MoSPI’s summary of the second all-India time use survey, with the published averages for paid work, unpaid domestic work and care that our tables reproduce. Start here for what the survey covers and how a day was recorded.
Deep Dive · Work & Labour
Who is still at work at five o’clock
On 27 October 2026 the Commerce Minister told Dutch investors that no one in India will tell them it is five o’clock, or remind them it is the weekend. India’s own time use survey records what Indians did in every half hour of a day. This is what it shows, read against the law, against other countries and against the work nobody pays for.
India
Work & labour
Time use
7 readings
IM
ImpactMojo Editorial
Analysis by the ImpactMojo team
Every Indian figure below is computed from the unit-level records of the NSO Time Use Survey 2024. Before any chart was drawn, our tables reproduced all twelve of MoSPI’s published averages for paid, domestic and care work exactly. The method is set out at the foot of the page.
House Pick
What was said
Speaking at the India-Netherlands Business Forum in The Hague on 27 October 2026, Piyush Goyal offered India’s workers as a reason to invest: “Indians want to work hard. Nobody in India will tell you it’s five o’clock. And certainly no Indian will ever remind you it’s the weekend.” The remark drew criticism from opposition leaders and from the Karnataka State IT/ITeS Employees Union, and a defence from industrialists who called it an edge.
Both halves of the sentence can be tested. The NSO’s Time Use Survey of 2024 asked everyone aged six and over in a sample of households to account for the previous day, from four in the morning to four the next morning, in half-hour slots. We take the 198,969 diaries of people aged 15 and over whose usual principal activity is work, weight them as MoSPI does, and ask three questions. Who is still working at five. Who works on Sunday. And what an hour of that work produces.
One fact frames everything below. Salaried employees, the people the minister was describing to investors, are 24% of India’s workers. The other 76% work for themselves, in a family enterprise, or as casual labour paid by the day.
Section 01
Five o’clock
The share of workers doing paid work in each half hour of a working day, among those who did any paid work that day. The dashed line is 5 pm.
Salaried, privateSalaried, governmentSelf-employedCasual labour
At five in the afternoon, 72% of men in salaried private jobs are still at work, and half of them have not finished by 6:30 pm. At eight in the evening 16% are still working. For women in private jobs the figures are 57% at five, with a median finish of 6 pm.
The workers who most often stop at five are the government’s own. In government and public-sector employment 51% of men and 24% of women are still at work at five, and the median woman in a government job has finished by 4 pm.
Self-employed and casual workers sit between the two. Casual labour ends early: by seven in the evening only 8% of casual men are still at it. Among the self-employed, 28% of men work past seven.
Section 02
The weekend
The share of workers who did any paid work on the diary day, by day of the week. Each day of the week was sampled separately, so each point is a different set of people.
MenWomen
Salaried work has a weekend. On a Wednesday 95% of salaried men do paid work; on a Sunday 61% do. For salaried women the Sunday figure falls to 43%.
The workers who do not stop on Sunday are the ones with no paid leave. 84% of self-employed men and 83% of casual men work on a Sunday. A casual labourer is paid by the day, and a day not worked is a day not paid. These two groups are 76% of the workforce, so when the minister says no Indian will remind an investor that it is the weekend, he is describing a workforce most of which has no paid weekend to claim.
Section 03
The hours, and the law
Paid hours in an average week, estimated from the diaries of all seven days.
MenWomen
The Occupational Safety, Health and Working Conditions Code, 2020, in force since 21 November 2025, limits a worker to eight hours a day, and the rules framed under it set the working week at 48 hours. Salaried men already average 48.7 paid hours a week, above that limit before a minute of commuting is added. Across all workers the average is 40.1 hours: 43.4 for men and 30.2 for women.
The ILO, using the Periodic Labour Force Survey, puts the average Indian worker at 45.5 hours a week in 2025. The two measures differ in method. The labour force survey asks people how many hours they worked in the past week; the time use survey records one day in detail and spreads its diaries across all seven. Of the 18 countries in the next chart, only Pakistan (46.8 hours) and Bangladesh (46.7 hours) report a longer average week.
Section 04
What an hour produces
Average weekly hours worked against output per hour worked, for 18 countries, latest year for each.
The minister was speaking to Dutch investors. A worker in the Netherlands averaged 30.2 hours a week in 2024 and produced $91.2 of output per hour. A worker in India averaged 45.5 hours in 2025 and produced $10.8. An hour of Dutch work yields about 8.4 times as much. The Dutch week is short partly because the Netherlands has the largest share of part-time workers in the European Union, so the comparison is of an average worker, and an hour is the fairer unit.
The countries that work the longest weeks are the ones that produce least per hour: Pakistan (46.8 hours, $7.2), Bangladesh (46.7 hours, $8.6) and India. In this chart long weeks go with low output per hour. Output per hour depends on capital, skills, infrastructure and how work is organised, and none of those rises because people stay at work later. The research on fatigue points the other way: past a threshold, each extra hour adds less output than the one before, and very long weeks carry a measured cost in heart disease and stroke.
Section 05
The shift nobody counts
Hours of paid work and of unpaid housework and care for household members, on a working day, for workers who did paid work that day.
Paid workUnpaid housework and care
The minister’s sentence counts only paid hours. Women who go out to work come home to a second job. A woman in a government post does 5.7 hours of paid work and 3.5 of unpaid work on a working day, 9.2 in all, against 7.7 for a man in the same job. A woman in a private job works 9.4 hours in all, against 8.5 for a man. Casual women work 9.5 hours against 7.7 for men. In every kind of employment the woman’s day is the longer one.
Section 06
Readings
The data behind this page, the law it is read against, and the research on what long hours do to output and to health.
The records this page is computed from: 10.2 million half-hour activity entries from 139,489 households. Free to download after registration. The layout file and the note for data users explain the weights and the codes.
PRS’s summary of the Code that now governs working hours. It records that the 2020 Bill fixed the maximum working day at eight hours. PRS’s note on the draft rules under the Code sets out the 48-hour week and the twelve-hour limit on the spread of a working day.
The ILO’s global review of how long people work, how working time is regulated, and what long and irregular hours do to workers and firms. Read it for where India sits among countries on the length of the working week.
A study of munition workers, most of them women, whose output and hours were recorded in detail. Below a threshold output rises in proportion to hours; above it, output rises at a decreasing rate. The paper links the finding to research on accidents and injuries in long working weeks.
The WHO and ILO joint estimates. In 2016, 488 million people worldwide worked 55 hours a week or more, and an estimated 745,194 deaths from ischaemic heart disease and stroke were attributable to those hours.
Essays on how time use surveys measure the unpaid work that national accounts leave out, and how that work shapes women’s paid employment. The background to Section 05.
How these figures were made
Data. NSO Time Use Survey 2024, Schedule 10.6, unit-level person records (MoSPI Microdata Catalogue 236), read from the released Nesstar file. Estimates use MoSPI’s weight, MULT divided by 100. Where a half-hour slot holds more than one activity, its time is divided equally among them, as in MoSPI’s own tables.
Check against the published figures. Our code reproduces, to the minute, all twelve published averages of minutes per participant per day for employment, unpaid domestic services and unpaid caregiving, for men and women aged 6 and over and 15 to 59 (eSankhyiki, TUS indicator 7).
Who counts as a worker. A person aged 15 or over whose usual principal activity status is self-employment (codes 11, 12, 21), regular salaried work (31) or casual labour (41, 51). Paid work is ICATUS 2016 divisions 11 to 15; commuting (division 18), seeking work and setting up a business are left out. Unpaid work is major divisions 3 and 4.
Government or private. The survey records the employer only for paid work done on the diary day, so a salaried worker can be placed in government or private employment only on a day they worked. That split is used only in Sections 01 and 05, which are limited to workers who did paid work that day.
Limits. Each person kept a diary for one day, so weekly hours are an average across people, not a record of anyone’s week. Two stratum variables in the released file could not be read with the software available to us; they affect standard errors, not these estimates. The aggregates are in data/deep-dives/working-hours.json, and the code that makes them is in scripts/working-hours/ in the ImpactMojo repository. The minister’s words are as reported by BW People and Infomance on 28 October 2026.