It is necessary that we examine the methodology of generating the data, going into sectoral details and original sources.
Studies based on the KLEMS data are being widely quoted to counter claims of poor job creation in the country. This database has been developed as part of an international project and has a very respectable parentage, nurtured by scholars from the Delhi School of Economics and ICRIER since 2009 and housed in the Reserve Bank of India since 2022. Hence, it is necessary that we examine the methodology of generating the data, going into sectoral details and original sources.
The KLEMS database consisting of data on capital (K), labour (L), energy (E), material (M) and services (S), currently available for the period 1980 to 2024, is meant to provide a “measurement tool to monitor and evaluate productivity growth at the industry level as well as the aggregate economy”. It uses data from different rounds of the employment-unemployment surveys (EUS), the periodic labour force surveys (PLFS), the National Account Statistics and the Annual Survey of Industries. In the absence of yearly data from the National Statistical Office, the available data are used as benchmarks and interpolated for other years.
As per the methodology, the EUS and PLFS data are used to determine the sectoral distribution of workers by the usual principal and subsidiary status (UPSS) for four groups — rural male, rural female, urban male and urban female. Since the surveys do not provide the absolute number of workers, the estimated worker-population ratios (WPR) for the four groups from the survey are multiplied by the total population. Population for the survey years can either be interpolated using Census numbers or taken from the population projections of the National Population Commission under the Ministry of Health and Family Welfare (MoHFW).
In the methodology segment of the RBI report, it is noted that for 2017-18, 2018-19 and 2019-20, the all-India figures for the employed persons are taken from the Economic Survey 2021-22. For 2020-21 onwards, population projections by the MoHFW are used. But these projections are available for males and females only, and consequently, a uniform growth is applied for projecting populations in rural and urban segments. The worker numbers are then distributed among the industry groups, considered in KLEMS, according to their shares in employment as in PLFS.
Importantly, the population figures projected by MoHFW are on the higher side due to a sharp decline in fertility rate during the period from 2010 to 2020. This implies that the total labour and workforce, obtained through the multiplication of the projected population with the WPR, would be overestimated. The estimated rural population would be higher also because it is assumed to grow at the same rate as the urban population while, empirically, the rate in rural areas is much less. Since WPR in rural areas is higher than in urban areas, the total employment generated in the years of the twenties would work out as higher than the actuals.
The forgoing explanation clearly shows that the RBI does not produce any employment figures independently. The UPSS-based WPR is used on a projected population to obtain the numbers. There is a significant drop in WPR, as per UPSS, from 2011-12 to 2017-18, as we shift from EUS to PLFS and the KLEMS assumes there is no problem of temporal comparability. The WPR, however, has gone up significantly for rural women with some increases also for the other population segments, in the subsequent years. These WPR values applied on somewhat higher population estimates, as discussed above, would produce inflated employment numbers.
In the KLEMS database, employment in agriculture increased from a near stagnant 20 crore before 2018-19 to 25 crore in 2022-23. Correspondingly, the service sector employment went up from 17.2 crore to 20.2 crore. Manufacturing employment grew from 5.5 crore to 6.3 crore.
The number of workers would go up systematically due to the population increase and the methodology of projection even when the WPR remains the same. Similarly, manufacturing employment goes up although the ratio of manufacturing to total workers goes down, as per the PLFS. It is important to note that the employment data includes those with subsidiary employment, implying the inclusion of persons having a tenuous connection with work. A large majority of them are engaged as unpaid family workers. Using EUS/PLFS data along with the projected population to claim employment generation would thus be misleading, without any reference to the nature and quality of work.
A study by economists at the SBI compares the projected total employment based on the Annual Survey of Unincorporated Sector Enterprises (ASUSE) data with the numbers available in the RBI KLEMS database. The ASUSE survey covers a subset of all unorganised enterprises and excludes those in construction, the corporate sector and government, besides those registered as factories and cooperatives. The survey estimated the number of persons employed in such enterprises to be only 10.96 crore. This is being inflated to claim that total employment in 2022-23 is 56.8 crore, close to KLEMS data. This needs to be investigated.
Employment in enterprise surveys indicates a position in the enterprises. It is not easily relatable to information on individuals, collected in household employment surveys, that are considered superior for employment data. Independent estimates from these two sources do not match for reasons well known. Similarly, data on the registration of MSME units in the Udyam portal usually does not imply new job creation, nor do the monthly changes in EPFO subscription mean additional employment generation.
Considering the methodological limitations of these data, it is surprising that claims should be made of a rapid growth of employment, that too of decent jobs.
Kundu is Professor Emeritus at L J University, Ahmedabad and Mohanan is a former member of National Statistical Commission
