Now, all those Ola and Uber operators and Swiggy delivery boys, all the pakora-frying street hawkers, agricultural and industrial labourers, entrepreneurs and of course also the organised sector workers -- all live in households. Since this is a household survey, they are all covered. Unless, of course any of them live on trees.
Is a sample size of 174,000 households adequate to represent the country of 240 million households?
People who are not trained in probability and statistics may wonder how a sample of 174,000 households can give fairly accurate population estimates involving about 240 million households. But it does, and reflects mankind's mathematical genius. This is the method used by developed countries to measure unemployment. And, it is the only method by which any society can measure unemployment.
Magically, statistical theory tells us that the proportion of households in the sample is not important. The size of the sample does matter but only upto a point and then there are diminishing returns to increasing the sample size. What is important in a diverse country like India is to select an appropriately stratified sample and ensure that selection process is random. CPHS ensures these conditions are met.
Why not use a payrolls database that covers millions of jobs instead of a sample of less than 0.2 million households? And, what about the millions of jobs seen increasing in the EPFO database?
There are several problems with the EPFO, NPS and ESIC databases. There are duplications, for example. But, what is more pertinent is that EPFO only covers the organised sector which provides less than 20 per cent of all jobs. So, it cannot provide the complete picture. It is very likely that those Ola and Uber operators and Swiggy delivery boys are not covered under EPFO. Secondly, growth in EPFO is more likely a reflection of formalisation of jobs and not creation of new jobs on a net basis. When people working in unorganised sectors move to formal jobs, they show up in the EPFO database. But, when they do so, they do not join the stock of employed for the first time. They merely moved from an informal job to a formal one.
Developed countries use payrolls database to measure jobs created. That makes sense because they have moved jobs out of the unorganised sectors into the organised sectors -- that maintain a payrolls database. It is foolish to emulate them without first shifting jobs to the organised sectors.
What about the millions of trucks, cars, three-wheelers being produced? Surely, they are creating jobs by the hordes. Why does that not show up in the CPHS results?
This is a typical cost-accountant's thinking. But, an economy is more complex than a factory. Those new truck drivers are mostly internal transfers -- to use an accountant's terminology. Old drivers driving new trucks, old helpers becoming drivers and agricultural labourers becoming truck helpers. Then, 80 per cent of the new cars are the small cars which are mostly self-driven. The only way we can find out whether, on a net basis, we are creating new jobs or merely shifting jobs from one sector to another is to conduct a household survey. And, the household surveys tell us loud and clear that on a net basis we are not creating new jobs.
But when GDP growth is 7 per cent per annum, how can there be no employment growth?
It's high time we now turn this question on its head -- when there is no growth in jobs for several years, how can the real GDP grow at 7 per cent per annum? There is credible evidence from NSSO and CMIE that unemployment has risen and jobs are stagnant. The CSO needs to explain, in detail, its GDP growth calculations in the face of these facts.