Hazards of relying on labour bureau’s employment data to evaluate job creation
A detailed look at Labour Bureau’s Quarterly Employment Survey (QES) methodology and design suggests that one shouldn’t read too much in these numbers
New Delhi: Opposition parties and many commentators have been citing the Labour Bureau’s Quarterly Employment Survey (QES) numbers to attack the government over failure to generate employment). How credible are such claims? A detailed look at QES methodology and design suggests that one shouldn’t read too much in these numbers.
According to the QES figures, number of jobs created in 2015 and 2016 are the lowest since 2009, the first year for which these figures are available.
The latest QES report is based on responses/records from economic establishments employing 10 or more workers in eight sectors: manufacturing, construction, trade, health, education, restaurants and accommodation, information technology and business process outsourcing, and transport. QES reports are based on the sixth economic census (2013-14) database, which is perceived to be a comprehensive database of non-agricultural economic establishments in the country.
A cursory look at economic census tables shows that these eight sectors roughly account for 92 million workers, constituting 85% of India’s 108 million workers engaged in non-agricultural activities. These numbers are approximations as detailed data has not been made public for the latest economic census. So, it is difficult to separate employment in health from human health and social work activities, the category for which data is available in economic census tables.
The actual number of workers in the sampling frame of QES reports is much smaller. It was just over 20 million according to the first report for new series, which was released in September 2016. The reason is QES’s criteria of only looking at economic establishments with 10 or more workers.
An earlier Plainfacts column showed that the average number of employees in an Indian economic establishment was only 2.24 in 2013-14. Nearly one-third of non-agricultural employment is in establishments which do not employ even one worker . The share of employment in economic establishments with 10 or more workers has been falling and was just above one in five for the sixth economic census.
Even in qualitative terms, the QES sample is hardly representative. According to the September 2016 report, an overwhelming majority of workers in the QES sample are regular (more than 80%) and full-time (96%). Only 75% of India’s workers were employed for more than six months in a year as per 2011 census data. The latest National Sample Survey Office (NSSO) survey (2011-12) shows that only 18% of India’s workers had regular wage/salaried employment. This further underlines the point that the QES sample is not representative of India’s workforce
A non-representative sample is not the only problem of QES employment estimates. QES reports clearly state that these statistics are based on reporting/records by firms and are not verified. This makes their credibility extremely suspect in contrast to NSSO or census figures which are based on responses from workers and not employers. In fact, the report itself is categorical in warning against using it to assess the unemployment situation.
“The QES is basically an establishment survey for collecting information on employment in the unit; therefore, it does not provide any information on unemployment in the country. Unemployment is generally captured by household survey”, it says.
So credible assessment of employment trends must wait for results from NSSO’s next large sample round. That lack of high-frequency employment statistics is a serious handicap for policymaking is a point which cannot be overemphasized. To be sure, the government has set up a committee under NITI Aayog vice-chairman Arvind Panagariya to evolve a methodology to generate timely and reliable employment data. However, it would be delusional to assume that such feats can be achieved without a systemic overhaul of India’s beleaguered official statistical machinery.