Last week, the Ministry of Labour and Employment released the results of a new Quarterly Employment Survey (QES) for April-June 2021 for the organised (formal) sector. It represents establishments (or units) employing ten or more workers. The surveyed sectors were manufacturing, construction, trade, transport, education, health, accommodation and restaurant, Information Technology/Business Process Outsourcing (IT/BPO), and financial services. The survey reported the following. First, “It is heartening to note that the estimated total employment in the nine selected sectors from the first round of QES works out as 3 crores and 8 lakhs approximately against a total of 2 crores and 37 lakhs in these sectors taken collectively, as reported in the Sixth Economic Census (2013-14), implying a growth of 29%.” Secondly, employment fell from 3.078 crore before the first nationwide lockdown in March 2020 to 2.848 crore post the lockdown on July 1, 2020. Thus, the report showed that 24 lakh jobs lost during the lockdown in 2020 came back by the first quarter of 2021.
The new QES is a welcome step as it will help generate timely employment estimates for the larger units. However, the above analysis is fraught with caveats and the interpretation of the survey results calls for caution.
Limited coverage
As is widely known, establishments with ten or more workers account for a small proportion of all non-agricultural establishments — a mere 1.66% as per the Economic Census (EC) of 2013-14. Also, a disproportionately large share of workers — 81.3% as per the Periodic Labour Force Survey (PLFS, 2018-19) — worked in the unorganised sector. With its limited coverage, the QES based on data for formal sector enterprises cannot provide a total picture of employment dynamics. Hence, drawing inferences about overall job losses during the lockdown based on this survey is simply misleading.
Data from the Centre for Monitoring Indian Economy (CMIE) showed that during April 2020, 12.1 crore workers lost their jobs. Most of these workers were informally employed — 9.1 crore job losses reportedly occurred amongst small traders and casual labourers. The official PLFS indicated the extent of distress in the labour market during the nationwide lockdown. The unemployment rate by current weekly status (i.e. the activity status during seven days preceding the survey date) in urban areas increased from 9.1% in January-March 2020 (before the lockdown) to 20.8% in April to June 2020.
Significantly, the new QES data suffer from many methodological shortcomings such as an outdated sample frame, non-comparability of employment numbers obtained from the EC-2013 with sample estimates obtained from the QES for only a quarter, and differences in methods used for gathering the information. We elucidate these problems below.
Conducting a scientific sample survey requires a “sample frame”, i.e. the list of units (e.g., persons, households, businesses, etc.) in the survey population. Since the basis for sample selection is this list, the frame is of utmost importance for the survey design. To generate a frame for its enterprise surveys, India has been conducting the ECs since 1977, albeit at long and irregular intervals. The most recent one was in 2013-14, which the QES has used.
The new QES has a sample of approximately 11,000 establishments. The datedness of the frame implies that the QES does not include units set up after 2013. Further, the EC-2013 has based itself on the “enumeration blocks” of the Population Census, 2011 as the primary geographical units. The outdated nature of the sample frame, which has been acknowledged for a while now, renders the new QES employment estimates irrelevant.
To understand the QES’s origin, the Labour Bureau (LB) conducted the first such survey in 2009, with a modest sample size of around 2,000 manufacturing units for eight select labour-intensive and export-oriented industries in 11 States. The survey sought to assess the employment effects of the global financial crisis and was conducted till December 2015. In April 2016, the LB replaced this series with another quarterly series, which had a larger sample size and enhanced sectoral coverage. The EC-2013 served as the sampling frame for this survey, too. However, it was soon abandoned as a government-appointed Task Force on Improving Employment Data (2017) recommended doing away with the QES on grounds of its limited coverage and an outdated sample frame.
Given the above background, the rush to produce a new QES that draws its sample from the EC-2013 frame seems baffling. It would have been more prudent to await the release of a newly updated frame in the EC-2020 and then canvass for the QES. In fact, given the outdated frame, drawing inferences about employment changes even for the organised sector during the lockdown is misleading. It is also worth noting that QES was primarily a telephonic survey and verification of responses of establishments has not been done (as has been the usual practice for the LB’s quarterly surveys).
Moreover, the comparison of the employment estimates obtained from the new QES for April to June 2021 with the employment number based on the EC-2013 reported in the first paragraph is confounding. The latter is a census, conducted over an entire year, to provide a frame. The former is a sample survey conducted with a short reference period. The questionnaire of the QES asks establishments about employment details for a specific quarter, in this case, April 1, 2021. In contrast, the EC-2013 questionnaire asks establishments about the number of persons working on the last working day prior to the date of fieldwork in the establishment.
Conceptual problem
In addition, there is a conceptual problem in comparing employment numbers of the EC with the QES. Although the former asks questions about the number of persons working in an enterprise, it is not a good instrument for estimating the size of the workforce or for analysing employment trends as the principal objective of the EC is generating a frame, not estimating employment.
The initiative to produce quarterly employment data for selected industries in the organised sector is desirable. However, in a rush to generate high-frequency estimates, there cannot be a compromise on data quality and its reliability. Hence, the inference drawn using the new survey data, as reported in the opening paragraph, is misleading. The LB’s hurried effort has only created (avoidable) confusion and undermined the potential value of the QES.
Radhicka Kapoor works with the Indian Council for Research on International Economic Relations, New Delhi. R. Nagaraj works with the Centre for Development Studies, Thiruvananthapuram
