# Formalization and the measurement of Indian employment

Economy Study · India · Labor Markets

Formalization describes several different changes: a business can register, a worker can enter a payroll system, a contract can become written, or employment can acquire social insurance. These transitions need not occur together. An assessment of Indian job creation should identify the particular transition being measured before treating an administrative or survey series as evidence of a broader change in employment quality.

The distinction is substantive. A worker can perform an existing job after joining a benefit scheme. That is potentially an improvement in protection, but it is not necessarily an additional job. A registered enterprise can employ people without written contracts. A self-employed person can become a salaried employee while total employment remains unchanged. Each outcome has economic significance; none should be substituted for all the others.

## What the labor survey actually measures

The Ministry of Statistics and Programme Implementation's [PLFS: Changes in 2025](https://mospi.gov.in/sites/default/files/publication_reports/PLFS_Changes-in-2025_rev.pdf) explains two reference periods. Usual status considers activity over the preceding 365 days; current weekly status considers the preceding seven days. The redesigned survey introduced monthly all-India rural and urban indicators and extended quarterly coverage to rural areas from 2025.

A person with intermittent work can therefore be classified differently across reference periods without either measure being erroneous. Usual status provides a longer view of attachment to economic activity. Weekly status is more sensitive to recent work and availability. The choice should follow the question: structural participation and a short-term employment interruption are different phenomena.

The 2025 redesign also changed sampling and rotation arrangements. A comparison crossing that boundary must examine the official comparability guidance. Greater frequency does not, by itself, make a new monthly observation directly comparable with an earlier annual or urban-only estimate. Reference period, geography, population age, and sample design should be aligned before drawing a trend.

These conditions matter for businesses assessing labor availability. A high participation rate does not establish that workers with the needed skills are available in the intended location. An unemployment rate does not measure all people who would prefer additional paid hours. Occupation, hours, earnings, and mobility provide necessary context.

## Regular employment is not a complete measure of protection

The [Annual Report, PLFS 2025](https://www.mospi.gov.in/uploads/publications_reports/publications_reports1780040415321_0624fb13-fb47-40bc-b470-7c7e9635c3ef_PLFS_2025_F_REV_29052026.pdf), statement 10 and table 36, separates written contracts, paid leave, and specified social security benefits for regular wage or salaried employees in non-agricultural work. For the combined rural and urban population in calendar 2025, it reports 58.2 percent without a written contract, 47.3 percent without paid-leave eligibility, and 51.7 percent without the specified social security benefits.

These percentages describe different conditions within a defined employee population. They are not shares of all Indian workers and should not be added together. The same worker can lack more than one protection. Nor does the complement of any one percentage establish comprehensive formal employment: a written contract does not necessarily imply all the other benefits.

The evidence is useful precisely because it resists a single label. It shows why employment status and employment conditions need separate monitoring. An increase in salaried work can coexist with limited contractual protection. An improvement in social insurance can occur without a change in the distribution between self-employment and wage employment.

For policy evaluation, the relevant transition should be observable. If a reform intends to improve insurance coverage, the assessment should measure active coverage, contribution continuity, and access to benefits. If it intends to create jobs, employment levels or hours need a defensible counterfactual. A rise in registered participants cannot answer both questions automatically.

## Enterprise coverage and worker coverage are different

Household surveys follow individuals and can capture work outside the establishments included in a business survey. Establishment surveys provide information about jobs, payroll, and production within their coverage. Administrative systems record people or transactions that meet the system's registration and reporting conditions. Differences between these sources are expected, not evidence that one must be dismissed.

A household can report several income-generating activities. A business survey might count jobs at establishments, including multiple jobs held by one person. An administrative account can remain registered even when contributions stop. The unit needs to be identified before counts are compared.

Enterprise registration can change the measured perimeter. Existing activity may move into a recorded category after a reporting reform. The resulting increase improves statistical visibility but cannot all be assigned to new production. The same consideration applies when an enterprise expands beyond a survey threshold and enters the covered population.

A transparent reconciliation distinguishes continuing units, newly created units, previously existing units newly captured, and closures. Where linked microdata are unavailable, the analysis should present that limitation directly. It should not force totals from incompatible systems to match by treating unexplained differences as an estimate of informal work.

## The arithmetic of an administrative increase

Suppose, in a hypothetical administrative program, active registered workers increase from 100 to 112. During the period, 15 existing workers register for the first time, five genuinely new jobs enter the system, and eight participants cease active registration. The net increase is 12, but the number of new jobs identified in this accounting is five.

Even that five is not yet a causal employment effect. Some of those jobs might have been created without the policy. The difference between registration additionality, employment additionality, and improved protection remains important. Each requires its own baseline and evidence.

Now suppose some people hold two registered jobs. A job-based series may rise without a comparable increase in the number of employed individuals. If contribution records are cumulative rather than active, exits may not reduce the published total. A net addition series should therefore state how repeat registration, switching, dormancy, and re-entry are treated.

The example does not describe the construction of any particular Indian administrative series. It demonstrates the reconciliation needed before a headline is interpreted. An analysis must consult the relevant system's methodology rather than assume all registries follow this accounting.

## Employment quantity, intensity, and returns

Counting people employed answers only one part of the labor-market question. The number of hours, regularity of work, and real earnings determine how employment translates into household resources. One person moving from occasional work to a stable full-time position can experience a substantial improvement without changing the employment headcount.

Conversely, an increase in employment can include low-hour or low-remuneration activities. Such activity should not be disregarded, but it should be distinguished from secure employment that provides a predictable income. The assessment becomes more useful when it reports the distribution of hours and earnings alongside employment status.

Self-employment deserves particular care. It includes employers, own-account work, and assistance in household enterprises with different financing and income arrangements. A change in its share can reflect entrepreneurial opportunity, a response to a lack of salaried work, or changes in household labor allocation. The category alone does not identify the cause.

Gender and location also affect interpretation. A change in recorded work among rural women can have different implications depending on whether it represents paid employment, own-account activity, or participation in a household enterprise. Evidence about remuneration and the respondent's role is needed before assigning a uniform improvement in economic independence.

## A reconciliation that preserves the distinctions

| Claim under examination | Most relevant observation | Inference that requires additional evidence |
|---|---|---|
| More people are employed | Comparable household employment measures | The policy caused the entire increase |
| More jobs are registered | Active, deduplicated administrative records | Every registration represents a new job |
| Work is better protected | Contracts, leave, contributions, and benefit eligibility | All registered workers receive identical protection |
| Labor income improved | Earnings and hours with a price adjustment | Headcount growth implies higher income per worker |
| Enterprises became more visible | Registration and reporting transitions | All newly recorded production is newly created |

An effective monitoring system can combine these observations while retaining their different meanings. It should document revisions, changes in coverage, and the period to which each observation refers. Disagreement between sources can then become a research question about employment transitions rather than an invitation to choose whichever figure supports the strongest headline.

## Precision and the scale of a reported change

Survey estimates also carry sampling uncertainty. A small difference between two reported rates may be consistent with ordinary sampling variation, particularly for a narrow state, occupation, or demographic group. Publishing more decimal places does not make that difference economically or statistically conclusive. The appropriate comparison uses the survey design and, where available, uncertainty estimates for the change itself.

Revisions should be preserved in an assessment record. A revised estimate can alter a trend without any new employment event taking place in the economy. Comparisons should use compatible vintages or explain the revision. This is particularly important when public discussion combines a recent monthly estimate with an annual estimate extracted before a methodological update.

A transition into a registered establishment can change the statistical visibility of work before changing the worker's hours or output. Indian employment comparisons should therefore preserve the distinction between administrative enrollment and survey employment. Following the same workers over time would help establish whether registration is accompanied by earnings progression, more stable hours, or improved protection rather than merely a different reporting channel.

## Implications for economic assessment

The PLFS evidence establishes that regular wage employment and employment protection cannot be treated as equivalent categories. The 2025 sampling changes also require care when interpreting newly available high-frequency indicators. Together, these features favor a multidimensional assessment of formalization.

The practical objective is to identify improvements in work, income, and protection without counting the same transition several times. Employment creation, greater statistical coverage, and stronger social insurance can each be valuable. A mature assessment reports which of them has occurred, for whom, and on what evidence, while reserving causal conclusions for an evaluation that can support them.


---

American Economic Senate