Robust job growth, not fake news

My concern here is not to forecast the next election. Rather, my goal is to test the veracity of the claim that the Indian economy is in a terrible shape and that despite economic reforms like demonetisation and GST (or because of them), it is fighting to create jobs. The number most frequently touted is that the economy needs 8 to 12 million jobs a year, to keep unemployment and social tensions at bay. It is “clearly” producing very few jobs, and hence economic growth, and job creation, is the faultline in the Indian economy.

At the time of the national elections in May 2014, India’s macro-economy was in a shambles. CPI inflation was running at 9.4 per cent, and had averaged 7.8 per cent over the previous 10 years of UPA rule. The previous five years, 2009-2013, had witnessed an average inflation rate of 9.7 per cent per annum. GDP growth had also slowed down — from 7.7 per cent in UPA-I to 7.1 per cent in UPA-II. Not surprisingly, Modi’s election campaign centered around the macro-economic malaise and corruption. The campaign emphasised growth, lower inflation, employment generation and the promise of “acche din”. Four years later, a legitimate question arises — how has the reality lived up to the promise (and expectations)?

Regarding the macro-economy, there is no question that the situation is considerably better today. In the last fiscal year (2017/18), CPI inflation is 6 percentage points (ppt) lower at 3.7 per cent; GDP growth is just 0.4 ppt lower at 6.7 per cent.

But what about jobs? Are we in a period of jobless growth, and if so, Modi’s popularity, and vote, is likely to be considerably dented. Going by reports in the media, this is exactly what the political opposition believes. They believe that jobless growth is a new Modi phenomenon. Is this the reality, or fake analysis?

I also believe that job growth is very important for votes, but am willing to take a historical perspective before passing judgement. However, there is a sense in which Modi is held to a higher standard than other politicians — he must deliver job growth, because he promised to do so. This higher expectation is flattering (to Modi). But how many of you know that during India’s peak growth period, UPA-1 and 2, job growth was at a very slow rate of 0.6 per cent per annum — so labour productivity was at a China-beating pace of over 7 per cent an annum for seven years. That is not the reality — hence, the only reasonable conclusion is that the NSSO data is understating job creation, that is employment growth, during the UPA years. By how much is a subject for research.

In order to properly assess what has happened to employment in 2017, we want to base our assessment on two separate age-groups — 15-24 and 25-64. The reason for this separation is that when you have increased enrolment in education by the young (and especially young women who are catching up with, and exceeding, the enrolment of men), then the employment (and labour force) definition should include the fact that you are attending school or college.

We employ two sources of data — the recent provident fund (EPFO) data (as first popularised by Soumya Ghosh and Pulak Ghosh) for the age-group 15-24, and the CMIE employment data for the 25-64 age group (see table). The EPFO data refers to the net additions of employees in the formal, provident fund payment sector; however, not all net additions are new employment. But for the young, given the difficulty, and attraction, of a formal sector job, it is unlikely that the 18-21 group will have any job-hoppers. For the six-month period, September 2017-February 2018, total net additions in the 18-21 age-group was 1.1 million; in the 22-25 group (college graduates?) net additions were lower at 0.8 million. Even after allowing a reasonable amount of job-hopping (20 per cent), the EPFO data suggests that in 2017, 3 million jobs were added in the 15-24 age group.

For the 25-64 age group, the CMIE data suggests that total employment creation in 2017 was a robust 12 million. Total job-creation in 2017, 15 million. While this estimate may not be the “truth”, it is unlikely to be far way from the truth. But as the FT quote suggests, beware of trolls.

The table also reports female labour-force participation (LFPRF) rates, an important ingredient in estimates of employment for India and selected countries in the world. This reveals that there are major problems with the CMIE data (and the problems are such as to bias the estimate of employment downwards). The CMIE data suggests that employment opportunities are so few in India that one half of the population, the women, have almost completely withdrawn from the labour market. For the 15-64 age group, CMIE’s estimate of LFPRF is 12.5 per cent — only one out of every eight women are offering themselves for work. And that the LFPRF is the lowest in the world — and declining! In 2015, the lowest LFPRF in the world was Iran, 14.4 per cent; the next lowest, Saudi Arabia, at 21.4 per cent. NSSO data (for 2011/12) had India’s LFPRF at 27.2 per cent. But CMIE has India’s LFPRF, circa 2017, at 12.5 per cent.

If you believe that, I have some snake oil to sell. Employment survey questions are the easiest to get answers to. Age, sex, and whether I was working yesterday or not. Contrast that with over 200 questions (including how much salt you bought) routinely asked in NSSO questions. A respondent can forget, lie, get confused, when you ask her questions about her consumption or income. But how can we forget our sex, age, and whether we worked yesterday? We can’t, and we don’t.

The CMIE survey is also a large sample survey (more than 5,00,000 respondents), and the survey is done every month. How can such a low LFPRF be obtained? The only manner in which such a large outsized anomaly can be obtained is via incorrect weights. (For the not so statistical, weights is the blow up factor to go from the sample to the population.)

It appears likely that 15 million jobs were created in 2017, not much different from the Vajpayee average of five years. Between 1999-2004, 11 million jobs were created each year (weekly status definition of employment, closest to the daily status CMIE definition; the usual status definition gives an increase of more than 12 million a year). For reasons unknown, but deserving investigation, the UPA-I and II era (2004 through 2011) reveal an employment gain of less than 4 million a year.

Any non-partisan interpretation of the data would suggest that in economic terms, the Modi period 2014-2018 has delivered the best macro-economic performance ever in India — several economic reforms, steady GDP growth (and growth that can, and should, accelerate to an 8 per cent+ potential), low inflation (shoo — don’t tell the MPC that!), and robust job-creation. But woman does not live by bread alone. While social tension in the form of riots has definitely declined post 2014, there has been a worrisome increase in “communal tensions”. Every political party in the world has its fringe — and the BJP has both a left fringe (with economic policies similar to the communists and the Congress left) and an ultra-nationalist and an ultra-religious “right” fringe — those who would kill a Muslim in the name of the cow. How to rein in the fringes is the biggest challenge for India, and PM Modi.