Indian institutions are operating with the same mindset and technology and outlook that the pioneers did 70 years ago. But the world, and technology, has moved on. Not us, though.
There is an old saying—no, it is confirmed that it is not a Chinese proverb—that says “be careful what you wish for, it might come true”. At present, the political opposition has made jobs and unemployment their number one issue to unseat the Modi government. If I were a Congress strategist, I would do exactly what they are doing. Jobs, and the economy, is where the BJP are maybe the most vulnerable. And the perception of vulnerability is enhanced by BJP’s misguided stubbornness to not release the Periodic Labour Force Survey (PLFS) data on the labour market—employment, unemployment, wages, etc. These data were for the period July 2017-June 2018 and is the last official government survey on the labour market. It yielded the result that unemployment, at 6.1%, was at a 45-year high.
Statisticians and economic experts, who should know better and probably do, are also siding with the opposition with the belief (wish!) that the government will be intransigent enough to not release the data and thereby allow the perception to continue that the economy is really as weak as indicated by the PLFS data. This is where being careful what you wish for comes in; once released, the PLFS data is likely to show that their conclusions are extremely flawed, and embarrassingly inaccurate.
In the next few paragraphs, I report on labour market data for 2011-12 and 2017-18. As the reader peruses the evidence, she will realise that the PLFS data, and results, are in a class of their own—they don’t conform to any known economic behaviour. But we won’t know that until the data are released.
While the media (obviously!) has concentrated on the reported 45-year high in the unemployment rate (6.1% in 2017-18 vs the 2.2% observed in 2011-12), it is prudent to interpret, in a detached manner, other economic trends contained in the ‘leaked’ report.
Count the inexplicable anomalies as we list them. First, some background about labour market reports in India. The NSSO believes (erroneously) that the ratios obtained are sacrosanct, but not the survey method on which they are based. For example, to derive the important data on jobs, NSSO documents advise users to use the ratios that they provide (labour force participation rate, worker participation rate, etc) and multiply the ratios by census-based population. Good advice—but how do you believe the building blocks of the survey-based ratios?
For example, first the PLFS reports an urbanisation rate (a ratio) of 2017-18 of 29.3% compared to an estimated 31.2% in the NSSO’s own survey for 2011-12. A decline of 2 percentage points, when the reality (yielded by every other observer) is for an increase in urbanisation; how did this simple cross-check escape the NSSO or its oversight body, National Statistical Commission (NSC)?
Secondly, the estimate for overall population in 2017-18 is off the charts—i.e, too low. The NSSO 2011-12 survey reported a population of 1088 million; in 2017-18, it reports a population that is 14 million less. On a survey/census ratio (since the NSSO and NSC are fond of ratios) basis, the 2017-18 ratio of 82% is the lowest observed for the last 40 years, if not ever. This is both an Indian and world record for population underestimation.
Third, what explains this statistical junk? As lucidly pointed out by Avik Sarkar (bit.ly/2ULHNuA), it is the sampling technique, stupid. Inexplicably, and seemingly without much consultation with outside experts, the NSSO decided to radically change their sampling method—it was to be now based on educational attainment rather than household per capita consumption. It is one of the biggest statistical no-nos to select a sample based on a criteria that affects the behaviour you are trying to explain. Education affects labour market participation—the last thing you want to do is ‘select’ on the basis of education. James Heckman got a Nobel prize in 2000 for obtaining reliable results when the sample behaviourally selects itself. And here we are imposing a selection bias.
Simply put, the PLFS 2017-18 survey is a huge statistical embarrassment and someone should inquire as to how it passed all the statistical checks and balances of the experts. My best ‘recovery’ of what happened—the expert masters felt that the radical change in sampling method would not make much difference to the results.
A large number of households with lower educational attainment (with no member of the household having more than a secondary level of education) were missed out in the PLFS survey. These households are likely to be poorer, and more at work than members of the middle class. Missing them means that you are likely to skew the wage pattern in unbelievable directions. For example, poorer casual workers (those working on daily wages, including agricultural wages) show a big decline in their share in the workforce—a 20% decline from 30.5% of the workforce in 2011-12 to 24.9% in 2017-18. The higher income salaried workers show a large increase in employment share—from 19.3% in 2011/12 to 22.8% in 2017-18—again, a near 20% increase. Such large changes are unprecedented and suggest that, contrary to the unemployment rate jump, the economy is doing very well.
But wait, as there is more to ‘release’—all puns intended. Let us look at what has happened to the real income (wage) increase for the two sets of workers comprising almost half the working population (the rest are self-employed, like the farmers, from whom the NSSO abstains from collecting any income data. [Again, revealing a provincial mindset].
The higher income middle class salaried workers show a cumulative 18% decline in their real incomes. Again, unprecedented. Well, there can be an explanation. Look, everybody is getting educated (the sample selection is based on education), there is excess supply of educated people, and the demand hasn’t kept pace. Ostensibly because of GST and demonetisation. But wait—the real wages of the poorer casual workers, the ones who are out of jobs because of demonetisation, GST, and a lack of demand, have increased by 29%. If true, the PLFS data would indicate that the Modi government has unleashed the most inclusive growth anywhere, and at anytime in human history. There should be a special Nobel prize for this achievement. Forget farm distress—you are not counting income right. All the farmers have become farm labourers, in order to enjoy these high wages. And inequality wallahs, please note—Modi has engineered the highest inequality decline in history, that too in just five years.
The PLFS results can be described in two words—shockingly untrue. The statistical masters who passed the PLFS data report with a ‘good housekeeping seal’ should hang their heads in statistical shame. All the data are presented in the table—all the data are from the much leaked (and vaunted) media reports. Please, will any of the concerned citizens who have voiced shame, censorship of government data, and decline of public institutions, explain the array of evidence presented?
My own personal view is that the PLFS data is a statistical embarrassment to India and the NSSO organisation. We all complain about how institutions have gone down, and there are a myriad complaints about the authenticity, or transparency, of Indian statistical data. The NSSO was a much-respected institution, as was the National Statistical Commission, which was first formed in 2006 under the chairmanship of the late Suresh Tendulkar (I was proud to serve under his leadership for the first three years of the Commission 2006 to 2009).
Sadly, and extremely unfortunately, many, no, make it most, no, make it all Indian institutions, are still operating with the same mindset and technology and outlook that the pioneers did 70 years ago. But the world, and technology, has moved on. Not us, though.
(The author is contributing editor, Indian Express and consultant, Network 18. Views are personal Surjit tweets @surjitbhalla)