Imagine a town that wants to understand the working lives of its adult residents. It chooses one week and observes the same population under one clear rule.
Some residents worked during the week. Some did not work, but they were trying to find work or were available to take it under the rule being used. The rest were neither working nor meeting that search-or-availability test. Many in this third group were studying, caring for family members, retired, ill, living on pensions or remittances, or unable to take available work. Some may have wanted work but failed the particular test because they had stopped searching or were temporarily unavailable.
The town must classify these people before it calculates a rate. Those who worked form one group. Those without work who met the search-or-availability test form a second group. Everyone else remains in a third group for that period.
The first two groups together show how many people were participating in the labour market. The third group was outside that participation measure. A resident can move from one group to another, so these are not permanent labels attached to human beings. They describe a person's activity under a stated rule and period.
This simple classification explains why “people without jobs” and “unemployed people” are not the same number. It also explains why one unemployment rate can never tell the whole employment story. We must know who was counted, over which period, under which rule and with which denominator.
First decide what counts as employment
Everyday language uses the word work very widely. Cooking for a family, caring for a child and helping a neighbour all require effort and create value. Labour statistics nevertheless need an operational boundary. Without one, different interviewers could classify the same activity differently and their totals could not be compared.
A broad statistical idea of work can include activity such as producing for one's own use, unpaid trainee work and volunteering. Employment is a narrower category connected to production under the rule used by the labour-force measure. A socially valuable activity can therefore be work in the broad sense without placing the person inside measured employment.
Under the Calendar Year 2025 PLFS methodology, India's labour-force measurement treats production for the market as economic activity. It also includes selected production for a household's own use, such as primary goods and the construction of certain fixed assets. A person who helps without regular wages in the household's farm or non-farm enterprise can therefore be counted as employed. A worker who keeps an attachment to a job or enterprise may also remain employed during a temporary absence.
Unpaid domestic services and care produced only for one's own household generally fall outside this employment boundary. That exclusion does not mean that such activity is idle, easy or socially unimportant. It means only that the labour-force statistic has drawn a particular production boundary.
Payment alone therefore cannot classify a person. An unpaid helper in a household shop may be employed, while a person doing substantial unpaid care inside the home may remain outside the labour force. The activity, production boundary and reference-period rule must all be known.
A person is also different from a job. A job is a set of tasks performed for an economic unit. One person may teach at a school and run a small evening business, holding two jobs while still counting as one employed person. Vacancies, payroll entries, jobs and employed people are consequently different statistical units.
The three groups create the labour force
For one fixed population and period, a survey places each person in one of three broad groups. An employed person performed qualifying economic activity or retained the required work attachment. An unemployed person had no employment and met the applicable search-or-availability condition. A person outside the labour force met neither classification.
An employed person is also called a worker. The workforce commonly means the employed group under the stated frame. The labour force is wider because it includes both that workforce and the unemployed.
The labour force contains the employed and the unemployed. If `E` means employed people and `U` means unemployed people, the labour force `L` is `E + U`. People outside the labour force do not enter this total.
The population used for comparison must be named. A total-population rate includes children unless it says otherwise. A working-age rate uses a specified age range or lower age limit. A survey may also have its own eligible population. “Working age” is an analytical boundary, not a claim that everyone inside it can or should work. Comprehensive adult measurement need not use one universal upper age limit. In the Calendar Year 2025 reporting frame, results for all ages and for people aged 15 years and above have different denominators and cannot be exchanged silently.
This gives us two denominators. One is the stated population. The other is the labour force inside that population. Most confusion among the main employment rates comes from switching between them.
Three rates ask three different questions
Return to the town and make the example precise. Picture exactly 100 imaginary adults. Place 54 in employment, 6 in unemployment under the selected rule and the remaining 40 outside participation. Adding the first two groups gives a labour force of 60. The full 100 forms the denominator for participation and worker shares, whereas those 60 participants form the unemployment denominator.
The labour-force participation rate, or LFPR, asks: what share of the stated population is in the labour force? Its numerator is employed plus unemployed people. Its denominator is the stated population. Thus `LFPR = labour force ÷ stated population × 100`.
The worker population ratio, or WPR, asks: what share of that population is employed? Its numerator is employed people, while its denominator remains the stated population. Thus `WPR = employed people ÷ stated population × 100`.
The unemployment rate, or UR, asks: what share of labour-force participants is unemployed? Its numerator is unemployed people, but its denominator is the labour force, not the whole population. Thus `UR = unemployed people ÷ labour force × 100`.
A fourth ratio can divide unemployed people by the stated population. We can call this the unemployed share of population, or PU. It answers a valid question, but it is not the unemployment rate. The difference matters whenever participation changes.
The LFPR is 60 divided by 100, or 60 per cent. The WPR is 54 divided by 100, or 54 per cent. The UR is 6 divided by 60, or 10 per cent. Unemployed people are only 6 per cent of the whole imaginary population, which shows why the population share and UR are different.
These figures also check one another. When rates use exactly the same population, period and status frame, WPR equals LFPR multiplied by one minus UR, with the rates written as decimals. Here, `0.60 × (1 − 0.10)` equals `0.54`. The unemployed share of population similarly equals LFPR multiplied by UR: `0.60 × 0.10` equals `0.06`.
A lower unemployment rate can arise in different ways
Now suppose three of the six unemployed residents find work. Employment rises from 54 to 57, unemployment falls from 6 to 3, and participation still totals 60. LFPR is unchanged at 60 per cent. WPR rises to 57 per cent and UR falls to 5 per cent. The fall in UR accompanies a clear employment gain.
Instead, suppose the three residents stop searching and are not available under the chosen rule. They move from unemployment to outside the labour force. Employment remains 54, unemployment becomes 3 and the labour force shrinks to 57. LFPR falls to 57 per cent, WPR stays at 54 per cent and UR falls to about 5.3 per cent.
Both paths lower UR, but only the first raises employment. A fall in UR can therefore accompany improvement, or it can accompany withdrawal from the labour force. Reading UR with LFPR and WPR reveals the difference.
Even one person's movement changes the measures. If one of the original six unemployed residents finds work, employment becomes 55 and unemployment becomes 5. Participation stays at 60 people, leaving LFPR unchanged at 60 per cent. WPR becomes 55 per cent and UR becomes about 8.3 per cent. The same accounting works in a population of any size.
The reverse movement matters too. Suppose three people outside the labour force begin looking and become available, but have not yet found work. Unemployment rises to 9 and the outside group falls to 37. The labour force rises from 60 to 63 while employment remains 54. LFPR rises to 63 per cent, WPR stays at 54 per cent and UR rises from 10 per cent to about 14.3 per cent. The higher UR reflects new participation in job search, not a loss of those 54 jobs.
The three-person movements make the contrasting paths easy to see, while the one-person case shows that the logic is not tied to a large jump. The central lesson is causal: identify the category movement before judging the rate.
A headcount is not a history of job flows
The three groups form a stock of people in each status for the chosen frame. They do not directly count every event that moved people between the groups.
A person may be hired, separated from a job, leave the labour force, enter it again and find another job. Hires and separations are flows of events. Entry to and exit from the labour force are category movements. Duration records how long a spell continues. A survey headcount at one time or over one frame cannot by itself reconstruct all these histories.
The distinction also explains why a payroll addition is not automatically one more employed person in the whole economy. One worker may change employers and create a payroll entry without becoming newly employed. Another may hold two covered jobs. Many workers may remain outside the particular payroll system altogether.
Employment includes several relationships
Employed people do not all work under the same arrangement. A person may work independently, hire others, help in a household enterprise, receive a regular salary or work under a casual contract. These categories describe work relationships; they do not rank people's effort or worth.
An own-account worker runs an enterprise independently or with partners and generally does not hire a regular worker. An employer also runs an enterprise but generally hires labour. A helper in a household enterprise contributes to that enterprise without receiving a regular wage or salary. All three fall within the broad self-employed group used by India's household survey.
A regular wage or salaried employee works for another economic unit and receives wages or salary on a regular basis rather than through daily or periodic renewal. A casual worker also works for another unit, but the work and payment follow a daily or periodic arrangement. Regularity does not by itself prove adequate earnings or security, and casual status does not reveal every feature of the job.
Self-employment is especially varied. It can describe a successful professional, a small producer building an enterprise, a household helper with little control, or a worker creating an activity because suitable wage work is unavailable. The label alone does not prove entrepreneurship, distress or disguised unemployment. Those interpretations need more evidence.
The reference period changes what the survey can see
Return to a resident who farms during one season, takes occasional construction work and has no work in several weeks. A year-long view may classify this person as employed because work occupied the major part of the relevant time. A particular week may classify the same person as unemployed if no work occurred and the search-or-availability rule was met. A day-by-day record may reveal both workdays and days without work.
None of these classifications is necessarily wrong. They answer different time questions. A long frame shows a person's broad position across a year. A short frame is more sensitive to intermittent work and sudden change. Daily information shows variation that a single status for the week can hide.
India's Periodic Labour Force Survey, or PLFS, uses these different views. Its method for Calendar Year 2025 provides a useful current reference for understanding them.
Usual status gives the longer-period picture
Usual status looks at the 365 days before the interview. The activity on which a person spent relatively more time determines the usual principal status. The method first decides whether the person was mainly inside or outside the labour force and then, for a person mainly inside it, whether employment or unemployment occupied more time.
A person may also have pursued an economic activity for at least 30 days during those 365 days even when it was not the principal activity. That qualifying work becomes a subsidiary economic activity. The combined label `ps+ss` means principal status plus subsidiary status.
This addition matters for intermittent work. A person mainly engaged in study or domestic duties may still perform enough economic activity to enter the worker count under `ps+ss`. Principal status alone would miss that work.
Usual status is valuable for a structural annual picture. It is not a simple average of monthly rates, and its long frame can smooth brief unemployment or seasonal shortage. The reference period must remain attached to the estimate.
Current Weekly Status gives the shorter-period picture
Current Weekly Status, or CWS, looks back across the preceding week. It gives employment priority over unemployment, and unemployment priority over outside-labour-force status.
Under the Calendar Year 2025 method, a single hour of work on any day in that week is enough for CWS employment. Qualifying attachment to work can also produce that status despite temporary absence. If a person performed no economic activity, the survey next checks search and availability. Seeking work or being available for an hour on any day in the week can produce CWS unemployment. A person meeting neither condition is outside the labour force under CWS.
The priority prevents double counting. A casual worker who worked briefly on one day and searched for more work on the remaining days is classified as employed for the week. That is a status rule, not a claim that the person obtained enough work. One hour of work does not establish full-time hours, adequate income or stability.
Current Daily Status reveals variation within the week
Current Daily Status, or CDS, records a person's activity for each day of the reference week. It can show person-days of employment and unemployment hidden by one weekly label.
Suppose the casual worker worked on Monday and found no work on the other days. CWS calls the person employed because employment has priority. The daily record preserves the fact that work occurred on only part of the week. Person-days and employed persons are different units, so their totals should not be merged.
Long, weekly and daily frames are complementary. A careful answer never treats usual status, CWS and CDS as interchangeable estimates of one identical thing.
Unemployment does not cover every shortage of work
An employed person can still want and be available for more hours. Time-related underemployment identifies such a person when actual hours fall below the threshold defined for the measure. It remains a condition within employment, not unemployment.
Intermittent work can produce the same concern. A worker may obtain a few days of work, remain employed under a priority rule and still lack enough work over the week or year. Seasonal unemployment is different again: work regularly becomes unavailable during part of a production calendar.
Some people outside the labour force also remain connected to the labour market. The potential labour force can include people who want work and either searched without being currently available, or were available but did not search, under the applicable definition. Discouraged people may stop searching after repeated failure. Care duties, illness, unsafe travel or a temporary constraint can also break either the search or availability condition.
Marginal attachment is a broader descriptive idea for people outside the labour force who want work and meet some, but not all, conditions for headline unemployment. Its exact boundary depends on the rule being applied. It must not be turned into one universal extra category.
Exact treatment depends on the survey rule. India's CWS operational rule, for example, uses its own stated search-or-availability test. A label developed for another measure cannot be imposed without checking compatibility. Nor should unemployment, underemployment and potential labour force simply be added into one invented rate; their conditions and denominators differ.
Open unemployment describes people who have no employment and satisfy the applicable unemployment test. Disguised unemployment asks a different production question. Imagine a small family farm shared by six workers. If the acreage, equipment and crop stay unchanged, perhaps four family members could maintain its output. The survey may still classify all six as employed, although some add little or nothing to production under those conditions.
Disguised unemployment is therefore an analytical claim about labour and output. An interviewer cannot measure it merely by asking which worker seems unnecessary. The diagnosis requires a credible comparison of production with and without some labour while other relevant conditions are held in view. It is not a synonym for all farm work or all self-employment.
Unemployment labels explain possible mechanisms
The survey first measures status. Terms such as frictional, structural and cyclical unemployment try to explain why unemployment exists. One measured unemployed person may fit more than one mechanism.
Frictional unemployment arises during search and transition. A new entrant looks for a first suitable job, or a worker moves between jobs. Workers and vacancies may coexist because matching takes time and information.
Structural unemployment arises when workers and available jobs remain mismatched in skill, location, sector or other requirements. Technology may contribute when tasks change faster than workers, firms or places can adapt. The survey does not assign “technological unemployment” as a separate status merely because technology is involved.
Cyclical unemployment follows weak economy-wide demand and output during a downturn. Firms sell less, reduce production and hire fewer workers. The business-cycle mechanism may overlap with structural problems if displaced workers later face lasting mismatch.
Seasonal unemployment follows a recurring production or demand calendar. Long-term unemployment describes duration, while educated unemployment describes a subgroup. Neither duration nor education reveals the cause by itself. A young graduate can simultaneously face a long search and a structural location or skill mismatch.
These labels aid diagnosis; they do not form a set of boxes whose percentages can be added. The idea of one fixed “natural” or desirable unemployment rate is not needed for India's measurement framework and does not belong to this chapter's scope.
Participation and hiring respond to more than wages
Firms and other producers demand labour because they expect workers to help produce goods and services. Labour demand is therefore derived demand. It depends on demand for output, worker productivity, wages and other costs, technology, access to finance and the ability to combine labour with land, machinery, energy and materials.
People's labour supply is also constrained. A person may compare the wage with the value and cost of their time, but that is only part of the decision. Health, education, care duties, safety, transport, migration costs, social norms, job information and the actual availability of suitable work can all affect participation and hours.
This is why non-participation cannot automatically be called voluntary leisure. A student may be building skills. A retired person may have completed a working life. A caregiver may be unable to combine paid work with essential care. A discouraged worker may want employment but stop active search. The rate cannot distinguish these stories without further information.
Women's work makes the boundary problem especially visible. Subsistence production, home-based production and unpaid help in a family enterprise may be missed when questions or respondents treat them as extensions of domestic duty. At the same time, unpaid care for one's own household can consume many hours while remaining outside the employment boundary.
Care burdens, safety, mobility, suitable work, wage conditions, household circumstances, norms and reporting can interact. No single cultural or economic explanation can be read from a participation gap. Better activity questions can change measured employment even when lived activity changes less, while a genuine expansion of opportunity can change both.
Rural and urban location, age, sex, education, region and social position are valuable comparison lenses. They are not causes by themselves. Every comparison must match the age universe, period, status frame, geography and survey design. Small subgroup samples and imprecise estimates require additional caution.
Employment quantity is different from job quality and formality
Employment status answers whether qualifying work occurred. It does not by itself reveal hours, earnings, stability, safety, social protection, bargaining power, skill match or future prospects. WPR may rise while many workers obtain too few hours or insecure work. UR may fall without a broad improvement in job quality.
An informal-sector enterprise is classified by features of the production unit. Informal employment is classified by features of the job relationship or protection. The two can overlap, but they are not synonyms. Informal employment can exist outside an informal enterprise, and an enterprise can contain workers with different arrangements. The detailed mechanisms and institutions belong to the next labour chapter.
Output growth and employment growth must also be kept separate. Labour productivity broadly relates output to labour input. It can rise when workers use better tools, skills, organisation or technology, but it can also change with hours, capacity use and the movement of workers across activities.
Employment elasticity of output compares the percentage change in employment with the percentage change in output over compatible periods. It is not the number of jobs divided by the level of GDP. A value can be difficult to interpret when either change is close to zero, definitions change or short-run shocks dominate.
“Jobless growth” is often used loosely. It should not be taken to mean that literally no job exists or no employment was added. A more precise expression is job-poor growth: output expands while employment grows weakly relative to the labour force, population or policy need. Productivity, working hours, sector composition and the starting base all matter to the diagnosis.
A larger working-age population therefore creates only a potential demographic dividend. People need health, learning, suitable work, capital and capable institutions. A population structure cannot guarantee employment or productivity.
Different data systems answer different employment questions
A household labour-force survey asks people about their activity during a reference period. It can cover salaried workers, casual workers, self-employed people and many people outside registered establishments. Its unit is usually the person or household, and its estimates depend on classification responses and survey weights.
An establishment survey asks businesses or workplaces about workers or jobs within its coverage. A payroll or registration system counts entries and records in a defined administrative system. A census-type exercise aims for broad enumeration but occurs less often. None is automatically a superior version of the others because each asks a different question and covers a different universe.
The results need not match. A household respondent may report a person who holds two jobs, while establishment data may record both positions. A payroll system may exclude unregistered activity. An administrative addition can reflect formalisation, a change of employer or a delayed registration rather than a newly employed person in the whole economy.
Survey estimates have their own limits. Recall affects how activity is reported. Seasonality affects which work appears in a period. A household member may answer for someone else, creating proxy-response error. Question wording and interviewer classification can affect borderline activities. Non-response and sample design affect representation.
Raw respondents are not the national workforce. Survey weights translate the sample into population estimates. Absolute worker totals add a population-projection layer and inherit its assumptions. Rounded rates can hide small differences, while sampling error makes some apparent changes uncertain. Relative standard errors and effective subgroup samples help show this precision.
Questionnaires, classifications, weights, sample frames and population projections can change. An estimate labelled provisional or revised must retain that vintage. A time comparison must therefore ask whether the measurement instrument stayed compatible, not just whether the printed rate changed.
India’s survey architecture changed in January 2025
PLFS began in April 2017. Under its original architecture, annual reports covered rural and urban areas using usual status and CWS, while quarterly reports supplied urban CWS estimates. Each old annual round covered an interval beginning in July and ending the following June.
The design changed from January 2025. The sample structure, allocation, rotation and parts of the enquiry were redesigned. From Calendar Year 2025, the annual cycle moved to January–December.
As of August 2026, the redesigned system publishes three product frequencies. The monthly series provides all-India rural and urban estimates under CWS. Quarterly products provide rural and urban CWS estimates. Annual products cover both usual status and CWS in rural and urban areas. These frequencies are not new activity-status definitions.
A monthly bulletin does not count everyone hired or dismissed during that month. It combines household responses, each classified from the week immediately preceding that interview. A quarterly estimate similarly describes status in responses collected through the quarter; it is not an administrative job-flow total.
The core concepts remained broadly continuous, but changes in sampling and estimation can affect estimated levels. Results from before January 2025 and after the redesign should not be joined into an unqualified trend. The comparison must preserve age, sex, geography and status frame and must acknowledge the design break.
The rotating design creates another limit. Some households appear in adjacent survey periods, so consecutive estimates are not wholly independent samples. Seasonal activity also means that a raw adjacent-month change can reflect the calendar. Matched periods, compatible methods and any valid seasonal adjustment are needed before a strong claim is made.
Read labour data as a complete measurement sentence
Begin every interpretation with the population, age universe, place, reference period, status frame and survey vintage. Then identify the numerator and denominator. Ask which movement between employment, unemployment and outside the labour force could have produced the rate.
Next separate people from jobs and status headcounts from hires, separations and duration. Check whether the source is a household survey, establishment record, payroll system or census-type count. Look for design changes, seasonality, revisions, weighting, projection assumptions and sampling uncertainty.
Finally, remember what the main rates cannot establish. Participation does not prove access to suitable work. Employment does not prove adequate hours, earnings, productivity, stability, security or formality. A low UR does not by itself prove inclusion, and a high UR does not by itself prove that existing jobs disappeared.
The basic story remains simple. Classify people consistently for a stated period. Build each rate from the correct groups. Trace changes to movements between those groups. Then add the wider evidence needed to judge whether an economy is creating enough productive and secure work.