Labor force participation of women left behind in Tajikistan

ABSTRACT

This paper examines the impact of male migration on the labor force participation of the women left behind in Tajikistan. Studies from many countries show that when men migrate, female labor force participation decreases and this is largely explained by the income effect from remittances. Our study challenges this finding. Using panel data from 2007, 2009 and 2011, we find that, in Tajikistan, migration has no significant effect on the number of hours that women work. We use panel data which allow us to control for unobservable heterogeneity, rather than the cross-sectional data used by others. We analyze several countervailing factors that may have neutralized the income effect, such as the need to substitute for the missing labor in the household. We also find that women work more when the household has a farm, regardless of the presence of a migrant in the household.

KEYWORDS:

• Female labor force and employment
• migration
• gender
• Tajikistan

JEL CLASSIFICATION:

• J21
• R23
• J16

Acknowledgments

We would like to thank Stephan Klasen, Susan Steiner, Simon Lange, Rahul Lahoti and Inmaculada Martinez-Zarzoso for critical comments at various stages of research. This paper also benefited from helpful comments from the editors and two anonymous referees, and conference participants at the 26th Annual Conference of the International Association for Feminist Economics, the 2016 International Sociological Association RC28 Conference, the Life in Kyrgyzstan Conference 2015 and the 2015 Human Development and Capabilities Association Conference.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. The focus of this paper is on outward international migration rather than the very limited domestic (rural-urban) labor migration. According to the 2007 TLSS, only around 9% of internal migration (defined as living in a different place from where one was born) was due to employment or looking for employment.

2. According to the 2007 Tajikistan Living Standards Survey, others migrate to Kyrgyzstan, Kazakhstan, Ukraine and other former Soviet states, and less than 1% migrate to countries outside of the region.

3. A report by the OSCE (2012) argues that the end of subsidies from Moscow in 1991, combined with the outbreak of civil war from 1992 to 1997, led to a severe economic depression. It also led to the deterioration of social services and education, and slowly drew women back to the home. Independence also revitalized Islam, and with it, greater gender inequality (Falkingham, 2000). Women are also considered primary domestic caretakers, and working outside of the home is often discouraged and even stigmatized. Even in 1991, only 29% of the economically active female population were in the workforce (Falkingham, 2000). Haarr (2007) adds that, of the women who work outside of the home, 81% work on collective farms and earn little or nothing.

4. An oblast is an administrative geographic region in Tajikistan. There are five oblasts: Dushanbe, Regions of Republican Subordination (RRP), Sogd, Khatlon and Gorno-Badakhshan (GBAO).

5. We use the active working-age population instead of the 15–64 years of adult age, because using the latter may introduce selection bias at both ends of the spectrum: the younger population may still be in school, and the older population may be retired.

6. We also ran our main analysis on an unbalanced panel sample, which yielded consistent results with the balanced panel.

7. Household size is the number of members physically living and sleeping in the household at the time of the survey. Therefore, while migrants are associated with the household, they are not included as part of the total household size.

8. In total, 15.5% of all working people reported working in unpaid family positions.

9. The hours worked over the past 14 days refers to the time immediately preceding the survey and most likely provides a strongly representative trend of the hours worked by women. The surveys were also conducted in the fall between September and November, ensuring that seasonality is consistent and that the results are not driven by seasonal variation, such as the extra work needed during harvest season.

10. Employment includes all forms of economic activity, including self-employment and unpaid work in a family business. Wage earner status asks the respondent whether they received any compensation (wages, salary, cash payments) from the employer, from the business or for their work. Because of the large informal economy, many people are employed but not wage earners.

11. Several descriptive studies (Falkingham, 2000; Haarr, 2007; Harris, 2005) explain that Tajikistan is a highly patriarchal society where brides move to their husband’s homes, and her in-laws control each instance she leaves the home, and whether she works. Women are expected to be submissive, and often verbally, psychologically and physically abused by members of the in-laws if she challenges family structures.

12. Here, it is important to note that our dependent variable is censored, as some women work zero hour. In such cases, the Tobit model is considered a more consistent estimator than OLS (Amemiya, 1973); however, it cannot be used for fixed effects models. While Honore (1992) provides an alternative semi-parametric estimator for censored regression models with fixed effects, it cannot accommodate instrumental variables. As such, although recognizing these shortcomings, we are limited by the available econometric tools.

13. The purpose for testing a random effects model was to account for women living in households with long-term migrants. For these households, migrant status remains constant, warranting the use of a random effects approach to capture differences between households. At the same time, there are likely many omitted variables that correlate with the predicted variables that we only control in the fixed effects model.

14. The official exchange rate of local currency to US dollars is based on a monthly average of the 2007 period, $1 = 3.44 TJS (World Bank, 2014).

15. We use consumption rather than income since we seek to understand the effect of poverty, and consumption may be more reliable in separating poor from non-poor households. Also, the 2011 survey omits questions about consumption, and we therefore keep 2009 consumption levels constant for 2011, assuming limited variation in consumption across years. We anticipate that consumption does not change as rapidly as income, and believe that it is a reasonable basis for which to impute the missing data for 2011.

Additional information

Funding

Funding from the German Research Foundation RTG Globalization and Development (1723) and DFID/IZA Growth and Labour Markets in Low Income Countries Programme is also gratefully acknowledged.

Notes on contributors

Sophia Kan

Sophia Kan is a postdoctoral researcher at the Department of Economics, University of Göttingen. Her main research field is development economics.

R. Emre Aytimur

R. Emre Aytimur is a lecturer at the Department of Economics, University of Manchester. His main research field is political economy.