The Influence of Patriarchal Norms, Institutions, and Household Composition on Women's Employment in Twenty-Eight Muslim-Majority Countries

ABSTRACT

The low level of women's employment in Muslim-majority countries is often explained by patriarchy, while disregarding variation among and within these countries. Using a new theoretical framework, this study translates patriarchy as a concept to macro- and micro-level explanations of employment. It formulates and tests hypotheses for societal norms and institutions and household composition, including how the latter's effects are context dependent. The study analyzes data from surveys (1997–2008) for twenty-eight countries, 383 districts, and 250,410 women and finds that men's public dominance over women decreases women's employment. Presence of – in particular non-foster – children and elderly people at home withholds women from labor market entrance. However, presence of other women in the household stimulates labor market entrance. Absence of a partner, male household head, or other adult men pushes women into the labor market, and thus, for example, male breadwinners' absence has a weaker negative effect in contexts of male public dominance.

KEYWORDS:

• Employment
• patriarchy
• household
• Islam

JEL Codes:

• D1
• J12
• Z1

NOTES ON CONTRIBUTOR

Niels Spierings is Assistant Professor of Sociology, specializing in political and gender sociology. The topics he studies involve economic and political participation, Islam, economic and political development, social science research methods, gender and inequality, and migration. He has published in journals such as Journal of Marriage and Family, Electoral Studies, Review or Religious Research, Electoral Behavior, Politics & Religion, and European Journal of Women's Studies.

ACKNOWLEDGMENTS

I am very grateful to Mieke Verloo, Karen Anderson, the participants and organizers of the Feminist Economics workshop on “Gender and Economics in Muslim Communities,” and the anonymous reviewers for the useful and inspiring suggestions on this study. I would also like to thank the participants of the Ninth International Conference of the Middle East Economics Association in Istanbul, Turkey, for commenting on an early draft of this article. The study is made possible by PAPFam and Measure DHS, whose data are used, and The Netherlands Organisation for Scientific Research (NWO) for providing a research grant (400-07-136).

Notes

¹ Elissa Braunstein's (2014) contribution to this issue is a welcome exception.

² This database collects and merges large-scale representative household surveys conducted in developing countries (DDW 2010).

³ In fourteen of the twenty-eight countries in this study, there are legal restrictions in public movement; and in twenty countries, women's economic access is limited (for example, individual entitlement to loans or bank accounts).

⁴ The results cannot be generalized to never-married women and women of other ages. The label “not married” should thus not be read as being single in the never-married sense.

⁵ The online Supplementary Appendix is available at http://dx.doi.org/10.1080/13545701.2014.963136.

⁶ This variable and marital status are not multicollinear.

⁷ The last three factors were not applicable to all women (for example, households without children). The missing values are replaced by the weighted average of the other women, and a variable was included to indicate no children were present (Paul D. Allison 2001).

⁸ The variable cannot tap into women's own human capital, either, because all models are controlled for the individuals' educations.

⁹ Using employment might seem problematic. However, theoretically the concept of prescriptive norms directs to the inclusion of the genderedness of employment, which shapes whether it is considered acceptable for women to have a job. Reversed causality is minimal, as the presence of a single woman will not substantially change the prescriptive norm. Methodologically, the bias can only be minimal: employment makes up only half of the district-level variable and is divided by men's employment, making the figures relative, not absolute. Moreover, each district estimate is based on (on average) about a thousand women. A single woman accounts only for 0.1 percent of that aggregate. Finally, the models are controlled for women's educational attainment, as well as for the partners' occupation. The district-level variable cannot tap into that.

¹⁰ In a factor analysis, these items load on one factor. The district-level average or proportion of these indicators is rescaled: maximum 1 (most patriarchal), minimum 0 (least patriarchal). The index is the mean of these four.

¹¹ The GID data are unique in presenting comparative data for Muslim-majority countries. Using the data to compare Muslim-majority countries with other countries is problematic because the chosen indicators focus on manifestations of gender inequality that are most prevalent in Muslim-majority countries (such as veiling; Mieke Verloo and Anna van der Vleuten [2009]).

¹² Areas with higher unemployment among men, fewer service and light manufacturing jobs, and more wealth have lower employment rates among women. The last, after control for other socioeconomic variables, might tap into the absence of a necessity to work.

¹³ Models without the number of children non-biologically related to the head of household show largely similar results for the macro-level variables. Only the country-level public patriarchy variable showed a larger and more significant effect. Using a dummy for the presence of non-biologically related children delivers the same results.

¹⁴ Statistically significant effects: higher education (+); living in a city (+); having a partner with an upper white-collar job (+), with an agriculture job (-), and who is older than the respondent (-); and age (inverted u-curve).

¹⁵ Models without the number of children non-biologically related to the head of household show highly similar micro-level results. Using a dummy for the presence of non-biologically related children delivers the same results.

¹⁶ The sex ratio among children was statistically significant, but not economically (0.3 percentage points in the average situation).

¹⁷ This effect might seem a repetition of the marriage effect. That the other adult women in a household can help in care tasks might be related to the fact they these women are unmarried. Indeed, in the data used, 95 percent of the women in “single adult women households” are married; that percentage is between 86 percent and 90 percent for women in “multiple adult women households.” This indicates that marriage plays a role in this relationship, but the observation that married women benefit from the presence of other (perhaps) unmarried women in the household is still in line with the theoretical reasoning on this variable.

¹⁸ The age of these children is normally distributed around the teenage years, and they are roughly equally divided in boys and girls.