Getting Good Jobs Data Is Not Rocket Science: Here’s How To Do It, If Government Has The Will For It

The Bharatiya Janata Party (BJP)-led National Democratic Alliance government has been muttering darkly about the unreliability of unemployment statistics for the last two years or more. But its position is untenable because it has done almost nothing to make new, more reliable, data sets available. To be fair, it has been providing monthly payroll data using provident fund subscriber numbers (Employees' Provident Fund Organisation or EPFO) as proxy. But EPFO data is no substitute for a country-wide jobs number.

It does not take two years to set up a new structure or format to generate better jobs data. If you put a handful of economists and statisticians in a room and ask them to come up with a workable plan for generating jobs data, within a week you will have one. The only thing the government needs to do is fund the infrastructure for collecting the data.

India needs several kind of jobs data, from broadbased household surveys to enterprise surveys, among other things. We also need the data to be sliced and diced differently, as in the US, where there are six different measures of employment, from U1 to U6, with U3 being the main official unemployment figure. The U3 definition of unemployment is those without jobs and who have been actively looking for work in the last four weeks. U1 measures longer term unemployment, and includes those without a jobs for 15 weeks or more; U2 is about the percentage of the labour force who lost jobs or have finished temporary work contracts; U4 is U3 plus those who have stopped looking for work due to economic conditions; U5 is U4 plus “other marginally attached workers”, or those who would like a job, but are not looking for one actively; U6 is U5 plus part-time workers who want full-time jobs.

In India’s case, collecting such detailed jobs data is tougher and costlier, because we can’t get these details over the phone from those surveyed. We have to physically send surveyors to gather this data from rural and other homes in order to be sure we have a good sample and good data.

Ideally, all the data should be gathered on the same day, so that it is accurate and current. Collecting jobs data on the same day, month after month, will give us a better picture of seasonal employment, and also the changing mix between formal and informal jobs, and payroll and non-payroll jobs and/or self-employment. Over time, we will get a good time series too.

The National Sample Survey Office (NSSO) and the Centre for Monitoring Indian Economy (CMIE) do their employment-unemployment surveys over several months, which means if someone is contacted on Day 1 in Month 1 and he reports he has no job, he will be counted as jobless four months later, even if he has found work by the time the report is published. Ultimately, we must collect jobs data either on the same day, or on days very close to a central date for the survey. It can’t be spread out over four months or a year, as is the case with CMIE or the NSSO.

In sum, we need:

#1: A national household survey with over 3 lakh respondents (CMIE and NSSO use about half that number) spread out over all states, done monthly on a specific day. We can start this process quarterly and then make it monthly. The NSSO is the right organisation to do it, but it would need additional staff and part-time surveyors to pitch in. It needs a healthier budget for this task. The money would be well spent, for without good jobs data, no government can work out a sensible job-creating policy environment.

#2: The household surveys must seek details on current employment status, type of employment (payroll, self-employment, informal and part-time work, etc), period of unemployment, and whether one is actively seeking work. In the US, you won’t be called unemployed if you have not shown serious efforts to find a job. In India, the assumption is that government must somehow find a job for you. The US way of finding out if you are actively looking for a job includes evidence of the following: whether you have actively sought an interview with a potential employer, whether you are formally registered with an employment agency, whether you have contacted friends or relatives on job openings, whether you have sent resumes to potential employers, etc. The questions we need to ask in India, where the informal sector is large, may be different, but there are ways to establish whether a respondent is actively seeking work as opposed to merely sitting idle and demanding a job as an entitlement.

#3: Enterprise surveys should combine data coming from the Annual Survey of Industries, the EPFO, central, state and public sector employment, and jobs reported by various professional bodies (like CAs, lawyers, etc), among other sources. We must find a way to de-duplicate the data from multiple sources so that we get a consolidated number on formal employment. This way we may get realistic non-farm payroll data.

#4: We need separate data on farm work, and this can only be done through quarterly sample surveys.

We can get the best jobs data within a year if we really want it. If the government just blows hot and cold on the quality of existing data, it is not sincere in finding a way to get better data. That would be a pity. A bonus: creating a large and permanent job survey infrastructure will itself provide worthwhile jobs.