Methodology
The survey's own words, as the observatory uses them: what each measure counts, how a person's activity is coded, how a sample stands for a country, and how sure any number can be. How we count sets out the rules the observatory keeps; this page defines the terms.
IThe measures
- Labour force participation rateLFPR
- working, or seeking or available for work, as a share of everyone in the groupThe labour force is everyone who works and everyone who wants work and cannot find it.
- Worker-population ratioWPR
- working, as a share of everyone in the groupThe plainest measure of work: of every hundred people, how many work.
- Unemployment rateUR
- seeking or available for work and finding none, as a share of the labour forceWhere most people cannot afford to be without income, a low rate often means people take whatever work there is, not that work is good.
- Usual statusUPSS
- the past year: the activity of most of it (principal status), or work of thirty days or more beside it (subsidiary status)The observatory uses usual status, principal and subsidiary together, unless a page says otherwise.
- Current weekly statusCWS
- the past seven days: at work if an hour or more on any dayCatches short and seasonal work that a year can hide.
- Regular wage or salary
- paid regularly by an employer, not by the day or the taskRegular does not mean secure: many regular workers have no written contract, paid leave or social security.
- Casual labour
- paid by the day or the task, with no promise of the next dayIncludes public works, MGNREGA among them.
- Own-account worker and employer
- running an enterprise or a farm, hiring no one regularly (own account) or hiring workers (employer)
- Unpaid helper
- working in a household's own farm or enterprise without a wage of one's ownCounted as work. Keeping the house, cooking and caring for one's own family are not: the survey records them as domestic duties, outside the labour force.
Unless a page says otherwise, the measures are for persons aged fifteen and over, in rural and urban areas together, on usual status.
IIThe activity codes
Every person in the survey is given one activity status for the year, and another for the week. These are the codes, and the groups the observatory reads them in.
| Code | Activity | Read as |
|---|---|---|
| 11 | own-account worker | self-employed |
| 12 | employer | self-employed |
| 21 | unpaid helper in a household enterprise | self-employed |
| 31 | regular wage or salaried employee | regular wage |
| 41, 42 | casual labour in public works (42: MGNREGA) | casual labour |
| 51 | casual labour in other work | casual labour |
| 81 | sought work | unemployed |
| 82 | did not seek but was available for work | unemployed |
| 91 | attended education | outside the labour force |
| 92, 93 | domestic duties only, or with free collection of goods for the household | outside the labour force |
| 94 to 99 | rentiers, pensioners, those unable to work, and others | outside the labour force |
IIIIndustries and occupations
Industries follow the National Industrial Classification, occupations the National Classification of Occupations. Both changed over the forty years the observatory reads, and a change of classification is not a change in the economy.
Industries. From the 2008 classification, publishing left manufacturing for information and communication, and the repair of computers and household goods moved to other services. The observatory reads every round on the 2008 divisions, with manufacturing as divisions 10 to 33, so the factory does not seem to shrink at the join.
Occupations. The labour survey moved to the 2015 classification of occupations in 2017-18. Its groups do not map one to one onto the older ones, so the observatory does not draw an occupation's line across that change.
IVHow a sample stands for a country
The surveys are drawn in two stages: first villages in rural India and town blocks in urban India (the first-stage units), then households within each. Every person's record carries a multiplier, the number of people like them that the survey stands for. Every rate and share on the observatory is computed with those multipliers, by the survey's own rule for each round: where a round's documentation divides the multiplier, or combines two sub-samples, the observatory does the same.
The weights are made for rates, not for head counts: summed, they fall short of the census's projection of the population. The observatory therefore prints rates and shares, and a "thousand" is always a thousand people in proportion, never a count of millions.
VHow sure a number can be
A sample is not a census, and any estimate could have come out a little differently with another draw of villages and blocks. The observatory computes each estimate's standard error from the survey's own design, its strata and its first-stage units, and prints the 95% interval: the estimate, give or take about two standard errors. Where two intervals overlap, the survey cannot say which is higher, and the atlas does not rank them.
The survey is designed to speak for India and its States. A district holds only a few first-stage units, so the district atlas reads each district by its relative standard error (the error as a share of the rate):
- Reliable up to 25%, drawn as a solid wash.
- Indicative up to 50%, or where the district has few first-stage units: the wash under hatching.
- Unreliable beyond that, where one unit carries most of the weight, where the sample is suppressed, or (2025) where the district is not its own stratum: hatching alone, and no number.
VISeams
Three changes of instrument run through the record. In 2004-05, side work began to count only if it lasted thirty days or more. In 2017-18, the employment surveys of the NSS gave way to the Periodic Labour Force Survey, with a new design. In 2025, that survey was redesigned again. At each seam the observatory stops the line and marks the join in turmeric, the colour of the state: the change is the state's, not the workers'.
VIIHow to cite
Kumar, A., and Pradhan, A. (2026). India Labour Observatory. https://indialabourobservatory.com
Cite a monograph by its title and the release it appeared in, as printed at its foot.