Hi(s)tory affects Her-story: Evidence on Polygamy in Benin

2017, Supervised by Remi Jedwab


Several countries have banned and criminalized polygamy. Yet, it is commonly practiced among certain populations. One such example comes from Benin, where, despite a legal ban in 2004, more than one-fourth of the married population is in a polygamous union (DHS 2012). This paper attempts to identify mechanisms underlying such persistence of polygamy. More specifically, it investigates whether the observed persistence could be attributed to: (i) religious identity (predominance of Islam), (ii) ethnic identity (attachment to traditional polygamous beliefs) and (iii) agricultural identity (economic motivations arising from women being seen as additional workers on the field). Using extensive ethnographic mapping and a set of fixed effect models, exploiting cohort-level variation in exposure to the ban, I find that among the three characteristics, agricultural identity surfaces as a significant factor. Households that practiced shifting cultivation – a form of agriculture, which was historically female-dominated - faced significantly higher probability of being in a polygamous union. This finding sheds light on how historically embedded economic motivations could influence current-day conjugal practices. Further, it indicates that interventions beyond de-jure measures would be required to alter deep-seated norms and practices.

Impact of Polygamy on Fertility: Evidence from Ghana

2017, Supervised by Ben Williams

This paper examines the impact of polygamy on fertility. Theoretically, there are 2 competing effects: (i) substitution effect: men in a polygamous union prefer lower number of children owing to income constraints, (ii) overbidding effect: women in a polygamous union want more children to claim their stake on husband’s time, affection and property. Further, there is plausible feedback between polygamy and fertility; desire for more children (or sons) can cause selection into polygamy. In this paper, I investigate the net effect of polygamy on fertility in the context of Ghana using an instrument variable (IV) approach. I test the effects utilizing multiple instruments, each of which is a plausible determinant of polygamy (but not fertility). Preliminary IV estimates suggest that over-bidding effect dominates, i.e. polygamy increases fertility. This finding resonates how critically women may depend on reproductive power to alleviate their socio-economic standing, which is arguably threatened in a polygamous union. This result is especially noteworthy in the context of Ghana where both, maternal mortality and fertility rates are considerably high.

Let her live! An Investigation of Gender Bias in Child Mortality in India

2016, Supervised by Robert Phillips

This paper investigates the effect of underlying gender biases on child mortality in the Indian context. In so doing, it applies a set of models - linear probability, logit and probit model - on child-level panel data from Andhra Pradesh, India. Estimates from across models suggest that households with higher number of girl children face significantly higher (7pc-10pc) probability of child mortality, controlling for an array of socioeconomic factors such as dependency ratio, education and living conditions. This finding reiterates the need to account for norms and preferences in formulating strategies to reduce child mortality and more broadly, enhance child health outcomes.  

Gender Inequality in Health: Experiences from India

2015, Supervised by James Foster

This paper explores variation in health status stemming from differences non-genetic components such as access to nutrition and healthcare. More specifically, it attempts to capture embedded gender inequality in health status across regions in India. In doing so, it uses child height as an indicator of health and employs two methodologies: a percentile cut-off method (comparisons with WHO reference population) and an inequality decomposition method (bifurcation of inequality into within and between group using coefficient of variation). The findings of the paper suggest differences in male and female child height, as early as infancy, across all sampled regions, albeit in varying degrees.

Note: The above four papers were submitted towards completion of doctoral course requirements. Drafts available on request.