ISSN (Online) - 2349-8846
-A A +A

Counting Jobs in India

A Detailed Review of Labour Database

Jitender Singh ( is with the Department of Financial Services, Ministry of Finance, Government of India; and Arup Mitra ( is with the Institute of Economic Growth, Delhi.

A detailed review of various sources of labour statistics in India highlights the lack of long-time series data on total employment. The Labour Bureau’s attempt since the last couple of years in this respect has been helpful. To gauge the accuracy of these estimates, it is desirable to have data from at least two sources. Dependence on one specific source can be risky.

The authors are indebted to the anonymous referee for constructive comments.

Internationally, countries depending on their size and degree of development compile quality information on job creation. In the United States (US), Bureau of Labor Statistics (BLS) conducts monthly household and establishment surveys to generate information on employment and unemployment. Besides, a Quarterly Census of Employment and Wages (QCEW) also generates estimates on employment.

The establishment survey (also called payroll survey) covers non-farm wage and salary jobs. On a monthly basis, it covers approximately 1,46,000 establishments. The survey generates information on employment, hours of work, and earnings across industries and geographies in the country. Simultaneously, a monthly household survey is also carried out, covering approximately 60,000 eligible households every month. In addition, the QCEW is conducted and published with a six months’ time lag for the evaluated quarter. This shows that a sound system of statistics is in place for the generation of updated high frequency information on employment in the US. Many European countries also estimate employment quarterly.

Employment Statistics in India

The Government of India spends huge resources on collection of employment statistics. Substantial human labour is employed to generate reliable and timely estimates. The mechanisms and systems are in place: various central government departments/organisations are directly involved and appropriate statutes have been enacted to enforce information collection.

Interestingly, in India, the sources of employment data are many. The National Sample Survey Office (NSSO) carries out the most comprehensive quinquennial Employment and Unemployment Survey (EUS). The survey collects information from households, across industries and geographies in the country, but only once in five years though recently a decision has been reached that annual labour force surveys will be carried out by the NSSO instead of the Labour Bureau. A huge time gap of five years reduces the utility of the data in assessing individuals’ livelihood opportunities that may be changing fast in response to economic variations. This, in turn, dampens the scope of initiating new policies in the short-run based on such low frequency data. In fact, the quinquennial surveys are useful only for five-year planning and research.

The Labour Bureau, under the Ministry of Labour and Employment, has been conducting labour force surveys every year though the sample size is quite small. The Labour Bureau has also been conducting Quarterly Employment Survey (QES) since January 2009. However, the scope of QES is limited as it covers only few industries, including textiles, metals, automobiles, gems and jewellery, transport, information technology/business process outsourcing, and mining. The latest survey is available up to September 2015. Its sample size is also small. Nevertheless, QES is useful to monitor the impact of global slowdown on employment in export-oriented industries.

The Labour Bureau compiles useful information on industrial disputes, covering closures, retrenchments, and lay-offs. However, this information is limited to factories employing 100 and more workers.

Besides, Labour Bureau compiles a Statistics on Factories report every year. However, only 40.82% of the total working factories submitted their annual returns for 2012, indicating gross underestimation. Besides, the results are released with a time lag.

Since 1959, the Central Statistics Office (CSO) has been conducting the Annual Survey of Industries (ASI). It is the most comprehensive data set for registered manufacturing. It covers factories employing 10 or more workers with power and 20 or more without power. Micro, small and medium enterprises (MSME) census reports comprise another source of information on employment. The CSO conducts Economic Census every five years and the latest is for 2013–14 for the reference period January 2013 to April 2014.

The Directorate General of Employment and Training (DGET), Ministry of Labour and Employment, produces two reports. One is Employment Exchange Statistics and the second is Employment Reviews. The first is annual and collects information on registration, placement, live register and submission from about 978 employment exchanges across states. The second report provides information at short intervals about the structure of employment in public and organised private sector at the state and national levels.

The Employees’ Provident Fund Organisation (EPFO), Ministry of Labour and Employment, collects information on monthly basis from 187 specified industries/classes of establishments as specified in schedule I of the Industries (Development and Regulation) Act (IDRA), 1951 or any activity notified by the central government in the official gazette. Every year, the Department of Public Enterprises (DPE), Ministry of Heavy Industries and Public Enterprises, publishesthe Public Enterprises Survey reports based on 290 central units. The Department of Industrial Policy and Promotion (DIPP) receives information on potential job creation through Industrial Entrepreneurs Memorandum (IEMs) under the IDRA, 1951. Besides, MCA21 data is recorded by Ministry of Corporate Affairs (MCA) for the companies incorporated under the Companies Act, 1956. In addition, National Technical Manpower Information System (NTMIS) collects data from students, educational institutions, and users of technical manpower to estimate the demand and supply of technical manpower in India. The Department of Commerce, Government of India, publishes information on employment in special economic zones (SEZs).

The following inferences can be drawn from the above elaboration:

(i) Multiple departments/organisations under the central government are engaged in collecting information on employment/jobs in India. The NSSO, CSO, DGET, EPFO, Labour Bureau, Office of Development Commissioner (MSME), DPE, DIPP, and MCA are some of the prominent ones.

(ii) Multiple statutes are backing the collections of information. The IDRA, 1951; the Employment Exchanges (Compulsory Notification of Vacancies) Act, 1959; the Employees’ Provident Funds and Miscellaneous Provisions Act, 1952; the Factories Act, 1948; the Industrial Dispute Act, 1947; the Collection of Statistics Act, 2008 and the Companies Act, 2013 are important.

(iii) Multiple approaches are followed to collect information such as “enterprise,” “establishment,” “household,” and “individual.” The MCA21 (MCA) andthe Public Enterprises Survey (DPE) follow the “enterprise” approach. The “establishment” approach is followed by QES (Labour Bureau), Statistics on Industrial Disputes, Closures, Retrenchments and Lay-offs (Labour Bureau), Statistics of Factories (Labour Bureau), ASI (CSO), MSME census reports (Development Commissioner), Economic Census report (CSO), Employment Market Information Programme (EMIP) (DGET), Employees’ Provident Funds Data (EPFO) and IEMs (DIPP). The “household” approach is followed by NSSO’s quinquennial EUS and Labour Bureau’s Annual EUS. Information is collected from individuals in Employment Exchange Statistics by DGET.

(iv) In some of the sources, the responsibility of reporting lies with the employer, especially in cases where employers should file compliance reports under some statute. In some other sources, the collecting agency approaches the employer for collecting information. Data on employment exchange is based on self-targeting where individuals register themselves.

(v) The frequency of data sources also varies. The sources for monthly information include Statistics on Industrial Disputes, Closures, Retrenchments and Lay-offs, EPFO data and IEMs. The sources for quarterly information include QES (Labour Bureau) and Employment Reviews under the EMIP (DGET). The annual EUS and Statistics of Factories by Labour Bureau, ASI by CSO, Employment Exchange Statistics by DGET,the Public Enterprises Survey reports by DPE and MCA21 by MCA collect information every year. There are also quinquennial and decadal reports produced by the CSO, the NSSO and the Ministry of MSME.

(vi) Various sources also differ in their scope, coverage, and other details that may arise due to the differences in the purpose of collection.

Data Sources

Broadly, these data sources fall into two categories. First, when the onus of collecting information lies on the government, it can be called “survey based.” Second, when, the onus of submitting returns is on employer/establishment/enterprise/individuals it can be called “return based.” The survey-based sources include NSSO surveys, QES and EUS (Labour Bureau), ASI, and Census and Economic Census by CSO and MSME census. The return-based data sources include factory statistics, EMIP, employment exchange data, Statistics on Industrial Disputes, Closures, Retrenchments and Lay-offs and EPFO data. The survey-based data is comprehensive and is used to feed into the statistical system in India. However, the return-based data is for specific purpose and is generated as a by-product of enforcement of an act.

The survey-based data is more updated than return-based. Survey-based data is also more complete and reliable. However, its frequency is very low: decadal, quinquennial, yearly. On the other hand, the frequency of return-based data is high. Multiple organisations are responsible for the collection of return-based data, using different approaches and frequencies. But it is incomplete, outdated, and unreliable due to its poor response rate and slow processing.

Estimates of Employment

The employment estimates given in Table 1 varies across sources due to variation in their reference period and difference in their coverage, and so on. The latest employment estimates available as of now is up to September 2015 by QES, Labour Bureau. However, it is limited only to a few industries.

The employment figure in the Statistics of Factories reports appears to be grossly underestimated. The ASI estimates employment for 2012–13 (financial year) at 129.5 lakh persons, while the Statistics of Factories estimates only 13.12 lakh persons for 2013 (calendar year). The difference exists even though the coverage in ASI and Statistics of Factories is almost the same. Moreover, despite the Statistics of Factories being return-based, its latest estimates are available up to 2013 only, while ASI estimates are more updated and available up to 2014–15.

Comparable estimates by the EMIP for 2012 are 295.43 lakh persons. The EMIP covers all establishments in the public sector irrespective of their size and non-agricultural establishments in the private sector employing 10 or more persons. The estimates on employment from various sources are difficult to reconcile. The return-based data are also incoherent and inconsistent with survey-based data.

The quality of return-based data is poor due to low response from employers. Only 66% of the total registered working factories submitted returns in the Statistics of Factories, 2013. The low response is probably because of weak mechanisms to ensure compliance in filing timely returns. The delays in aggregation and publishing information can be attributed to limited administrative capacity, duplicity of tasks, and multiple aggregation points. The technology used in processing this information is also obsolete. All these factors taken together make the return-based data on employment/jobs irrelevant, non-actionable, incomplete, unreliable, incoherent, and inconsistent.

The multiplicity of acts requires an employer to file the same information for different regulatory agencies. This overburdens the employer. It also hampers the ease of doing business in the country. The 2014 Labour Bureau report also acknowledges the fact that labour statistics in its present form is dated and of poor quality, thus limiting its reliability and use. The report also examined various laws and rules, and recommended that they be simplified. It also suggested that data should be captured electronically. However, the scope of the report was restricted only to labour laws monitored by the Labour Bureau. There is a need to take a comprehensive review of return-based sources of employment estimates for collection of timely, reliable, and actionable information on employment/jobs.

The workforce participation rates (WPR) (principal and subsidiary workers) given by Labour Bureau (Table 3) do not match with the corresponding NSSO rates (Tables 2 and 3). Part of the difference can be attributed to the fact that the NSSO rates in parentheses in Table 3 correspond to ages 15 to 59 years while the Labour Bureau rates cover the age groups 15 years and above.


Though the Labour Bureau WPR do not exactly tally with the NSSO WPR, a reasonably good panel data set has been emerging that can be utilised to identify the determinants of participation. The state-level cross-section-time-series pooled data can be used to determine the impact of growth fluctuations on employment—an issue that has been bothering both policymakers and researchers for a while.

The other control variables could be state level ones like state government expenditures that can vary every year, and other cultural, social, demographic, and infrastructure variables that are not likely to show huge changes in the short run. A detailed analysis based on state-level data brings out a wide range of interesting results with important policy implications (Mitra and Okada forthcoming).

However, as mentioned before, the Labour Bureau’s surveys on labour force has been discontinued. The NSSO, instead of conducting the EUS once in five years has now initiated the annual labour force surveys. Since the NSSO has already acquired the expertise of carrying out the EUS and has established a reputation among the users with regard to the reliability of the employment figures, similar exercises annually may not be difficult. However, in the recent past, the 2009–10 estimates were questioned, and thus, the EUS survey had to be repeated in 2011–12 much before the five-year period got over.

The engagement of contractual investigators who have replaced the regular investigators to a large extent might have resulted in such discrepancy between the estimates and the expected outcome for 2009–10. But if that is the case, the same argument can hold for the annual surveys to be conducted by the NSSO. On the other hand, Labour Bureau had acquired reasonable experience to handle such surveys after repeating similar exercises for the last several years. Maybe for the initial years, the figures showed some discrepancy, but over time the Labour Bureau figures were tending to converge with the NSSO figures. If Labour Bureau surveys were to continue, the users could have more than one source of labour and employment statistics at least once in five years to compare and determine the reliability of the aggregate nationwide employment data. Now the users will be left with no choice: the NSSO data will have to be accepted without any questions. If in a given year, the biases are strong in comparison to the immediate past year, the differences in the estimates will be interpreted as the actual change in employment.


The technology used in collecting information is evolving very fast. The use of computer network, digital gadgets, customised software and mobile web portal can reduce the cost of data entry and aggregation considerably. Faster processing of the data can save time. Technology can enable storage of larger data sets, easy transformation in usable form, easy retrieval, and query-based outcomes. There is a need to examine the further use of technology in collection, processing, storage and dissemination of labour/employment statistics in India.

Let us assume there is a digital product, which can be a mobile app, and establishments/enterprises have to feed only few figures every month/quarter/year. These figures can be verified if needed through random checks or putting possible error alerts. The digital product can also automatically remind the respondent to feed the information and help locate the respondent through the global positioning system.

It will provide the information of the contact person and can maintain the list of contact (active or passive) persons. These few figures are the main indicators and can be churned out from various
returns (production, labour, disputes, etc) a respondent is required to file at various times to different agencies. Since the information sent through this digital product can be accessed by all agencies simultaneously, including the monitoring agency in the central government, the information can be compiled concurrently at the centre and state levels without any time lag. It will avoid duplicity of data entry and the entry at the respondent level will be sufficient. It will save time, money, human labour, and provide updated data on employment/job creation in time.


Labour Bureau (2014): “Report on Simplification of Returns under Labour Laws,” Chandigarh.

Mitra, Arup and Ava Okada (forthcoming): “Labour Market Participation in India: A Region and Gender Specific Study,” Singapore: Springer.

Thomas, Jayan Jose (2015): “India’s Labour Market during the 2000s: An Overview,” K V Ramaswamy (ed), Labour, Employment and Economic Growth in India, Cambridge.

The authors are indebted to the anonymous referee for constructive comments.

Updated On : 13th Mar, 2018


(-) Hide

EPW looks forward to your comments. Please note that comments are moderated as per our comments policy. They may take some time to appear. A comment, if suitable, may be selected for publication in the Letters pages of EPW.

Back to Top