Employment, or the lack of it, has been a major issue of debate among economists and policy makers in India in recent years. Recently, Prime Minister Narendra Modi claimed that India created “eight crore new jobs in the last three to four years”. The Prime Minister was using data from the India-KLEMS database hosted by the Reserve Bank of India (RBI). As per this database, the total number of workers in India rose from 56.6 crore in 2020-21 to 64.3 crore in 2023-24, that is, a net rise by 7.8 crore workers. Tailing this claim, the research team of the State Bank of India (SBI) published a validating report that claimed a match between the total number of workers in the India-KLEMS database and in the NSSO’s Annual Survey of Unincorporated Sector Enterprises (ASUSE), 2022-2023.

What lent an element of surprise to these claims was the rise in the number of workers over the two COVID-19 years and after. According to the International Labour Organization (ILO), the employment-to-population ratio between 2019 and 2023 was stagnant, if not falling, in East Asia, South-East Asia and the Pacific. Given such trends elsewhere, analysts have had serious methodological and empirical suspicions in relying on the India-KLEMS database to posit an outlier status for India in employment creation.

The India-KLEMS project began as an academic exercise financed by the RBI in 2009. From 2022, the RBI hosts the database. KLEMS stands for Capital (K), Labour (L), Energy (E), Material (M) and Services (S). It is a framework used to measure industry-level “total factor productivity” (TFP), which is considered by mainstream economists as a measure of the efficiency of all the inputs to produce a unit of output.

In other words, the objective of the KLEMS framework is not to produce data on employment. The employment figures are merely inputs into the database’s modelling framework. Further, the the RBI does not directly collect data on any input, including employment, that enter the India-KLEMS database. It sources sectoral data on employment, input usage and output from official sources, including the Central Statistics Office, Census of India, Annual Survey of Industries and the Periodic Labour Force Surveys (PLFS). It is amusing then that data sourced by the RBI from other official sources, and used as inputs to estimate TFP, are portrayed as “RBI jobs data” to make political statements on employment generation in the economy.

The method in India-KLEMS

India-KLEMS borrows employment data from the PLFS, but not as absolute figures of the number of workers. The PLFS provides only the share of workers in the population, or the Worker Population Ratio (WPR). To obtain the number of workers, the WPR is multiplied with the total population. This is where the problem begins, as there is no official population figure for India after 2011.

Also read: PM Modi speaks about an increase in job avenues for youth, data shows otherwise

To obtain a population estimate for the intercensal years, demographers typically interpolate population numbers from the last available Census. But here, India-KLEMS adopted a strange solution. The estimates of population in 2017-18, 2018-19 and 2019-20 were borrowed from the Economic Survey (ES), 2021-22. The ES projected these populations by assuming that population growth rates between 2001 and 2011 were the same for the years after 2011. The WPRs were multiplied by these population projections to obtain the number of workers for each corresponding year.

But for the years between 2020-21 and 2023-24, India-KLEMS used a totally different source and method. It used population projections from 2011-2036 published by the Ministry of Health & Family Welfare (MoHFW) in 2020. From the Census figures of 2011, this publication arrived at annual population projections using demographic models that factored in the Total Fertility Rates (TFR) and the mortality rates reported in the Sample Registration System (SRS) of 2017. The simple question is why the India-KLEMS database did not use the MoHFW’s population projections for all the years after 2017-18. It appears that while the RBI adds new estimates to the series after 2022, it does not correct or update older estimates published before it began hosting the database.

There are two major issues here. Firstly, population projections from the ES and the MoHFW disregard the sharp fall in fertility rates in India over the last decade. The replacement TFR is canonically assumed to be 2.1 children per woman. However, results from the most recent National Family Health Survey (NFHS) show that India’s TFR had fallen to 2.0 in 2019-21. Similarly, a 2024 study published in The Lancet argued that the “reference TFR values in Bangladesh and India are projected to decrease below 1.75 by 2026 and 2027, respectively”. These falls in TFR are not considered in the population projections in the ES or by MoHFW.

Secondly, the population projections in and by the ES and the MoHFW are not available separately for rural and urban areas. So, the India-KLEMS managers took the national sex-wise populations, assumed population growth rates for rural and urban populations and obtained separate rural and urban population projections. However, it is well-known that India’s rural population is growing at a slower rate than the urban population. Assuming uniform growth rates for both is likely to lead to an overestimation of the rural population. For these two reasons, the population figures with which the WPRs were multiplied by in India-KLEMS, and the number of workers obtained thus, are likely to be overestimates.

Shifts in employment structure

When PLFS data are readily available for analysis, one fails to understand the need to depend on India-KLEMS for a temporal analysis of employment. PLFS data show that India’s WPR fell from 38.6% in 2011-12 to 34.7% in 2017-18, and then rose to 41.1% in 2022-23. The rise in overall WPR was largely due to a rise in the rural female WPR, which rose from 17.5% in 2017-18 to 30% in 2022-23. WPRs for other population segments also rose, but not as much as for rural women.

These changes are the basis for two claims of the government: one, that crores of new jobs were generated during and after the pandemic; and two, that this phenomenon was gender friendly as women occupied the jobs vacated by men in the rural workforce.

Both the claims are flawed. The rise in rural female WPR was largely due to an increase in unpaid forms of self-employment among rural women in agriculture. Between 2018-19 and 2022-23, the share of rural women employed in agriculture rose from 71.1% to 76.2%, and the share of rural women who were self-employed rose from 67.8% to 78.1%. Among female workers in agriculture, the share of those who were employed purely on a subsidiary basis (that is, those who worked only irregularly, and on a minor scale) rose from 15.6% in 2018-19 to 27.7% in 2022-23. And within all subsidiary employment in agriculture, the share of unpaid family work was about 65% in 2022-23.

But a rise in unpaid subsidiary work can show up as higher WPRs for women. When these rising WPRs are multiplied on with an increasing projected population, we obtain a steady rise in the total number of workers. Even if the WPRs were constant, one would have obtained a rise in the number of workers because of the increase in the projected population. This is what we see in the projected workforce figures in India-KLEMS. In short, there was little expansion of meaningful and paid employment in India after 2017-18. The departure of men from agriculture hardly changed the status of rural working women.

The ASUSE comparison

This leaves us with one outstanding matter — the SBI report’s claim that the number of workers in India-KLEMS and ASUSE 2022-23 broadly match. The ASUSE covers only unincorporated non-agricultural establishments in manufacturing, trade and other services. Apart from agriculture, it explicitly excludes a range of manufacturing and trading establishments from its sampling frame. The number of workers in the unincorporated non-agricultural establishments — defined and covered as above in ASUSE — was 11 crore in 2022-23. The SBI report, however, estimates the total number of workers from ASUSE as 56.8 crore, and claims comparability with the figures in India-KLEMS.

Clearly, the SBI report assumed a certain number of workers employed in sectors not covered in ASUSE — such as agriculture, construction, registered factories, corporate sector, government and cooperatives — using other household surveys that employ different concepts and methodologies. It then added those numbers to the number of workers in ASUSE to arrive at the inflated estimate of 56.8 crore workers. But there is no scientific basis for such an indirect method, that too to make an inane and motivated validation.

To sum up, data from India-KLEMS, which was designed for very different purposes and uses questionable methods, are being used to drive a specific political narrative on employment generation. But the real culprit in this episode is the Government of India, which has refused to organise the new decadal Census till date.

The absence of accurate population figures has led analysts and institutions to use many erroneous projections based on heroic assumptions. Consequently, we end up needlessly politicising economic debates and restricting the space for reasoned studies of important trends in the Indian economy.

P. C. Mohanan is former member, National Statistical Commission and R. Ramakumar is Professor, Tata Institute of Social Sciences, Mumbai.