Economy
Are people finding work and earning more?
A little more work, formal hiring slower, and households unconvinced. Unemployment was 5.0% in August 2026 by the monthly labour survey, but new formal registrations with ESIC are 11.9% lower than a year ago, and MGNREGA provided 228 crore days of rural work in 2025-26.
Written for anyone who wants to know whether India's growth is turning into jobs and incomes. Survey to August 2026.
A little more work, formal hiring slower, and households unconvinced.
India's labour survey finds 5.0% of the labour force unemployed in August 2026, 6.8% in towns and cities and 4.1% in villages. The monthly figure has moved between 4.7% and 5.6% in the 17 months since MoSPI began publishing it in April 2025. 52.8% of people aged 15 and over were working, against 52.2% a year earlier. Among the young (15 to 29) unemployment is 15.1%; among women, 34.8% are in the labour force at all, against 77.3% of men.
The longer view comes from the cities, where the quarterly survey starts in 2018. Urban unemployment fell from 9.7% in Q1 FY19 to 6.7% in Q1 FY27, below the 9.0% it averaged before the pandemic. Over the same years the share of urban women in the labour force rose from 18.8% to 25.0%. MoSPI redesigned the sample in January 2025, so the last few quarters are not strictly comparable with the rest.
Formal payrolls tell a slower story. New workers registered with ESIC, the insurance scheme for lower-paid formal jobs, numbered 141.4 lakh in the twelve months to July 2026, 11.9% fewer than a year before. EPFO's net new subscribers were up 0.7% on the same basis in July 2025, the last month EPFO has published: its monthly releases stopped after that. NACH bulk credits, the channel for salaries, pensions, dividends and interest, grew 16.3% over twelve months, but a test we wrote down first found they do not track hiring (correlation with EPFO's figure −0.36).
In villages, the job guarantee is doing less. MGNREGA provided 228 crore days of work in 2025-26, against 294 crore in 2022-23 and 363 crore in 2021-22, a year swollen by the pandemic; households that worked averaged 43 days of the 100 guaranteed. Its daily wage, ₹267, rose 3.8% after rural inflation that year and 6.0% over the four years from 2021-22. Since 1 July 2026 its successor, VB-G RAM G, has its own dashboard: 41.7 crore days of work so far in 2026-27, which we keep separate rather than add to the old scheme's figures.
Households themselves are unconvinced. In RBI's September 2026 survey, the share of urban households saying jobs were worse than a year ago exceeded the share saying better by 21.6 percentage points, and in villages by 10.5 points, the gloomiest rural reading since the series began in September 2023. A pre-registered test found these feelings do not move with the measured unemployment rate (rank correlation of changes 0.10, p 0.75).
Scorecard
Eight readings, each against its own history. The monthly labour survey began in April 2025, so its rows are new series with no usual range yet; the payroll growth rates leave out April 2020 to March 2022.
usual range: the middle half of all readings latest 0: as many better as worse Each line runs from the lowest reading to the highest.
This month
5.0%unemployment, August 2026
Against 5.1% in July 2026. Urban 6.8%, rural 4.1%. The monthly series is too young for a usual range.
This cycle
−11.9%new ESIC workers, 12 months to July 2026, on a year earlier
EPFO's last twelve months on record, to July 2025, added 154.1 lakh net subscribers, 0.7% more than the year before.
This decade
6.7%urban unemployment, Q1 FY27
Down from 9.7% in Q1 FY19, the first quarter of the survey, and 9.0% on average before the pandemic.
The monthly survey
MoSPI's monthly estimates use current weekly status: anyone who worked for at least an hour in the week before the interview counts as employed. On that measure unemployment was 5.0% in August 2026, higher in towns and cities than in villages, where most people work on farms or for themselves.
Unemployment rate, monthly
Current weekly status, persons 15 and over, from April 2025.
How to read it. A low rate in villages does not mean plenty of good work: people who cannot afford to be idle take whatever work there is, often on the family farm.
Loading chart…
Cities since 2018
Only the urban quarterly series reaches back before 2025. Unemployment there peaked at 20.8% in Q1 FY21, in the first lockdown, and is 6.7% now. Youth unemployment in the cities was 18.2% in Q1 FY27, against 22.9% in Q1 FY19.
Urban unemployment rate, quarterly
Current weekly status, from April-June 2018. MoSPI redesigned the sample from January 2025.
How to read it. Gaps are quarters whose published rates do not add up and are left blank. Compare quarters before and after the redesign with care.
Loading chart…
Formal payrolls
EPFO and ESIC count workers joining their schemes, which only formal employers must register. They count registrations, not new jobs: a worker moving from an informal job to a formal one, or a firm newly registering its staff, counts too. ESIC's twelve-month total was 11.9% lower than a year before in July 2026.
Formal payroll additions, twelve-month totals against a year earlier
EPFO's line ends in July 2025, the last month it published. ESIC is missing February and March 2024 at source.
How to read it. Twelve-month totals smooth EPFO's large revisions and seasonal swings; read the direction.
Loading chart…
Rural work
MGNREGA guarantees 100 days of paid manual work a year to every rural household that asks. The days it provided fell 22% from 2022-23 to 2025-26. Fewer days can mean less need for the work or less money to pay for it; these figures cannot tell which.
Days of work provided under MGNREGA
Crore persondays by financial year (April to March), from the scheme's dashboard as on 2026-10-25. Complete years only.
How to read it. 2021-22 was still swollen by workers who had returned to villages in the pandemic.
Loading chart…
What households say
RBI asks households whether jobs are better or worse than a year ago. In cities the answer was net negative in September 2026, lower than in 87% of rounds since 2012, even though measured urban unemployment is lower than in 2018. Our pre-registered test found the two do not move together.
Households' view of jobs
RBI's consumer confidence surveys: % saying employment is better less % saying worse, each round.
How to read it. Below zero, more say worse than better. People judge jobs by their own and their neighbours' work and pay, which the unemployment rate does not capture.
Loading chart…
What the numbers can't tell you
- Low unemployment is not plenty of good work
- Someone who worked a single hour in the survey week, helped without pay on the family farm, or sells from a cart counts as employed. Most Indian workers are self-employed or casual, and many work fewer hours than they want. The unemployment rate counts only those with no work at all who are looking, so it misses underemployment and low pay, which is where most distress lies. More.
- Most women are outside the count
- Fewer than half of women of working age are in the labour force on any of these measures. Unpaid work at home is not counted as work, and a woman who would take a job but is not looking is not counted as unemployed. Rising participation can mean more opportunity or more households needing a second income. More.
- A young series, a new design and no error bars
- Monthly figures begin in April 2025, on a sample redesigned in January 2025, and the quarterly urban series changes design at the same point. MoSPI publishes no standard errors, so a move of a few tenths of a point may be sampling noise. The survey's usual-status measure, which looks at the past year, is annual and not shown here.
- Formal payrolls count registrations, and EPFO has stopped
- EPFO and ESIC count workers newly enrolled, including jobs that existed before and were formalised, and revise heavily. EPFO's monthly releases stopped after July 2025; nothing has replaced them. More.
- Salary credits are not jobs
- NACH bulk credits carry salaries but also pensions, dividends and interest. A test we wrote down first found they do not track formal hiring, so they are shown as a flow of income, never as a jobs count. More.
- Wages are the missing piece
- No monthly wage series is held. The MGNREGA wage, deflated by rural inflation, is the one real wage shown here, and it covers a guaranteed minimum, not the market. Labour Bureau rural wages and PLFS earnings are on the list to add.
What changed
Labour survey for August 2026 against July 2026; payrolls and the RBI survey against their previous release.
| Measure | Previous | Now | Change |
|---|---|---|---|
| Unemployment rate | 5.1% | 5.0% | down 0.1 points |
| In the labour force | 55.4% | 55.6% | up 0.2 points |
| Women in the labour force | 34.4% | 34.8% | up 0.4 points |
| Youth unemployment | 15.6% | 15.1% | down 0.5 points |
| Rural unemployment | 4.5% | 4.1% | down 0.4 points |
| Urban unemployment | 6.7% | 6.8% | up 0.1 points |
| New ESIC workers, 12 months (July 2026) | −12.4% | −11.9% | up 0.5 points |
| Urban households on jobs (September 2026) | −21.9 | −21.6 | up 0.3 points |
The workings
The question
Data
Every chart has a table view, a CSV download and a link to its published JSON. Bundles are listed in data status.