Not the data collection methods, but the outcome is really poor.
By Santosh Mehrotra & Jajati Parida
Indian economy is unable to create enough jobs for the increasing number of job aspirants, reflecting in rising unemployment rate. The ‘not in employment, education or training’ NEET population is growing (84 million in 2011-12; 103 million in 2015-16; 116 million in 2017-18). This is reflected in recent employment and labour force surveys of the government (both the Labour Bureau and the Periodic Labour Force Survey) and private (Centre for Monitoring Indian Economy) organisations. A labour market crisis is in the making, as the number of newly- and better-educated entrants into the labour force will only mount each year, at a rising pace till 2030, and decelerating thereafter till 2040.
Unfortunately, experts like Amitabh Kant and Surjit Bhalla hardly recognise this issue. Rather, they claim—in fact, several times—that the data collection methods of the government are not appropriate. Bhalla particularly appears obsessed with misguiding the government and public with stylised interpretations of his own view of statistics (‘Unemployment at 45-year high—a statistical embarrassment’, FE, April 13, https://bit.ly/2vvKB0q).
Unprecedented things are happening. First, the Labour Bureau’s fifth annual survey 2016-17 was discarded for seasonality issues, despite the fact that its first four rounds were well accepted for the same methodology. Moreover, PLFS (2017-18) of the NSS is now being attacked again by them, on methodological grounds.
Systematic efforts seem under way to undermine the credibility of the labour statistics system. First, the five-yearly Employment-Unemployment Survey of the NSSO is discontinued since 2011-12. Second, the quarterly employment survey of the Labour Bureau is discontinued. Third, annual surveys conducted by the Labour Bureau have also been discontinued since 2016. Moreover, the new high frequency PLFS conducted by the NSSO is being questioned by government experts. The current government has not released the results of the last Labour Bureau survey (2016-17), nor the results of PLFS, both of which have been cleared by the concerned authorities for public release. On the other hand, estimating new job creation based on the EPFO data is the most bizarre development in this space ever.
A statistical system, created by PC Mahalanobis, is believed to be the backbone of policy research and development in India. However, demolition of the statistical system would cost India hugely.
The Consumer Pyramids Survey of CMIE with a sample size of 160,000 households and 522,000 individuals too reported 5 million persons lost their jobs between 2016 and 2018, and the overall unemployment rate to be around 6% in 2018. This is comparable with the PLFS data of 6.1%. In addition to rising open unemployment among the higher educated, the less educated (mostly informal) workers have also seen job losses and reduced work opportunities since 2016. Moreover, with the squeezing job opportunities in agriculture and manufacturing because of mechanisation and rising capital intensities, young educated youth preferred to be NEET rather than openly declaring themselves as unemployed.
Although the services sector alone drove employment growth post 2011-12, the failure of a number of start-up projects (increasing every year since 2016), along with employment losses due to airline and other crises, have affected labour market outcomes in the services sector.
Countering Bhalla’s claim that PLFS (2017-18) of the NSS is useless because of population and urban sample issues, it seems once again he is innocent about the fact that the NSS-estimated population is always smaller than actual Census population (see table). A readjustment is normally made to the NSS estimates based on actual Census population, and the unadjusted population per se has never been used by anybody in the globe at any point in time. Hence, his argument of underestimation of NSS population is meaningless.
Moreover, he may have totally forgotten the basics of sampling and its relevance in research. The sample sizes in rural and urban areas normally vary slightly across the rounds of the NSS (see table), as these are based on multi-stage, stratified random sampling, the most appropriate sampling method for a geographically large country like India, in which highly socio-economically heterogeneous groups of people live. This sampling method is not only used for the Employment-Unemployment survey, but has also been used for a number of other household surveys including: the Consumer Expenditure, Housing Condition, Migration, Participation in Education, Disabled Persons, Land & Livestock Holdings, Debt & Investment, Drinking Water, Sanitation and Hygiene, Health and Morbidity, Particulars of Slums, Situation Assessment of Agricultural Households etc. Therefore, if we start believing Bhalla, then it could be concluded that all the NSS surveys conducted till date were useless. This would be the most foolish thing we ever did.
Raising these ill-informed questions seems either motivated, or at best smacking of innocence. Because, for its household surveys, the NSSO has been using basically the same sampling design over the years, with some fine-tuning made every year with the objective of improving the accuracy of important estimates. Moreover, labour market experts in India should dig deeper to explore the factors behind these poor employment results, rather than questioning the reliability of data.
If the current belief system of our experts were to continue, India could end up in a situation of high mass unemployment sooner rather than later. This will happen without the knowledge of our professed experts, with grave if not catastrophic consequences for our nation’s development.