Who has work, who wants it, who is counted.

Ask about the whole year, did you work, more often than not, and 3.2 per cent of the labour force says no. Ask about last week and the answer is roughly double. The vertical distance between these lines is the casualisation of Indian work: hundreds of millions who are employed in principle and idle in practice, week to week.

Both lines fall through the window. The post-pandemic recovery in work is real, on both stopwatches: the question each answers is simply different.

The monthly participation plate in §06 is seventeen months old and shows a snapshot. This annual companion runs seven agriculture years and shows the trend: participation has climbed steeply and steadily, on both stopwatches. Most of the climb is rural women re-entering the count, through self-employment, as the next section shows, not urban job-seekers flooding in.

The usual-status line sits above the weekly one because it catches attachment to work across the year, not just the week surveyed. Both rise in lockstep: the recovery in participation is real on either stopwatch.

In January 2025 MoSPI rebuilt the survey: a new sample, households visited four months running, villages added to the quarterly count, and annual figures on the calendar year instead of July to June. A new design is a new ruler. Its first year cannot be read as a rise or fall from the old one's last, however tempting the two numbers side by side.

The agriculture-year and calendar-year lines are the same old survey over two different twelve-month windows, so they overlap and nearly agree. The blue dot is the new survey's first full year. Where it sits next to the grey lines says how differently the two rulers read the same country, not what happened to jobs.

The quarterly bulletin was the oldest continuous stopwatch: urban India, last-week status, from 2018 to the end of 2024. The spike is the lockdown quarter: one urban worker in five counted unemployed in April–June 2020. What matters after the spike is the slow grind down: six years to retrace a few points.

From April–June 2025 the bulletin runs on the new survey design. Its urban line is drawn beside the old one, not joined to it: the two read the same cities through different samples, and the closeness of the levels is not proof that they agree.

Before 2025 the quarterly count stopped at the city limits. Now it covers the whole country, every quarter. Rural unemployment prints lower because village work is mostly the family farm or the family shop: a person helping on the land counts as working, even in a lean week. The rural line moves with the seasons; read it a year at a time, not quarter to quarter.

The gap between the two dots on a row is how much harder it is to be young and looking for work in that state. Everywhere the young dot sits to the right: a first job is the hardest one to find, and a young person with a family to fall back on can afford to wait for a better offer. The widest gaps are in states with many educated young people and few salaried openings near home.

The national rate is an average of four very different markets. Urban women seek work in a market twice as tight as urban men's; the rural lines bunch lower, where work is seasonal, family-based and rarely "sought" in the survey's sense. The monthly bulletin is young, so read the levels more than the wiggles.

This is the more important chart. Three-quarters of urban men are in the labour force; one urban woman in four is. Rural women participate more, farm and family enterprise absorb labour the city does not, but still at barely half the male rate. No single policy moves this line; childcare, safety, transport and social permission all move it at once.

Participation asks who is in the pool; this asks who is actually in a job. The two lines move together for men, because most men who want work find it. They part sharply for women: many women counted as participating are still searching, so the WPR line sits well below the LFPR line. That vertical gap, read alongside the UR plate, is the unemployment rate made visible.

The same four labour markets as the plates above, for the young alone. Every line sits well above its 15+ cousin: the young are new to the queue, and many are waiting for a particular kind of job. The top line is the one to watch: young women in towns and cities, often educated, looking for work that is safe, near home and salaried, and finding it rarely.

The unemployment rate counts people without work. This plate counts the kinds of work the employed actually have. Self-employment, the farm, the shop, the loom, is the Indian norm; the salaried payroll job is the minority; casual labour is the day-rate rest. The underemployment gap in §03 lives here: own-account work absorbs people a payroll never would, at whatever the week offers.

Rising self-employment is not simply rising enterprise. Part of it is unpaid: helpers in household enterprises are counted as self-employed workers, and earn no separate wage at all.

Split the person-level plate above by gender and the averages dissolve. Men hold steady across the three statuses: roughly half self-employed, a quarter salaried, a fifth casual, barely moving in seven years. Women do not. Female self-employment has climbed fifteen points, almost entirely into unpaid household-enterprise helper roles; female regular-wage work has fallen by five. Women are not joining the salaried minority: they are joining the family firm.

Read the two right-hand lines together: female casual labour has fallen faster than male. The woman leaving casual work is not moving to a salary; she is moving to a family enterprise that pays no separate wage at all.

Each row is a state, sorted by the share of its workers with a regular wage or salary: a payslip every month, whatever its size. The pale dot is casual labour, work hired by the day. Whatever is left of a hundred is self-employment: the farm, the shop, the auto-rickshaw. Small, urban territories sit at the top because their economies are offices and services; the large farming states sit low, where most people work for themselves.

The unemployment rate says who lacks work. This plate says what the work that exists is worth. Four lines, one shape: the salaried man on top; the salaried woman and the self-employed man close enough to trade places; the self-employed woman far beneath all three. The gaps are the structure: they have not closed anywhere in the window.

These are nominal rupees. Some of the climb is pay; some is the price level climbing with it.

This is the price of a day's labour for the fifth of Indian workers hired by the day. It has no leave, no contract and no next week attached. The male and female lines never touch: the day is priced differently depending on whose it is.

The same kind of job, a regular wage or salary, pays very differently depending on where it is. The top of the strip is mostly small territories and the north-east, where government and organised employers are a large share of the few salaried jobs. The big farming states pay least. The pale dot, women's pay, sits left of the dark one in almost every row.

These are nominal rupees, and the cost of living differs by state as well. A rupee buys more rent in Bihar than in Delhi.

A regular job is the best kind of work the survey counts, and still, for many, it comes without a contract, without paid leave and without provident fund, pension or health cover. The blue line is the hard core: employees with none of the three. It is the share of India's "good jobs" that a sick week or a manager's whim can end without notice or pay.

In most rich countries, unemployment falls as schooling rises. In India it rises with every rung. The reason is the denominator's cousin from §05: open unemployment needs a seeker who can wait. A worker who never went to school cannot afford a jobless week and takes whatever the day offers; a graduate, often young, often with a family behind them, queues for the formal post the degree was meant to buy. The ladder measures who can afford to search, and who cannot afford not to work.

Men's participation barely varies by caste or creed: about four men in five work or seek work, whatever their group. Women's varies enormously. Tribal women work most, often on their own land and forest produce, in communities where women's work outside the home carries less stigma. As social rank rises, a household's standing can come to rest on keeping its women at home, and participation falls. The religion rows tell a similar story about social norms, not ability or willingness.

"Others" in the survey means households outside the Scheduled Tribes, Scheduled Castes and Other Backward Classes: largely, though not only, the upper castes.

Each row is a State or Union Territory, sorted by the annual usual-status rate: the dark dot. The pale dot is the same year asked about last week instead; the two stopwatches sit side by side and are never averaged. The vertical rule is All-India. Blue marks the places printing at twice the national rate or more.

The top of the strip is not simply distress, and the bottom is not simply health. Open unemployment needs a seeker who can afford to wait: Kerala's graduates queue for formal work and are counted; Gujarat's farms and workshops absorb almost everyone at some wage, and the queue never forms. A high rate can mean a labour market worth waiting for. A low one can mean an economy too poor to permit the wait. The strip ranks the searching, not the suffering.

This is the answer to "how many people in my state work". Read along a row: the pale dot is the share of the state's men aged 15 and over who are working, the dark dot the share of its women. The men's dots form a tight column. The women's dots scatter across the chart. States where women farm their own land, trade and gather forest produce sit near the top; states where women's work outside the home is rarer, and the city-territories, sit near the bottom.