Several recent studies have documented that female labour force participation rate (FLFPR) in India has not increased over the last two decades. Estimates from the National Sample Survey data (1984-2012) show a declining trend in rural areas, while the urban FLFPR has remained flat. This is a puzzling phenomenon because one would expect women’s participation in the labour market to rise during a period when the country witnessed economic prosperity, declining gender gap in education, and falling fertility rate.
This has led to the issue being widely researched and discussed. On the one hand, the discourse has focussed on identifying the causes behind this stagnation/decline in FLFPR. The main arguments outlining the demand- and supply-side reasons have been succinctly discussed in Stephan Klasen’s I4I post. On the other hand, some studies have looked into the issue of accurate measurement of women’s work, indicating that women’s involvement in economic activities may not be adequately captured in the existing surveys (Deshpande and Kabeer 2019).
The dynamic nature of employment participation
In this context, an important aspect that remains largely unaddressed in the current literature is the fact that LFP is dynamic, that is, individuals enter and exit the workforce at various points in time. This is especially relevant for women who participate in multiple short-spell activities in the labour market. Hence, to properly understand the nature of women’s work, it is important to follow their activities over time. This requires information on the same individuals at multiple time points, that is, panel data.1
A majority of the existing studies have analysed aggregate trends and their determinants using repeated cross-sectional data2 where it is not possible to investigate which individuals are entering or exiting the workforce. In new research, we study the dynamics of employment status of working-age women in India using nationally representative data from the India Human Development Survey (IHDS), which collected information on the same individuals in 2004-05 and 2011-12 (Sarkar, Sahoo and Klasen 2019). On account of being a panel dataset, it allows us to estimate the rates and the determinants of labour market entry and exit of women in India.
It is important to distinguish a continual/persistent employment status (either employed or unemployed) from transitory employment because these have different policy implications. For example, those who are continuously out of the labour force may need to break social norms to enter the labour force, while those who are already in employment may need policies to support them to continue the employment.3
Besides, there can be intertemporal dependence in LFP decisions, that is, participation in current period may depend on the participation status in the previous period. Taking up employment is likely to involve costs of searching for the appropriate job and – for women – dealing with the cultural barriers that they face in their families and in society. On the other hand, being employed may empower women – it may raise their aspirations, preference for independence, or expected consumption standards. Thus, the processes driving the decision to enter the labour force may not be the same as those determining the decision to exit.4
Employment entry and exit rates
To estimate employment transitions, we consider the sample of women in the 25-55-year age group in 2005 and follow-up on them in 2012. We define the rate of entry as women who became employed in 2012 as a proportion of all women who were not employed in 2005. Similarly, exit rate is defined as women who dropped out of employment by 2012, as a proportion of all women who were employed in 2005.
In 2005, around 50% of women in the sample were employed.5 We find that 21% of the initially employed women have exited employment over the next seven years. In contrast, 90% of men in the sample were initially employed in 2005 and only 6.7% of them are no longer employed by 2012. This shows that women not only have a lower participation rate, but also a higher exit rate than men. Among the women who were not employed in 2005, only 25% were employed by 2012, while the entry rate for men was 52%. Therefore, over the period of seven years between the two rounds of the survey, women were three times more likely to exit employment and half as likely to enter employment, relative to men. The figures are especially stark for urban areas where only 24% of women were employed in the baseline, the exit rate was 29%, and entry rate was 13%.
Factors affecting dynamics of women’s employment status
Our analysis reveals that women experience substantial transitions in their employment status, suggesting a much lower labour market attachment than men. We also identify various determinants of women’s employment entry and exit, and test whether the factors which drive the entry decision are also the same factors driving the exit decision.6
Using the panel aspect of the data, we not only look at the effect of the baseline characteristics, but also the effect of factors that are changing over time.7 For instance, having a new-born child between the two rounds is associated with a 3 percentage point increase in the probability of exit. The exit rate also increases if an elderly member (above 65 years of age) moves into the household. This suggests that women may leave their jobs to meet childcare and elderly care needs in the household. Interestingly, if the mother- or father-in-law cohabitates, then women are less likely to leave their jobs. The in-laws may share the household responsibilities and thus enable women to continue working. However, none of the above factors are found to play any role in the entry decision of women.
We find that a household’s financial affluence strongly affects women’s entry and exit decisions. If assets or income of other members in the household increase, then women are less likely to enter and more likely to exit employment. This suggests that women are considered as secondary earners in the household – they participate in the labour market only when there is a need to augment the household income. On similar lines, we find higher entry and lower exit rates among women from households belonging to disadvantaged castes and where male members have low levels of education. Local economic development also encourages entry and reduces exit, possibly by enhancing labour demand, especially in the southern states.8
As a policy-relevant factor, we explore the role of the Mahatma Gandhi National Rural Employment Guarantee Act (MNREGA).9 The intensity of MNREGA implementation significantly reduces women’s exit rates from the workforce overall, but it has no effect on women’s entry. Since MNREGA provides unskilled manual labour work at minimum wage, it may not attract women who are away from the workforce and not used to such work. However, by expanding employment opportunities, it does seem to help those who are already in the workforce to continue participating.
In addition to MNREGA, several other factors, such as own education, household wealth/change in assets, household size, education of male members, change in the number of elderly members, and new births, have different effects on entry and exit. Thus, our analysis indicates that the entry and exit decisions are not symmetric.
The increasing importance of analysing transitions
The two rounds of data collected by IHDS are seven years apart. Therefore, our research analyses transitions that are fairly long term. What happens in the shorter term? We utilise data from the 2017-18 Periodic Labour Force Survey (PLFS), which is a panel dataset, making it is possible to estimate transitions across quarters in the same year for urban areas. We find that about 24% of women (vis-à-vis 92% of men) are in the labour force in a given quarter. Out of these women, over 10% exit in the next quarter, while the corresponding figure is only 1% for men. Among the women who exit, about 70% are likely to remain out of the labour force even in the following quarter.10
Therefore, both short- and long-term employment exit rates among women are quite significant. With all the disruptions caused by Covid-19 in the labour market, the issue of exit from and return to employment is being paid attention to by researchers. Our study highlights that even in normal circumstances, women’s attachment to the labour market is much lower than that of men – they are more likely to exit, and less likely to re-enter. Low FLFPR is indeed a major problem, but it is also important to consider women who break the barrier to enter the labour market and then leave at a high rate.
- Panel data are data that measure the same set of observations (individuals in this case) repeatedly across time.
- Repeated cross-sectional data collect information on different sets of individuals at various points in time; thus, how the status of a particular individual or group changes over time cannot be observed.
- Policies such as parental leave benefits or flexibility in working hours may directly help employed women to continue their job, while these policies may have lesser effect on women who remain out of the labor force due to social norms.
- Labour economists have referred to this phenomenon as the “lack of symmetry” in employment participation decision (Long and Jones 1980).
- A woman is considered employed if she reported working in salaried employment, casual wage earning, business, or family farm for more than 240 hours in the survey year.
- The identified determinants of employment entry and exit are individual characteristics such as age, marital status, education, number of children; and household characteristics such as asset, income, male education, social status, etc. We use appropriate econometric techniques to consider the possibility that women who are employed may not be comparable with those who are not employed in the initial data. We also address the problem of attrition as some women from the original sample could not be followed up in 2012.
- By considering the temporal changes of employment and its determinants, our analysis also accounts for individual-level differences in unobservable characteristics that are time-invariant.
- We measure local economic development using growth rate in night-time luminosity in the districts.
- MNREGA guarantees 100 days of wage-employment in a year to a rural household whose adult members are willing to do unskilled manual work at the prescribed minimum wage. We measure intensity of MNREGA implementation using district-level, labour-related expenditures.
- A set of households in the PLFS data were visited more than twice, therefore, it is possible to observe whether individuals who exited have returned to employment. However, IHDS is a two-period panel; thus, it is not possible to distinguish ‘entry’ from ‘return to employment’ in the analysis using IHDS data.
Soham Sahoo is Assistant Professor at the Centre for Public Policy, Indian Institute of Management (IIM) Bangalore.
Sudipa Sarkar is a Research Fellow at the Institute for Employment Research, University of Warwick.
This article was first published by the Ideas For India (I4I).