This piece is in response to more articles in the public domain that came in the last few days including one in Mint last week. Looking at the responses, we are glad that our payroll report has moved the debate on jobless growth to underreporting of payroll data in India. Looking at the reactions so far, barring a few, most of them are wide of the mark. Even so, we have taken care of most of the arguments. Here are our responses to the new ones in this piece.
First, regarding deduplication, we did not merge different databases and our intention was never to merge EPFO and ESIC and NPS to find duplication! The overlap between ESIC and EPFO was removed by (a) using an appropriate haircut and (b) even considering the payroll creation without ESIC and (c) restricting the age category to below 22 where overlap is lowest, if any. Within NPS, we also used the mutual exclusivity principle to only consider those government employees who have been enrolled post 2004 in NPS.
The issue with our research fraternity is that they don’t embrace technology and analytics—the need of the hour. While it is indeed true that ideally one should have a common identifier to merge multiple data sets, it is also true in the era of big data this is hardly an issue. There are plenty of methods to merge multiple administrative databases without common identifier in the machine learning and big data literature.
Second, we never claimed all employees who joined the workforce are new. We took only those who have nonzero contributions continuously, and new accounts—that also in the age group of 18-25—as new. Additionally, it is possible that during any year, existing employees younger than 19 would come into EPF and add to the payroll. However this is a normal event every year and we believe that this figure is not material on net basis as we have excluded first payments from joinees who are 25+ years old and who could have also joined the payroll for the first time (more than 5 million on record).
Third, regarding contract employees, while it is true that contract employees were working in 2016, they were made to join EPF only in 2017 due to better enforcement of law through possibly an amnesty scheme or otherwise. Thus in 2017 the number of employees will be higher and it may overstate the job creation. However, we have deleted around 1 crore payroll creation under amnesty scheme.
We also deleted around 3.2 crore records that were either incomplete or had zero contributions in addition to those under amnesty scheme.
Also, how does it matter even if it is contract (assuming again that all contract employees are in the 18-25 age, an improbable scenario) or not? As long they are on payroll, it does not matter. Also, since we are dealing with only non zero contributions, such contractual jobs are weeded out. This is exactly why we want that the debate should move to: creating quality jobs in India through proper reorientation of skilling.
(Authors Soumya Kanti Ghosh is the chief economic advisor of State Bank of India (SBI) while Pulak Ghosh is a professor at IIM Bangalore. Views are personal)