The reason India jobs data is not credible
Source: By R. Jagannathan: Mint
Thanks to the politicisation of unemployment figures ahead of the general elections, we now have numbers being flung at us from all directions. We have household survey data from the Centre for Monitoring Indian Economy (CMIE); we have the Azim Premji University’s State Of Working India 2018 report; we have monthly data on subscriber additions to the Employees’ Provident Fund Organisation (EPFO) and we have more enterprise data coming from the Confederation of Indian Industry (CII). We also have “leaked" data from the National Sample Survey Office (NSSO) for 2017-18.
In the long run, perhaps all the jobs data will converge to give us meaningful results, but right now, it is only adding to the confusion as all the studies have serious drawbacks. It’s a case of the blind men and the elephant.
The CII employment survey is based on a sample of 105,347 micro, small and medium enterprises (MSMEs). It suggests that employment is growing at a compounded rate of 3.3% annually. The net additions over the last four years amounted to 332,394 jobs. When extrapolated over the Labour Bureau’s entire macro database, this number apparently translates to 13.5-14.9 million jobs created annually. If this were to be true, the Narendra Modi government can clearly declare the jobs problem solved, but such huge extrapolations are unwarranted when we cannot be sure that the CII sample is representative of the entire Labour Bureau database.
At the other end of the spectrum is the CMIE’s four-monthly unemployment survey, which has pegged job losses at around 10 million between September and December 2016 (which includes the demonetisation quarter) and the same months in 2018. CMIE computes the current unemployment rate at 6.68%, which is not significantly different from the leaked NSSO report, which put it at 6.1%. The 2015-16 Labour Bureau’s Employment Unemployment Survey put the unemployment rate at 5%. The Azim Premji University’s State of Working India 2018 simply says that unemployment is now over 5%, with youth unemployment more than three times the overall rate.
This figure is compatible with the Labour Bureau’s survey of 2015-16, which indicated unemployment rates of 13-20% for those in the 15-17 age group, and 10-13% for the 18-29 age groups. Above 30 years, the unemployment rate falls to under 2%. The political brouhaha is about the leaked NSSO data reported by Business Standard suggesting that unemployment is at a “45-year high". The NSSO samples the population for employment information once in five years, and since there is no saying whether the year chosen is a particularly bad one for the economy or not, it is pointless trying to claim some figure is the highest or lowest in 45 years. Maybe the best years for employment came somewhere in between its two surveys.
In fact, let me throw another number to prove a point—a number that comes straight from Census 2011. The unemployment rate reported in the Census is as high as 11.18%. Yes, you read that right. The Census reported an employment figure of 482.88 million “main" and “marginal" workers, with the unemployed (those seeking jobs) placed at 60.7 million. This gives us an unemployment rate of 11.18%, since unemployment rate is the total number of the unemployed seeking jobs divided by the labour force participation rate. So much for unemployment hitting a 45-year peak. There is no data as comprehensive as the Census, where surveyors literally knock on each household door to get information, as opposed to the numbers provided by the NSSO and CMIE, which extrapolate from thin samples. Several conclusions are worth drawing from these flawed, incomparable and stand-alone sources of data.
First, India’s problem is clearly one of youth unemployment, and this is where the efforts must focus. The Modi government has eased employment under the Apprentices Act, and also allowed all sectors to offer fixed-term labour contracts. But thus far, employers do not seem to have taken the bait, possibly because many of them are still deleveraging and recovering from the double disruptions of demonetization and goods and services tax (GST).
Second, while enterprise data is improving (EPFO, CII, etc.), household jobs data is still dependent on the private sector CMIE and the NSSO, whose 2017-18 report is still to see the light of day. The main reason for the government’s reluctance to release the report seems to be the choice of year (2017-18), which is when the combined effects of demonetization and GST would have played out.
Third, both CMIE and NSSO suffer from one major drawback: they are compiled over months, and this means they do not capture data at a particular point of time. So, if someone is unemployed in January and finds a job in March, they will still show up in the data as unemployed. The only real way to report accurate data is to get the survey completed in one day—or within a week at best. This means putting more feet on the ground, and only the government can afford to do this. The CMIE and other sources of data are essential to keep government data honest, but their usefulness is limited.
