How Does Fertility Affect Female Employment? Evidence from Albania

Abstract

This article inspects the relationship between fertility and employment, providing an explanatory mixed-method analysis to gauge their nexus for rural and urban Albania over the 2000s. We instrument reproductive decisions with having two first-born daughters in a 2SLS model of employment. We show that having an additional child negatively influences employment probability for rural mothers, but not for women in urban areas. Heterogeneous analyses show that rural women who lack education or wealth are more dependent on their fertility decisions, whilst there are no relevant differences by occupation types. We further show that the fertility effect is reinforced by the characteristics of the rural household, such as the presence of seniors in the household or if the partner is working. We then examine qualitatively how structural and contextual settings influence women in their decisions. We inspect the experience of rural women and employers from three distinct rural municipalities. Interviews reveal several obstacles that create a gap between employer and employee expectations, resulting in females refusing job offers. The qualitative findings suggest that targeted policies on transport, childcare, and employment services are essential aspects of policy-making to favour rural and distant women’s integration.

Keywords:

• Albania
• female employment
• fertility
• rural labour market

Acknowledgements

The authors thank the IAFFE 2022 Conference participants, Claire Guiraud and Sophie Salomon for useful comments. Quantitative data are available on the DHS website at https://dhsprogram.com/data/available-datasets.cfm, whilst qualitative data are detailed in Expertise France (2021) report. Stata codes used for the analysis are available on request. Any results, opinions, arguments or errors reflect solely the views of the authors, they do not represent the official position of the institutions affiliated with the study or the report, and they do not necessarily reflect the official views of the OECD or its member countries.

Disclosure statement

The authors acknowledge Expertise France for allowing access to their qualitative data from the Expertise France (2021) ‘Needs Assessment Study’ report. The authors report there are no competing financial interests to declare.

Notes

1 For an extensive review, see Bhalotra and Clarke (2022).

2 Aaronson et al. (2014) inspect the extensive and intensive margins of a fertility transition model. As the price of investing in children reduces with an additional child, their model predicts increased investment in children and a decline in the chances of having an additional child. Moreover, the literature suggests that having more than one child results in greater inter-generational support for senior parents (see Oliveira, 2016 for an empirical application to China).

3 The NAS report was conducted between February 2020 and February 2021, by a team led by Expertise France in close relation with the National Agency for Employment and Skills, NAES, its local offices in three selected municipalities and the three local municipalities’ teams. The NAS report also benefits from a desk review and of informant interviews at the municipal level performed between 2020 and 2021.

4 The employer identification was conducted with the support of the NAES local offices and municipalities seeking for a balanced representation of company size and sector (agriculture and agricultural-processing, manufacturing, services industry, food production, hospitality and tourism.)

5 For instance, ambitious women may opt to have fewer children due to higher opportunity costs, resulting in a positive correlation between ambition and employment probability but a negative correlation with the number of children (β₁ overestimated). Conversely, some women may desire more children but face financial constraints, which are eased when they are employed, leading to an underestimation. In an emerging economy like Albania, urban women may lean towards the first scenario with diverse job opportunities, higher wages, and easier access to contraceptives. In contrast, rural women might be influenced by stronger gender norms and lower income, potentially leading to an underestimation.

6 Angrist and Evans (1998) and Agüero and Marks (2011) show that the bias from overestimation is reduced with an increase in education levels and for low-income countries. For extremely low-income countries and low-educated women, they find that OLS could even underestimate the childbearing pressure on the labour supply. Furthermore, Ebenstein (2009) compares the effect of childbearing costs on employment between Taiwanese women and US women, using the preference for male children as an instrument. He finds that, contrary to previous literature, OLS underestimate the real coefficient. He explains this result by the magnitude of the first stage coefficient: the lowest the education and stronger preference for boys, the largest the number of children and the largest the negative effect of children on employment.

7 The LATE conditions of the 2SLS estimator imply that fertility decision is to be interpreted only at the intensive margin (how many children to have, given the cost of investing in them). This means that the external validity of the analysis applies only to the sample of women who already have made the decision to have a child (Angrist & Evans, 1998). This subset of the population however has a broad external validity, corresponding to 80.71% of women aged 20+ in Albania.

8 We obtain similar results if we restrict the sample to women residing in rural areas (Supplementary material, Table A2).

9 In Table A5 (Supplementary material) we show that the results hold if we transform the independent variable – the number of children – in logarithms.

10 The coefficients of the number of children for the urban sample should be interpreted with caution because of the low strength of the first-stage estimations.

11 Table A13 (Supplementary material) compares employment means by type over time, showing between 2008 and 2018 a drop of 41.5 percentage points of women working for family members in rural areas. This decrease is only 9.7 p.c in urban areas. Likewise, more women work stably in rural areas (+17.1 pc, against +6.3 in urban areas) and they are more likely to be paid (+32.3 pc in rural areas, +0.3 pc in urban areas).

12 Figure A3 confirms the descriptive statistics of Table A13 (Supplementary material) by regressing employment outcomes with a time and rural dummy. The Figure shows that women in rural areas have overall worse employment conditions than in urban areas, but the gap is largely reduced in 2018. The convergence in the job market between cities and the countryside could explain the null results for rural areas in 2018 of Table A10 (Supplementary material). The opportunity costs of childbearing increase as the job market improves, thus rural women are less likely to be affected in their labour market decision in 2018 than in 2008 with an additional child.

13 Empirical evidence from Albania shows that male and female labour in rural areas acts as substitutes, sustaining the findings from the focus groups (Miluka, 2013).