit vacant at around one in four. Recruitment runs for years. The queue is not free; it is paid in the years of a life.
By chance. Where price and queue both fail, outcomes turn on luck. Which exam session, which official, which district. A system this crowded produces enormous variance in what two identical families receive.
Density does not cause any of this by itself. Plenty of dense places run good public systems. What density does is make every gap in a system expensive at scale. A shortfall that would inconvenience a small country becomes, across 1.4 billion people, a permanent second market that families pay for out of the income that was supposed to lift them.
This is also why national totals mislead. Record numbers can rise while the individual experience barely shifts, because the denominator is doing most of the work. A country can be growing and a household can be standing still, and both statements are true at once.
What would actually make gains stick
The useful question is not how to make people climb faster. They are climbing about as fast as anyone could. It is how to stop the falling.
Health cover that pays at the counter. The measure that matters is not how many people are enrolled in a scheme. It is what share of a hospitalisation a family actually pays on the day. Out-of-pocket share is the honest number, and it is still above four in ten.
Land titles that are guaranteed rather than presumptive. This is a decades-long administrative project with no political glamour attached, and it would do more for rural household security than most schemes that get announced. Ownership you can prove is an asset. Ownership you can only argue is a liability with a deed.
Exam calendars that hold. A published date, a result date, and a joining date that are treated as commitments rather than aspirations. The families are not asking to be given the job. They are asking to know when they will find out, so the years are spent on a decision rather than on waiting.
Entitlements that travel. If three in ten citizens move, benefits tied to a fixed address are designed for a country that no longer exists. Portability is not a favour to migrants. It is a correction to a design error.
Emergency credit at a survivable price. Somewhere between a bank that cannot move in three days and a lender charging rates nobody can outrun, there is a missing product. Whoever builds it, public or private, removes the mechanism that turns a bad month into a lost decade.
And measurement that admits the snake exists. We count how many crossed the line. We should also count how many crossed back, and how far above the line a household has to be before a hospitalisation stops being catastrophic. That second number is the real one, and we mostly do not publish it.
There is a version of this list that sounds expensive. It is worth comparing it against what is already being spent. Families are currently paying for all of these things privately and badly: private health cover in the form of gold and debt, private title insurance in the form of not improving the land, private exam insurance in the form of a second and third attempt, private social security in the form of keeping three income streams instead of one. The money is already leaving households. It is simply leaving in the least efficient way available.
What this is, in the end
Poverty in India is often discussed as a starting position. For a very large number of families it is closer to a recurring risk. They leave it, they build something, and then a normal event that any system could have absorbed collectively arrives at their door alone, and they go back.
The families are not failing. Ten years of climbing is not a failure. What fails is the absence of anything underneath them while they climb.
Every one of the four shocks in this piece is fixable. None of them requires a discovery. They require an insurance system that pays at the counter, a records system that proves ownership, an exam system that keeps its calendar, and entitlements that follow the person instead of the address.
Until then we will keep publishing the number of people who crossed the line, and quietly not publishing the number who crossed back.