Surjit bhalla contested cmie’s labour growth estimates by suggesting it had got the female labour participation rates wrong. cmie explains why it is right.
Firstly, I thank Surjit Bhalla for pointing out an error in my comparison ofCMIE’s Consumer Pyramids Household Survey (CPHS) estimates of labour marketstatistics with those from the National Sample Survey Organisation (NSSO)surveys. However, even after using the data he presents, it is easy to seethat the fall in female labour participation rate seen in the CPHS data isbroadly in line with the trends seen in NSSO data.
Besides, a crucial comparison made by him to magically double the number ofwomen employed is not kosher. We discuss this later below. CPHS estimates are based on the status of respondents on the day of the survey. This ensures almost no recall problem. The unambiguous nature of a daily recall and the power of fast frequencysurveys make CPHS an exceptionally powerful survey. It makes it possible forus to conduct natural experiments as we could in the case of demonetisation. NSSO surveys do not have a comparable measure of daily status. And, their surveys are not fast-frequency.
Their measures are of principal and secondary activities based on the status over a 365 day period and a 7 day period. Their results are of annual observations.We, therefore, do not generally try and make comparisons between NSSO andCPHS. But, comparisons are sometimes necessary and also useful. Here are somecomparisons. Consider the following—CPHS has a sample size of 178,000 households incomparison to NSSO’s 120,000. CPHS is necessarily conducted using hand-heldGPS-enabled devices while NSSO surveys are done using paper with no trackingdevices. CPHS data is validated 100% in real-time by supervisorswhile the survey is being executed. NSSO validations are offline and mostlypost-hoc. CPHS results are released within a few hours of the completion ofthe survey while NSSO survey results are released after 6-8 months of thecompletion of the survey.
All the record-level data and the source-code to derive the estimates isavailable for anyone to check the estimates. Given the much greater superiority of the CPHS machinery compared to the NSSO as described above, if there is any difference between CPHS and NSSO estimates, the benefit of the doubts, if any, should go to CPHS and not NSSO. Nevertheless, one must not expect too much of a difference between the broadall-India level estimates of large surveys, and both the NSSO and CPHS areeasily large surveys.So, what does the NSSO survey tell us about female labour participation rateif we use Surjit Bhalla’s estimates? He says that if we use the current daily status estimates, female labour participation rate fell from 31.1% in 2004-05 to 25.1% in2009-10 and further to 23.3% in 2011-12. So, it fell 7.8 percentagepoints in 7 years. This translates into a fall at the rate of 1.11 percentagepoints a year.
CMIE’s CPHS says that the female labour participation rate fell further atthe rate of 1.39 percentage points per year between 2011-12 (NSSO) and mid-2016. Mid2016 is the period when CMIE completed its last survey just before demonetisation. Demonetisation was a game-changer and therefore we must distinguish the period before and after this huge economic shock. Female labour participation rate fell from 23.3% in 2011-12 (as per NSSO) to 16.37% during May-August 2016 (as per CPHS).
The direction of change is the same as shown by Bhalla using the NSSO’s survey but, the rate of fall increased. But, the greater rate of fall is only expected since the economy slowed down in the latter period. We know that the period of 2011-12 through 2016-17 was one of a slower growth of less than 7% per annum compared to the preceding 7 years when the economy grew at more than 8% per annum. Thus, an increase in the rate of fall of labour participation—particularly female labour participation rate—is not unexpected.
So, CPHS estimates of female labour participation rates are comparable with the NSSO data as churned by Surjit Bhalla. While I stand corrected on the use of the specific indicator, using the data suggested by Surjit Bhalla does not change the fact that NSSO and CMIE give estimates of female labour participation rates which are broadly comparable. Ergo, we mustaccept that female labour participation continued to fall post 2011-12.Note that the increase in the rate at which female labour participation rate fell between 2011-12 and 2016-17 is not dramatic. A rate of fall of 1.11 percentage points per annum rose to 1.39 percentage points per annum. The dramatic fall comes after 2016, i.e., after demonetisation.
This is when the female labour participation rate fell by a massive 4.93 percentage points injust one year. It fell from 16.37% in the wave (as one round of a panel survey is called) of May-August 2016 to 11.44% in the wave of May-August 2017.Male labour force participation rate fell by a lower 2.5 percentage points,from 74.76% to 72.33% during the same period. Given that CPHS female labour participation rates are broadly comparable with Surjit Bhalla’s estimates derived from unit level NSSO data till mid 2016, he must accept the steep fall after the May-August 2016 wave.
This is because CPHS is a longitudinal panel survey. It repeatedly surveys the same households with the same questions. It also uses the same weights and estimation procedures.It does not help if we try hard to not recognise the problem. No country has given itself a shock of demonetisation like we gave ourselves and, so, it is not surprising that no country has ever observed the kind of withdrawal of labour from labour markets like we observed after demonetisation.
Given this huge shock of demonetisation and its evident impact on the labour markets, it is not correct to assume that the labour participation rate has remained unchanged since 2011-12. But, without any explanation, Surjit Bhalla makes this assumption and once again magically (or, simply) doubles the number of women employed!And, he does this by selecting convenient ratios. He states “FLFPR increased in 2014 by 250 bps over the 2011-12 estimate, and it increased by 40 bps in the second successive drought year of 2015-16”.
But, he comes to this conclusion by using Labour Bureau estimates for 2013-14 and 2014-15 and the NSSO estimate for 2011-12. Instead, if he had used LabourBureau estimates for all the years under comparison (25.4% in 2011-12 and 23.7% in 2015-16) he would not see an increase, but he would have seen a fall.And then, his doubling of female work force would vanish! This is not about politics. It is not about elections. The stories and accompanying pictures in The Big Picture feature of Sunday Express were not of a political rally. 2.37 crore applicants were trying their luck for 2 lakh jobs in the Indian Railways. Many had started losing hopes merely by seeing the level of competition for jobs. Many leave the labour markets when they lose hope. I believe that a much bigger problem of unemployment will be inherited by the new government.
By- Mahesh Vyas, MD and CEO of Centre for Monitoring Indian Economy