Twins, family size and female labour force participation in Iran

ABSTRACT

Despite the remarkable increase in women’s education levels and the rapid fall of their fertility rate in Iran, female labour force participation (FLFP) has remained low. Using the instrumental variable method, this article estimates the causal impact of number of children on mothers’ participation in the labour market. It finds that having an extra (unplanned) child would only reduce female participation rate for low-educated mothers and mothers with young children, thus having no causal impact on most mothers’ participation. This result explains why the rapid decline in fertility rates did not increase female participation; rather, other factors should be at play. It hence moves us a step forward in explaining the puzzle of FLFP in Iran. Policy implications are discussed.

Abbreviation: FLFP: Female Labour Force Participation; LFP: Labour Force Participation

KEYWORDS:

• Female labour force participation
• fertility
• Iran
• twins
• instrumental variable

JEL CLASSIFICATION:

• J13
• J22
• O53

Acknowledgments

The author wishes to thank Ragui Assaad, Insan Tunali, Meltem Dayioglu-Tayfuras, as well as the editor and the anonymous reviewer. All the remaining errors are mine.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

¹ Participation rates for urban women aged 20–65 have remained steady at about 15%, and rural participation has been fluctuating at rates around 22%. For a detailed discussion of these rates and more, see Majbouri (2010, 2015, 2016).

² See World Bank (2004) for a thorough discussion of low FLFP in the Middle East and North Africa.

³ Xie (1997) uses the generalized residual moment and shows that children are endogenous to the participation decision.

⁴ Angrist and Evans (1998) used a different instrumental variable for the number of children: sex-composition of children to estimate the causal impact of number of children on the mother’s participation in the labour market. In the US, parents like to have mixed gender children. That is, if the first two children are of the same gender, they are more likely to have a third child in hopes that the third will be of a different gender. In the context of Iran, however, Azimi (2015) shows that parents have a preference for sons rather than mixed gender children. Households whose first child or children are girls are more likely to have more children in hope of having a son. Hence, in this context, the instrument is son-preference. But as will be discussed later, this instrument may not satisfy the exclusion restriction in Iran.

⁵ Esfahani and Shajari (2012) also offered reduced form estimations of FLFP. Majbouri (2015) estimates a structural model of female labour supply. It finds a very elastic supply curve and attributes it to the fact that potential wages for women on the margin of working are similar to those who work. For a complete review of the literature, see Majbouri (2010, 2016).

⁶ Majbouri (2010) and Salehi-Isfahani (2005b) discuss these changes in educational attainment and their causes in detail. Majbouri (2018) argues that strong demand for education has been possibly the main driver for such increases in education.

⁷ Figure A1 in the Appendix uses the Iranian Annual Household Income and Expenditure Survey (HIES) from 1990 to 2013 and depicts the FLFP rates by education level over time. For all education levels, FLFP was either almost flat or declining. Women with college education and above are the only group who have an FLFP rate above 20% but their participation has been declining, partly because more women were getting college education. For more information on these trends, see Majbouri (2010, 2016).

⁸ For a detailed description of women’s labour market in Iran, see Majbouri (2010, 2016).

⁹ Following other studies, such as Jacobsen, Wishart Pearce, and Rosenbloom (1999) a third-order polynomial is employed. Controlling for higher polynomial order of age at first birth does not change the result.

¹⁰ Iran is one of the few Muslim countries in which these techniques are legal and widely used today.

¹¹ This rate was over 3.4% in 2011 in the US (Center for Disease Control website at https://www.cdc.gov/nchs/fastats/multiple.htm, accessed on April 2018).

¹² In addition, in an important study, Braakmann and Wildman (2014) show that the bias that is caused by not controlling for infertility treatments is small.

¹³ There are 29 provinces. That translates into 58 clusters.