Opinion Mahesh Vyas writes: Why the job shortage is for real

The government has also succeeded in dramatically improving its data releases in recent years. Since 2017-18, delays and irregularity in release of employment data by the government has reduced enormously. We can always ask for more. But, we must acknowledge the progress made by India’s official statistical system. Further, unlike in 2018, official data are not fretted over publicly by government officials anymore, as it was then. The government and the official statistical machinery have made progress and this must be acknowledged even as we demand further progress.

The authors claim that CMIE does not report monthly data on employment. This is incorrect. CMIE publishes monthly data on employment regularly. These are often discussed in the media. Preetha Joseph and Raashika Moudgill have used the CMIE-published monthly employment estimates in their work for the CEDA-CMIE Bulletin. CMIE stands by these estimates. These estimates imply that between January 2020 and September 2022, employment fell by 14 million or by 3.4 per cent.

Bhalla and Das compare this 3.4 per cent fall in employment with the 8.3 per cent rise in national GDP between 2019 and 2020 and claim that both numbers cannot be right. Both can be right, but it is not the best practice to compare monthly point estimates of employment with annual estimates of GDP. If we update the comparison to December 2022 and compare it to a like pre-Covid month, December 2019, we see a 4.6 million increase in employment. It is not advisable to compare point monthly estimates with annual GDP estimates. Joseph and Moudgill don’t make such a comparison.

Bhalla and Das then use what they call as the “non-interpolated non synthetic original CMIE data for the calendar years 2019 and 2022”. This again, is wrong. CMIE does not publish calendar year estimates of employment. The claim that these are original CMIE estimates is incorrect. The authors seem to have derived them as simple averages of results published by CMIE for three different waves in the calendar years 2019 and 2022. It is possible to derive calendar year estimates. To do that one needs to pool the raw data of the three waves and use appropriate weights. The superset of sample observations from three waves is larger than any of the individual waves and therefore needs a different set of weights. The shortcut of averaging of estimates of three individual waves may not yield the correct annual estimate. CMIE avoids such shortcuts.

CMIE publishes wave-wise estimates, fiscal year-wise estimates, quarterly estimates and monthly estimates. It does so by pooling the sample data as required for the estimates and using the appropriate weights. Bhalla and Das have done none of this.

Bhalla and Das claim that the CPHS data lacks credibility because the female labour force participation rate is much lower than that observed by the Periodic Labour Force Survey of the government. True, and the reason for this is a well-known fact. It is because CMIE uses a different definition of employment. It eschews the rather relaxed definition in the official system of classifying a person as employed if such a person was employed for even just one hour in a seven day reference period, to the exclusion of all other statuses during the same week. CMIE’s CPHS considers a person to be employed if such a person is employed for a better part of a day.

Because of CMIE’s different approach, India now has two different estimates of female labour force participation (PLFS and CPHS). Both are useful and one can use either of the two, or even both. We have a choice.

Finally, a clarification on the World Bank report alluded to by the authors. The paper by Sutirtha Sinha Roy and Roy van der Weide referred to in the World Bank report uses the CPHS data to emulate an NSSO survey for the purpose of estimating poverty in India. They have reconstructed the dataset (the sample itself) and transformed the data substantially to construct estimates that can possibly be comparable to estimates based on a methodology as close as possible to the 2011-12 consumption expenditure survey of the NSSO. These transformations are a researcher’s prerogative. Others have estimated poverty using the same CPHS datasets differently. We welcome this diversity.

There is no dearth of data on the Indian labour markets. There is no credibility problem with any of the large household survey based datasets on the subject either. Researchers are putting all this to great use. Both, the CPHS and the NSO databases are being used extensively by researchers to throw light on the Indian economy. Possibly, no other country can match such an incredibly data-rich environment – copious, diverse, official and also private.

The CEDA-CMIE collaboration is an effort to provide data-driven insights into the trends in employment in India. We should welcome and engage with such an initiative and look forward to many more such efforts.

The writer is Managing Director & CEO, CMIE