What Drives Female Labour Force Participation? Comparable Micro-level Evidence from Eight Developing and Emerging Economies

Abstract

We investigate the micro-level determinants of labour force participation of urban married women in eight low- and middle-income economies: Bolivia, Brazil, India, Indonesia, Jordan, South Africa, Tanzania, and Vietnam. In order to understand what drives changes and differences in participation rates since the early 2000s, we build a unified empirical framework that allows for comparative analyses across time and space. We find that the returns to the characteristics of women and their families differ substantially across countries, and this explains most of the between-country differences in participation rates. Overall, the economic, social, and institutional constraints that shape women’s labour force participation remain largely country-specific. Nonetheless, rising education levels and declining fertility consistently increased participation rates, while rising household incomes contributed negatively in relatively poorer countries, suggesting that a substantial share of women work out of economic necessity.

Acknowledgements

We are grateful to Veronica Frisancho, Esther Heesemann, Lisa Hockel, Stephan Maurer, Bruno Witzel-Souza, and participants at the 26th IAFFE Conference, the 2019 AEL and WIDER Development Conferences, and seminars of the Universities of Goettingen, Hannover, and Stellenbosch for comments and suggestions. For assistance with the Tanzanian data, we thank Novati Buberwa, James Mbongo, and Titus Mwisomba from Tanzania’s National Bureau of Statistics. For assistance with the Indonesian data, we thank Krisztina Kis-Katos, Christoph Kubitza, and Robert Sparrow. Friederike Schilling provided excellent research assistance. Supplementary materials are available with this article at the {\it Journal of Development Studies} website. Data and codes to replicate the results are available upon request.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplementary Materials are available for this article which can be accessed via the online version of this journal available at https://doi.org/10.1080/00220388.2020.1790533.

Notes

1. Some authors have also predicted that the labour deregulation policies of the 1980s would lead to a feminisation of the labour force (e.g., Standing, 1989, 1999).

2. For reviews see, among others, Elson (1999); Heath and Jayachandran (2018); Klasen (2019).

3. See also Kabeer (1997) and Elson (1999) on norms and restrictions on women’s employment.

4. Aaronson et al. (2017) instrument fertility with twin birth (Rosenzweig & Wolpin, 1980) and sibling sex composition (Angrist & Evans, 1998). Using infertility shocks as a different source of exogenous variation for 26 developing countries, Agüero and Marks (2011) find no effect of fertility on mothers’ labour force participation. Priebe (2010) argues that, in poor settings, child costs push women into the labour market; as fertility declines, this type of distress-driven FLFP falls. The author shows causal evidence of this mechanism operating in Indonesia.

5. Ganguli et al. (2014) analyse census data from 40 countries, presenting both macro- and micro-level results. At the micro level, the authors show that if the education gender gap, the marriage gap (LFP gap between married and single women), and the motherhood gap (LFP gap between mothers and childless women) were to close everywhere, a large unexplained gender gap in participation rates would still remain for most countries. However, Ganguli et al. (2014) assume that education and FLFP are linearly related. As we will show in this paper, the shape of the education-participation relationship is nonlinear in some countries.

6. Until 2008, India and Vietnam were also low-income countries, according to the Word Bank’s classification.

7. Note that South Africa’s GDP per capita grew much faster between 1995 and 2014, which is the period shown in Figure 1.

8. For Jordan, we consider both urban and rural areas because information on urban status is not available from the 2008 and 2014 surveys. In any case, more than 80 per cent of Jordan’s population lives in urban areas, in the period considered, according to data from the World Bank’s World Development Indicators.

9. For example, in its 2000 and 2006 rounds, Tanzania’s Integrated Labour Force Survey only recorded agricultural income in urban areas. Other well-known practical complications are unmeasured product variety and quality.

10. For simplicity, we refer to currently married/cohabiting women as married women.

11. Please note that these rates include participation in informal work and the surveys go to great lengths to adequately capture informal activities.

12. See Figure A2 for the FLFP rates by education level and year for each of countries in the dataset.

13. Notice that our data for Jordan only starts in 2006, thereby hiding the rapid education expansion of previous decades. Of the top 20 countries in the world with the largest increase in educational attainment between 1980 and 2010, three are included in our dataset: Brazil (rank 5), Jordan (rank 9), and Bolivia (rank 13) (Campante & Chor, 2012, Table 1).

14. We obtain similar results with a logit model.

15. For Indonesia and Tanzania, it was not possible to derive meaningful proxies for ethnicity or religion that were also comparable over time.

16. For South Africa, however, we use an alternative definition of household head education, since the head is not identifiable from the data. As a best approximation, we use the maximum educational attainment of any adult married man of age 18+, with an additional dummy whenever no such household member exists.

17. We also include survey wave dummies whenever there are several survey waves per year (as in South Africa after 1995, Jordan, and Tanzania).

18. These are: provinces in South Africa, Indonesia, and Vietnam, states in Brazil and India, governorates in Jordan, departments in Bolivia, and regions in Tanzania. As a robustness check, we remove as much spatial heterogeneity as possible by adding primary sampling unit (PSU) fixed effects to the model. PSU information is not available for all surveys. For Brazil, Bolivia, South Africa and Tanzania, we find similar results with either PSU or regional fixed effects. For India and Indonesia, adding fixed effects at the second highest subnational level – districts in India, regencies (Kabupaten) and cities (Kota) in Indonesia – also produces similar results.

19. The Indian and Indonesian surveys are representative at the second highest subnational level; this is the level of aggregation used for the regional employment share variables. For the remaining countries, we use the highest subnational administrative level to aggregate the employment shares.

20. In principle, one could pursue a country and year-specific IV approach, but the resulting local average treatment effects would be hard to interpret in a unified comparative framework, as the population of compliers would vary across settings and IVs.

21. See Klasen and Pieters (2015, pp. 460–461) for a discussion of the lack of robustness in estimates of own-wage effects in India, as well as a more detailed discussion of the challenges involved in such estimations.

22. The small size of the income effect should be interpreted with caution. The earnings variable available from the Jordanian surveys is very roughly measured: it is the mid-point of five earning brackets. We thus suspect the average marginal effects of household income to suffer from attenuation bias.

23. There is no obvious relationship between the U-shaped FLFP-education link and the feminisation-U hypothesis discussed earlier. Only if there were an identical, stable and universal U-shaped FLFP-education link in all countries (and education was seen as a proxy for income), could this deliver a feminisation-U in FLFP. This condition is clearly not met.

24. This finding resembles patterns that are taking place in OECD countries in the past decades. In the US, for example, Blau and Kahn (2007) and Heim (2007) show that income elasticities of married women labour supply have plummeted since the 1980s. Similarly, Bargain et al. (2014) find extremely small income elasticities for 17 European countries during the period 1998–2005.

25. See Priebe (2010) for causal evidence of this mechanism in Indonesia.

26. We loosely follow the notation of Fortin, Lemieux, and Firpo (2011), who review decomposition methods relevant to labour economics.

27. The covariate contribution is also often referred to as composition effect, endowment effect, or explained term.

28. The expression holds as an exact equality for logit models that include an intercept, whereas it holds very closely for probit models (Fairlie, 2006).

29. See Fairlie (2006) for more details.

30. For each decomposition, we draw 1000 random samples. In addition, at each sample draw, the ordering of the variables in the sequence of counterfactuals is randomly determined. This addresses the issue of path dependence: since individual contributions depend on the distributions of all other covariates, the ordering of the variables matters for the final result.

31. For point estimates of the decompositions, see Tables A18-A21.

32. We sorted similar covariates into variable groups: own education dummies are grouped as ‘Own education’, household head education dummies appears as ‘Hh head educ’, number of children by age group and sex appears as ‘Children’, age and age squared appear as ‘Age’, ethnicity and/or religious variables are grouped as ‘Pop group’, and regional and survey wave dummies appear as ‘Region dummies’ and ‘Survey waves’.

33. For Tanzania, results depend on the choice of counterfactual. Using the 2000 coefficients, the covariate effect accounts for 36 per cent of the LFP reduction between 2000 and 2014. Increasing household incomes drive the negative covariate effect, being partly offset by the positive effect of rising female education. In 2014, the negative average marginal effect of income shrinks by two thirds relative to 2000. As a result, the total covariate contribution becomes positive when weighted at 2014 coefficients.

34. In addition, we decompose the FLFP change in South Africa for the full post-apartheid period: 1995–2014.

Participation rates of urban married women rose substantially from 58.5 per cent to 68.1 per cent between 1995 and 2014. We find that women’s labour market characteristics account for around 70–74 per cent of this increase (Table A22). Rising education, declining fertility, and a relative increase in the share of black women (in urban areas) were powerful drivers of participation in this period.

35. Table A23 shows how the education and social group variables are created for each country.

36. See Figure A5. The only exception is Tanzania where the weighted FLFP rate is much higher than the unweighted rate. Accordingly, we interpret the results for Tanzania with caution.

37. For point estimates of the between-country decompositions, see Tables A24 and A25.

38. The FLFP rates in the paragraph are calculated without survey weights.

Additional information

Funding

This work was carried out with financial support under the Growth and Economic Opportunities for Women (GrOW) initiative. GrOW is a multi-funder partnership with the UK Government’s Department for International Development, The William and Flora Hewlett Foundation, and Canada’s International Development Research Centre. The views expressed herein do not necessarily represent those of the IDRC or its Board of Governors.