Payroll data and formal employment
We do need multiple sources of data on formal employment to be able to properly judge its significance and growth
The government of India has started producing monthly estimates of payroll count derived from the database maintained by the Employees’ Provident Fund Organisation (EFPO), which manages social security funds of workers in organised/semi-organised sectors. It recently released the first set of provisional estimates for the six-month period from September 2017 to February 2018. These show that during the six-month period, the new EFPO enrolment of employees was 3.11 million (3.27 million, according to a later release). Several commentators have cited this figure to claim that formal jobs increased very rapidly in the last six months. Far from being jobless, they have gone on to claim, economic growth has been impressively jobcreating in recent times.
Such claims are wholly unwarranted. The reliability of the estimates of payroll numbers remains open to doubt. The EFPO database is known to have serious limitations arising out of the existence of duplicate and inactive accounts. Apparently, some efforts have been made to clean up the database through the Aadhaar-linking of accounts and eliminating inactive accounts. Even if the database is now clean (we cannot be sure it is), the payroll numbers do not give us the number of new formal jobs. In the first place, additions to the payroll still represent the sum total of new formal jobs and newly formalised non-formal jobs and a way to separate these two types has to be found before we can get estimates of new formal jobs. Secondly, EPFOenrolled employees include temporary or casual employees, who cannot be counted as formal employees. Finally, even the EPFO-enrolled regular employees earn a monthly wage of up to ~15,000 while most of the regular employees earning more than ~15,000 per month are not EPFO-enrolled. The payroll count, therefore, can at best give us the number of low-wage formal jobs and not the number of all formal jobs.
The initiative to use the hitherto unused administrative databases (not just the EFPO database but also the Employees’ State Insurance Scheme and National Pension System databases) to produce payroll estimates is to be welcomed not because it will immediately provide the number of new formal jobs being created in the economy but because it will make these databases usable. Efforts to produce payroll estimates from the EFPO database, for example, cannot but involve efforts to fully clean up the database and to sort out the problems of interpretation. After a time, we can have reliable estimates of a certain kind of formal jobs in the economy, which would very usefully complement the estimates of formal employment that the national sample surveys generate.
Starting from 1999-2000, the surveys of employment and unemployment, conducted by the National Sample Survey Organisation (NSSO), have generated data that can be used to derive estimates of formal employment for the years 1999-2000, 2004-05, 2009-10 and 2011-12. A very similar survey conducted by the Labour Bureau for 2015-16 also provides the information necessary for estimating formal employment. And starting from 2017-18, the NSSO’s newly launched periodic labour force surveys will produce annual estimates of formal employment in the economy.
Curiously, these sources of data have gone unnoticed amid laments about the lack of high-frequency employment data. Let me cite some estimates that I have derived for the years 2004-05, 2011-12 and 201516. A formal job is defined as a regular salaried job in government establishments or private enterprises and one that comes with entitlement to one or more of the following social security benefits: Provident fund, pension, gratuity, healthcare and maternity. On this definition, the number of formal jobs in the economy is estimated to have been 31.2 million in 2004-05 (based on the NSSO survey data), 36.4 million in 2011-12 (based on the NSSO survey data) and 41.4 million in 2015-16 (based on the Labour Bureau survey data). Thus, formal employment was growing by 0.74 million per year, on average, between 200405 and 2011-12 and by 1.25 million per year, on average, between 2011-12 and 2015-16. So, formal employment increased faster in the second period (at 3.3 per cent per annum) than in the first (at 2.2 per cent per annum). We may note that if we assume formal employment to have continued to grow at 3.3 per cent during 2015-18, we would conclude that 1.4 million formal jobs were added in 2017-18.
The National Sample Surveys, of course, give estimates of total employment in the economy. We thus know that total employment in the economy increased by 3.46 million per year in the first period and declined by 2.82 million per year in the second. This means that non-formal employment in the economy increased by 2.72 million per year in the first period and declined by 4.07 million per year in the second period. It is hard to avoid the conclusion that the observed increase in formal jobs represented largely new formal jobs in the first period and newly formalised non-formal jobs in the second period. The faster growth of formal employment in the second period, therefore, did not imply better performance of the economy in terms of job creation.
I cite these estimates to make three points. First, while efforts to produce reliable estimates of payroll counts are welcome, we must remember that we do have and will continue to have other very useful sources of data on formal employment. Second, we do need multiple sources of data on formal employment to be able to properly judge its significance and growth. Payroll estimates, even when wholly reliable, would not tell us the number of and the change in formal employment in the economy. They can complement, not substitute for, estimates of formal employment generated by the national sample surveys. Finally, we do need to know the big picture on employment to understand the significance of observed changes in formal employment. And the big picture can never emerge from payroll estimates. Claims about creating formal jobs on the basis of observed trends in payroll numbers alone are and will always be false.