Rates Are Deceptive

EVER SINCE THE National Sample Survey Organisation’s (NSSO) Periodic Labour Force Survey was published, there has been a lot of discussion and debate on the state of employment in India. The NSSO does these large sample employment surveys once every five years. Perhaps one should say roughly once every five years. Accordingly, such surveys were done in 2004-2005, 2009-2010, 2011-2012 and 2017-2018. When I used the expression ‘a lot of discussion and debate’, I meant the 2017-2018 survey. Often, the NSSO surveys aren’t comparable with each other. That’s why statisticians and economists ignore 2009-2010. It wasn’t quite comparable. Any analysis of trends is done with 2004-2005 and 2011-2012 and now with 2017-2018. Some statisticians and economists have questioned the 2017-2018 survey, others have pointed out it isn’t quite comparable with preceding surveys. Those criticisms are legitimate, but I am not going to get into those in this column. There are people who have written a lot about the unemployment rate in the country, using 2017-2018 and earlier surveys. Personally, I don’t think the unemployment rate, however measured, tells us anything in a country like ours. I would much rather go by employment trends, not rates. Recently, Laveesh Bhandari and Amaresh Dubey have written a paper titled ‘Emerging Employment Patterns of 21st Century India’. Bhandari works for Indicus Foundation and Amaresh Dubey, Jawaharlal Nehru University. Those who work on such topics will immediately recognise these names. They have done a lot of work on labour and employment (and sundry other matters) in the past and have a lot of credibility. This paper will soon be published. It is just that I have had a preview.

Bhandari and Dubey have tried to render 2004-2005, 2011-2012 and 2017-2018 comparable, as best as they can. I am sure people will read the paper, when it is published. I am going to report some numbers and trends I found interesting. I am sure there are plenty of others I will miss. How can one possibly summarise a longish paper in a column? First, total employment increased from 412 million in 2004-2005 to 433 million in 2011-2012 and increased again to 457 million in 2017-2018. Over a seven-year period from 2004-2005 to 2011-2012, this translates into 3 million additional jobs a year. Over a six-year period from 2011-2012 to 2017-2018, this translates into 4 million additional jobs a year. If these numbers are correct, we aren’t creating something like 9 million jobs a year. But it isn’t that no jobs are being created. While 4 million jobs a year aren’t enough, this phenomenon of growth not leading to sufficient jobs isn’t new. For the period under consideration, it has plagued us since 2004-2005. Second, the rate of employment growth has been less than half the rate of population growth. Between 2004-2005 and 2011-2012, the annual rate of population growth was 1.68 per cent, while the annual rate of employment growth was 0.72 per cent. Between 2011-2012 and 2017-2018, the annual rate of population growth was 1.77 per cent, while the annual rate of employment growth was 0.89 per cent. As a share of population, total employment was 37.67 per cent in 2004-2005, 35.27 per cent in 2011-2012 and 33.47 per cent in 2017-18. To state the obvious, there is no denying that there is an employment problem and because of population growth, this has become a little bit more serious.
Third, this is primarily a rural sector problem. Between 2004-2005 and 2017-2018, urban employment increased from 110 million to 152 million. Over a 13-year period, that’s an annual average employment growth of 3.2 million. However, between 2004-2005 and 2011-2012, rural employment increased from 302 million to 305 million. This is a virtual stagnation. Fourth, there are clear male/female differences. Male employment increased by 38.1 million between 2004-2005 and 2011-2012 and by 21.6 million between 2011-2012 and 2017-2018. There has been a relative slackening of male employment growth between 2011-2012 and 2017-2018.

However, overall, between 2004-2005 and 2017-2018, male employment increased by 59.7 million. Female employment shows a different pattern. It declined by 16.8 million between 2004-2005 and 2011-2012 and increased slightly by 2 million between 2011-2012 and 2017-2018. For rural women, the decline in employment is across all age groups. However, for urban women, the decline is for those 24 years and below. Sometimes, the rural/urban difference is artificial, a point I have made earlier. But accepting that classification, we need a plausible hypothesis about declining female workforce participation rates. Notice these aren’t trends over the last few years: they go back to 2004-2005. Perhaps enrolment in education might be an explanation for decline in female workforce participation rates for women under 24 in urban areas. But what about rural women? My intention in this column is not to probe hypotheses and explanations. I will therefore stick to the factual, as reported in the Bhandari and Dubey paper.

Fifth, between 2004-2005 and 2017-2018, the bulk of employment creation has been in the 24-plus age group. The total increase was 78 million, of which 69 million was in the 25-59 age group. On the other hand, youth (15-24) employment declined by 28 million during this period. Why? Is that because of enrolment, or non-enrolment, in higher education? Sixth, there are clear links with education and this has policy implications. In 2004-2005, illiterates and those who had not completed primary education accounted for 48.9 per cent of employment. In 2017-2018, illiterates and those who had not completed primary education accounted for 31.1 per cent. Between 2004-2005 and 2017-2018, there was a decline of 59 million in the employment of those who were illiterates and had not completed primary education. But there was an increase of 104 million in the employment of those who had completed at least primary education. At an anecdotal level, that premium on education is known. What we now have is empirical substantiation and the scale of this shift surprised me.
Seventh, organised sector employment increased from 37 million in 2004-2005 to 64 million in 2017-2018, with the share in employment increasing from 8.9 per cent to 14 per cent. Unorganised sector employment increased from 153 million in 2004-2005 to 218 million in 2017-18, with the share in employment increasing from 37.1 per cent to 47.7 per cent. What has dropped dramatically is the share of agri-cropping, from
53.2 per cent of employment in 2004-2005 to 38.1 per cent in 2017-2018. Total employment in agri-cropping declined from 219 million in 2004-05 to 174 million in 2017-2018. More on this a little later.

Eighth, self-employment (or family employment) accounts for a smaller share, down from 53.9 per cent in 2004-2005 to 50.8 per cent in 2017-2018. What drives insecurity in the employment market is the increased share of non-contractual employment. The share of non-contractual employment increased from 18.8 per cent in 2004-2005 to 31.8 per cent in 2017-2018. Again, one could sense this shift, but the scale surprised me. One must be careful not to equate organised with contractual and unorganised with non-contractual. There are quite a few non-contractual jobs in the so-called organised sector too. Ninth, as shares of employment, agriculture, forestry and fishing have declined from 54.27 per cent in 2004-2005 to 41.62 per cent in 2017-2018. The absolute decline in this segment is of 34.2 million. The share of manufacturing has increased marginally, but there are no major shifts within any of its sub-categories. The share of construction increased from 6.16 per cent in 2004-2005 to 11.95 per cent in 2017-2018. Absolute employment in construction was 253 million in 2004-2005 and 54.1 million in 2017-2018. The share of services has also increased sharply. Tenth, in 2017-2018, the cropping sector employed 47.5 million fewer people than in 2011-2012. The bulk of the decline in employment in agriculture has occurred for crops like sugarcane, cotton, wheat and rice. All of these are crops where there are significant subsidies and significant government intervention. We know independently of this paper that these sectors haven’t done too well growth-wise. Now we know that they haven’t done too well employment-wise also, an understandable consequence. If a sector does badly when there is government intervention, that’s a good reason to think about the nature of that intervention. While I generally knew about the cropping sector, this specific identification (in the study) of sugarcane, cotton, wheat and rice surprised me. Eleventh, amongst major states, between 2004-2005 and 2017-2018, absolute employment has declined in Jharkhand, Karnataka and Tamil Nadu. These are caused by declines in the rural sector. Let me repeat: these are declines in absolute employment. In states like Jharkhand, Karnataka and Tamil Nadu? I wouldn’t have expected it beforehand.

There can always be quibbling about what Bhandari and Dubey have done and about their methodology. If you have alternative assumptions, perhaps the numbers will change a bit. But I don’t think the trends will change, even if you use an alternative methodology, unless of course you choose to dismiss the 2017-2018 survey altogether. Hence, I think people who are interested in employment and job creation should read the paper. As I said, it is about to be published. I only had a preview.

educationemploymentgrowth rateNSSOstatistics