Introduction

In 2019, the National Statistical Office undertook India’s first-ever national time-use survey, the results of which have recently been published (GoI 2020). Time Use in India 2019 (henceforth, TUS19) provides information on time spent by men and women in rural and urban areas of all States in different activities during one full day. From such a survey, we should be able to gauge the time spent on a variety of activities – including employment, learning, rest, and cooking and cleaning. In this note, I focus on what we can learn from TUS19 about women and economic activity in rural India.¹

The nature of economic activity in agrarian and informal economies – particularly among women – is such that work is often “scattered, intermittent, seasonal, temporary or unstable” (Hirway and Jose 2011). The borders between economic activity (that is, within the System of National Accounts (SNA) or production boundary) and other activity are blurred. Women are engaged in multiple activities, and regular labour-force surveys can make mistakes when assessing their activity status. One of the main objectives of a time-use survey is to record the details of all activity in order to better identify economic activity as well as household maintenance and care activities that are essential for social reproduction (or what is termed “extended SNA” in the literature). Data from a time-use survey taken together with labour-force survey data can be used to examine variations in time-use between seasons and across activities to understand the nature of unemployment. In sum, we look to a time-use survey to give a more accurate account of women’s work and employment in an economy such as that of India.

TUS19 is based on a survey of 138,799 households, 59 per cent of which were households from rural areas. In each household, all persons above the age of six were interviewed, a total of 447,250 persons. The recall period was 24 hours, from 4 a.m. the day previous to the survey to 4 a.m. on the day of the survey. The report uses a nine-fold classification of activity, following the International Classification of Activities for Time-Use Statistics (ICATUS) of 2016, which, in turn draws on the ILO (2013) resolution. For each activity, three indicators are estimated: (i) the percentage of persons participating in the said activity, (ii) average time spent per activity for those engaged in the activity, and (iii) average time spent per activity averaged over all persons. Information is also collected on the household and individual (called “contextual variables” in the literature). These data have some noteworthy gaps: for example, there is no information on wage or payment received for any activity.

Presentation of Data

Data from the survey are available in two forms, digitized unit-level data and data presented in the report under review. A full assessment of the database and findings from the data will require that scholars study the unit-level data. The report under review consists of more than 2000 pages, mainly tables. It contains no discussion of findings from the survey, or even of the reasoning behind the organisation of the tables.

There are problems also with how data are presented in specific statistical tables. For example, Statement 5 on the participation rate (number of participating people as a ratio of total persons) in different activities shows that among all rural girls and women (six years and above), 19 per cent engaged in employment, 25 per cent in production of goods for own final use, 19 per cent in learning, and 82 per cent in unpaid domestic services (p. 26). What can we learn from this table? We cannot gauge how many children, engaged primarily in schooling, also assisted in productive work or domestic work. We cannot gauge the extent to which adult women were engaged in employment and domestic work at the same time. Reporting data aggregated over all ages can be misleading in other ways. For example, according to Statement 6, 422 minutes (over seven hours) were spent by each participant in learning and only 317 minutes (over five hours) in employment! The former figure is evidently influenced by the number of hours spent in schools and educational institutions. An explanation for the latter figure could be that many respondents worked in the said activity (“employment”) intermittently or seasonally. Aggregate statistics as presented in the tables mentioned are of little use to social scientists or policymakers.

Concepts and Definitions

TUS19 does not clearly define “participation” in an activity. The title of the first table on participation (Table 4, p. 859) is “Percentage of persons aged 6 years and above in different 3-digit/2-digit/1-digit activity code of TUS activity in a day as major activity (considering only the major activity of the time slots).” Each respondent reported their activity during a time slot of 30 minutes or one hour. For each time slot, a maximum of three activities could be reported, with the respondent deciding which activity of the three was the “major” activity. In short, the choice of “major” activity is a subjective response. The report does not specify whether Table 4 includes participation in any “major” activity even if only for a single half-hour slot or if a minimum time (say, an hour) is required in order to count as participation. As per the new ILO (2013) recommendation, a minimum of one hour is taken to define participation in economic activity.

TUS19 has a nine-fold classification of activities as proposed by ICATUS (2016). In this classification, work for pay and profit (wage employment and self-employment) is included in Division 1 activities and work for own or home use is included in Division 2 activities. A single respondent may move between these two categories across different seasons and years (Swaminathan 2020). In the work schedule of a rural woman worker, there is no hard and fast line between own-use production and production for profit (self-employment). For example, a bad harvest could reduce the sale of rice to the market; in such circumstances, what may have been production for the market at the time of sowing (Division 1) becomes production for own use at the time of harvest (Division 2).² Further, when employment opportunities shrink, women may spend more time in production for own use and, conversely, when employment opportunities expand, women may spend less time in production for own use. In other words, participation in Division 1 or Division 2 activities may be independent of a woman’s choice, and depend on the availability of employment (Division 1). Furthermore, in regular labour-force surveys, participation in Division 2 activities is included within the production boundary (or economic activity). Division 2 includes, for example, construction for own use and gathering firewood and fuel, in addition to agriculture, forestry, fishing, and mining for own final use (p. 43).

Another limitation of the ICATUS classification is that there is very little disaggregation of activities that are important in rural areas, such as crop production and animal rearing. In the pilot time-use survey conducted by the National Sample Survey Organisation in 1998, crop farming had sub-activities such as ploughing, weeding, and transplanting, whereas animal rearing had sub-activities such as milking and grazing. ICATUS (2016), however, has only one three-digit activity code for all crop production: “growing of crops for the market in household enterprises.” Similarly, there is only one three-digit code for livestock rearing: “raising animals for the market in household enterprises.” The use of aggregate categories such as these lead to the exclusion of many activities done by women.³ The time-use survey was an opportunity to understand the gender division of crop production and animal rearing in the contemporary period – that opportunity went unutilised.

TUS19 identifies a person as a worker or unemployed or non-worker based entirely on self-reporting of usual principal status, that is, the major activity pursued during the reference period of one year (see Table 3 of the questionnaire schedule, p. 2131). We know that women are likely to report themselves as out of the labour force when they have no employment available. We also know that when employment opportunities are limited or seasonal, women may not report it as their principal activity status. Surprisingly, TUS19 did not even ask about subsidiary status during the reference year, further reducing the likelihood of capturing the actual participation of women in the labour force.

TUS19 has identified workers based on a reference year (usual principal activity status). In my view, the interpretation of data for one day (24 hours) would make more sense in the context of a reference week. We conducted a time-use survey among women in two villages of Karnataka, collecting data on a 24-hour basis for seven days consecutively in two seasons. Using the major time criterion, we then identified persons engaged in economic activity using a weekly status definition of employment (economic activity referred to all activities contributing to SNA). The result was striking: while there was large variation in participation in economic activity among women between the lean and harvest seasons, almost all women were in the workforce in the harvest season (Swaminathan 2020). The TUS19 schedule, however, does not have any information on activity status in terms of a reference week.⁴

Findings on Work and Employment Among Girls and Women (15–59 years)

Results in Table 4 show that 23.2 per cent of rural girls and women participated in employment and related activities (Division 1 in the one-digit ICATUS code) on a normal day, and 27.8 per cent participated in production of goods for own final use (Division 2). While a woman can participate in both Division 1 and Division 2 activities on the same day, the extent of overlap can only be checked with unit-level data. For now, if we count economic activity as participation in either Division 1 or Division 2, then at least 27.8 per cent and at most 51.2 per cent of women (assuming zero overlap between those who participated in the two activity groups) engaged in economic activity on the reference day. In 2017–18, according to the Periodic Labour Force Survey, work participation for rural women aged 15 and above, as per principal and subsidiary usual status, was only 25.5 per cent.

In the 15–59 age group, on average, a woman spent 74 minutes on Division 1 activities and 35 minutes on Division 2 activities on the reference day (Table 8, page 1279–80). Altogether, time spent on economic activity (Division 1 plus Division 2) was less than two hours a day, even when all activities in a time slot, irrespective of whether it was a major activity, were included. To anyone who has spent any time in rural India, these numbers are absurdly low and raise questions about the quality of data.

Another set of results are disaggregated for those in the labour force (that is, working or unemployed) and those not in the labour force (such as students or retired persons). Among those in the labour force, 67.9 per cent participated in Division 1 activities and 32.7 per cent in Division 2 activities (Table 15, page 1384–5). Of the unemployed, 12.6 participated in Division 1, and 21.6 in Division 2; among those not in the labour force, 6.4 per cent participated in Division 1 and 21.9 per cent in Division 2 activities. Further, among the self-employed, 61.2 per cent participated in Division 1 and 42.2 per cent participated in Division 2, whereas among the regular-wage or salaried employed, 81.9 per cent participated in Division 1 and 16.7 per cent in Division 2.

These results support the view that when women obtain wage employment, they spend less time in production for own use (Division 2) and vice versa. In short, these findings lend support to my argument that the separation of activities into Division 1 and Division 2 activities, as made in ILO (2013) and now in ICATUS (2016), is neither easy nor useful in a rural economy.

TUS19 has results disaggregated by whether the activity was paid or unpaid. As only activities in Division 1 are counted as paid activities, it is not surprising that, among girls and women aged 15–59, around 21 per cent participated in paid activities (Table 9, p. 1328 and Table 12, p. 1355). In other words, around one-fifth of the relevant population reported paid employment. This observation is consistent with our understanding that there is a lack of suitable employment opportunities for women in rural areas.

Lastly, results are provided for activities categorised as part of SNA and non-SNA production. No less than 86.6 per cent of rural women categorized as workers as per usual principal status participated in SNA production when only the major activity was taken into account. Each participant spent 348 minutes or 5 hours and 48 minutes (almost six hours) in SNA activity. Of the unemployed, 30 per cent participated in SNA production and spent 2 hours and 40 minutes a day on such activity. This is a finding of interest though reported almost at the end of the Report (p. 1754).

Summing Up

With respect to understanding economic activity among rural women, a critical drawback of TUS19 lies in the concepts and definitions used, and in the subsequent design of questionnaire.

First, the TUS19 has used the respondent’s subjective answer on usual principal status to identify workers. This is certainly going to result in an under-reporting of women workers in the survey. Further, neither was information on subsidiary activity status recorded, nor was any information gathered on activity status during the reference week.

Secondly, the classification of activities (ICATUS 2016) that has been used in TUS19 is not suitable for capturing women’s work (or men’s) in rural India. The two most important activities in Division 1 – crop cultivation and livestock rearing – have not been further disaggregated. And, as I have argued, women move between Division 1 and Division 2 activities depending on a host of external factors. For both these reasons, there are likely to be errors in reporting data on time use in Division 1.

Serious conceptual problems in TUS19 affect the quality of survey data, as well as the scope for further analysis.