It would be proper to wait for the next round of the PLFS (2018-19) that is under progress and compare its findings with the results of the PLFS (2017-18) to make a more correct assessment of the employment rate in the country
By Avik Sarkar
Last month saw a series of discussions related to employment numbers reported in the leaked Periodic Labour Force Survey (PLFS). I wanted to understand the underlying reasons and the validity of the numbers quoted in the media. I did not have access to the findings of the leaked report, so I looked at the survey methodology available in public domain and compared it to the methodology of the last Employment-Unemployment Survey (EUS) of 2011-12—both of which have been conducted by the National Sample Survey Office (NSSO) of the ministry of statistics and programme implementation. I further compared the methodologies across these surveys to ground realities in India, to identify the possible reasons for the low numbers that have been reported.
I found there to be a stark difference in the methods used to choose survey households. The PLFS is based on education level of households and the EUS is based on expenditure (urban) or livelihood (rural) of households. Any direct comparison of the survey results of the PLFS with the earlier EUS would lead to erroneous inference about the employment scenario. Further, the sample chosen in the PLFS was not quite representative of the underlying Indian population in terms of the achievement of secondary education leading to lower estimates for the population, labour force participation and employment. And here I will show you how that is the case.
The EUS, which was last conducted during the 68th round of the NSSO for the duration July 2011 to June 2012, is a comprehensive survey providing a complete scenario of the labour force, across sectors like agriculture, industry, services, etc, in both rural and urban areas.
In any survey, a sample of locations are chosen judiciously to represent the entire country. For the EUS 2011-12, the selection of locations for First Stage Units (FSU) in the sample, urban and rural classification was made based on the data from Census 2001 and each town with a population of more than 10 lakh was represented as a separate group in sample locations. For the Second Stage Strata (SSS), the criteria for choosing households in both the rural and urban areas was household affluence, as shown in the accompanying table.
In rural areas, 50% of chosen households are those with principal earnings from non-agriculture-based activities. There was good representation of households with at least one member engaged in non-agricultural activities; for example, employed in formal/informal sector.
For urban areas, the Monthly Per Capita Expenditure (MPCE) available from the previous rounds of the NSSO household surveys forms the basis for selecting households. The MPCE considers the realities of that particular area based on the previous round survey (by the NSSO) leading to the criteria for selecting households. Household expenditure is a good proxy to ascertain earnings for households through employment (formal or informal). The sample also has good representation of the middle class engaged in gainful employment-related activities.
The PLFS, also conducted by the NSSO between July 2017 and June 2018, is the first to focus on a detailed overview of the labour market in India. It provides continuous update on the employment situation in India (quarterly for urban and yearly for rural areas). This survey has, for the first time, used the Computer Assisted Personal Interviewing (CAPI) method to capture data—a great step towards technology adoption. The sample size of various NSSO surveys are comparable and may assumed to be in line with the last EUS survey of 2011-12.
For the PLFS, the selection of locations in the FSU for urban and rural areas was done as per Census 2011 classification. Each census town with a population over 15 lakh was formed into a strata (group) for further sampling. The remaining towns were classified into three strata based on population: less than 50,000, 50,000 to 3 lakh, and 3 lakh to 15 lakh. The representative samples were then selected from each of these groups. Keeping in mind the population growth and constraint of similar sample size, it was a good decision to move the town population threshold from 10 lakh to 15 lakh.
But there is a major change in the criteria for the selection of households in the SSS for both rural and urban areas, based on the number of members in the household having general education up to secondary level (10th standard).
At first glance, the household selection criteria for the PLFS seems aspirational in nature, as the choice of households is dependent upon the education level of the household instead of the earlier criteria of affluence/expenditure. It is true that mostly formal or better-playing employment is linked to the education level of the household members and this move by the PLFS is really aspirational in nature. However, it made me wonder if this selection of sample based on the level of education is really representative of the underlying population of India.
I started by looking at the data on Census 2011 and found that the percentage of people above secondary level as of 2011 is quite low, at 21.51%, which goes further down to 15.3% for the rural population but has a healthy number of 35.24% for the urban population. Not all informal or daily wage employment requires more than secondary level education. A healthy literate level of 63.07% implies that a large portion of the population has basic literacy, which is what is required for daily wage employment.
Household-level secondary education was not readily available from Census 2011 or other data sources to my knowledge. For this, unit level survey data of the EUS of the NSSO 68 Round in 2011-12 was used to derive the estimates of the number of households with zero, one, two or more than two members having general education level above secondary level (see table).
Thus, it can be seen that there are 66.42% of households (75.61% rural and 46.20% urban) with no family members with general education above secondary level. Whereas only 25% of households have been sampled based on these criteria, leading to a huge mismatch between the reality and the samples drawn. People from these households are mostly daily wagers or engaged in informal employment, which would also show lower employment estimates.
These numbers provided a good view on the education level of the people in India and showed that the stratification criteria used in the PLFS is not quite aligned to the secondary and above secondary levels of education in the country.
This under sampling, leading to under-representation of such households, is leading to lower estimates of the people for this group in labour force participation and employment rate. This is what is falsely coming through—lower employment numbers due under-representation of the population.
The percentages of households for the urban area, though far from the sample sizes are closer to the reality in urban areas compared to rural areas. It would be proper to wait for the next round of the PLFS (2018-19) that is under progress and compare its findings with the results of the PLFS (2017-18) to make a more correct assessment of the employment rate in the country.
-The author is Officer on Special Duty and Head Data Analytics Cell, NITI Aayog. Views are personal