Ce mémoire regroupe différentes études sur la préférence pour les garçons au Pakistan. Celles-ci analysent la fréquence et l'étendue de la préférence pour les garçons et son effet sur la grossesse des femmes. Le rôle que la préférence pour les garçons joue sur la participation des femmes au niveau des décisions au sein du ménage est examiné, tout comme son effet sur l'espacement des naissances, la probabilité d'une naissance à risque, ou l'impact de l'âge auquel une femme se marie sur l'équilibre entre les sexes et le développement de l'enfant né. Nous trouvons qu'il existe une préférence réelle et déclarée pour la progéniture masculine au Pakistan. La probabilité de poursuivre les grossesses diminue avec le nombre de garçons nés. De plus, les femmes qui ont au moins un garcon ont plus d'influence sur les décisions quotidiennes du ménage mais pas sur les décisions financières. La participation des femmes sur les décisions à prendre au sein du ménage augmente avec le nombre de garçons mais seulement jusqu'à la troisième naissance. D'ailleurs, les femmes ayant au moins un garçon attendent plus longtemps avant d'avoir d'autres enfants. Ce constat est plus particulièrement vrai dans le cadre d'une première naissance et il est moins présent à partir de la deuxième naissance. En outre, le fait de se marier à 18 ans ou plus tard influence de façon positive la préférence des femmes sur la composition de sa famille. Pourtant, peu importe qu’une femme se marie avant ou après 18 ans, cela ne modifie pas le biais en faveur des garçons dans leur investissement parental. A partir de ces résultats, nous proposons des préconisations politiques afin de lutter contre les inégalités entre les sexes au Pakistan. / This thesis is a collection of studies on son preference in Pakistan. The studies analyze the prevalence and strength of son preference and its effects on women’s childbearing. The role son preference plays in determining women's participation in intra-household decisionmaking is examined as are its effect on birth spacing, probability of risky births and role of maternal age at marriage in modifying gender-specific reproduction and development outcomes. We find strong evidence for both the revealed and stated preference for male offspring. The probability of continuing childbearing also decreases with the number of sons born. Furthermore, we find that women with at least one son have more say in ‘routine’ household decisions but not in financial decisions. Female participation in decision-making grows significantly with the number of sons but only up to the third parity. We find that women with at least one son are more likely to delay succeeding births. We obtain strong evidence at parity 1. The impact seems to dissipate beyond the second parity. Moreover, we find that marriage at 18 or later positively influence women’s preference for family’s sex composition. However, whether or not a woman married early or late does little to modify the male gender bias prevalent in parental investment. In light of these findings, we suggest policy measures that could help improve gender equity in the country.

This dissertation comprises four chapters, which mainly deal with female's participation in household decision-making, a very important aspect of female's bargaining power within the household and closely linked to female's empowerment. The first three chapters, which all deal with female's participation in household decision-making, are two sides of the same coin, in that while the first one delves into the determinants of female's participation in household decision-making, the second and third chapters deal with its beneficial consequences. The fourth chapter is linked with Chapter 1. As a matter of fact, the data used in Chapter 1 has been collected in Rural Burundi, in the framework of the FNRS/FRFC-funded project “Microfinance Services, Intra-household Behavior and Welfare in Developing Countries: A Longitudinal and Experimental Approach”, which funded my PhD scholarship. In 2012, the project funded data collection in Rural Burundi. In respect to the experimental component of the project, these are baseline data. The 2012 household survey targeted a sample of rural households that have been interviewed in 1998 and 2007. This is where the longitudinal design of the project comes into play. Independently from the experimental research, the longitudinal nature of the data, that is to say three waves of data (1998, 2007 and 2012), had the advantage of allowing panel analysis of interesting and relevant issues in development, including for example the long-term welfare effects of shocks at either individual or household levels.In Chapter 1, entitled “The Power of The Family: kinship and Intra-household Decision-making in Rural Burundi” and co-authored with Bram De Rock and Philip Verwimp, we delve into the determinants of female's participation in household decision-making, by laying a particular emphasis on the role of female's kinship. We show that in rural Burundi the characteristics of the female's kinship are highly correlated with her decision-making power. First, a female whose own immediate family is at least as rich as her husband's counterpart enjoys a greater say over children- and asset-related decision-making. Second, the size, relative wealth and proximity of the extended family also matter. Third, kinship characteristics prove to be more important than (standard) individual and household characteristics. Finally, we also show that the female's say over asset-related decision-making is positively associated with males' education, more than with female's education per se. All these correlation patterns can inform policies aiming at empowering women or targeting children through women's empowerment.In Chapter 2, entitled “The Returns of I Do: Multifaceted Female Decision-making and Agricultural Yields in Tanzania?”, I use the third round of the Tanzanian National Panel Survey to investigate the effect of multifaceted female's empowerment in agriculture on agricultural yields. The classic approach in the empirical literature on gender gap in agriculture includes the gender of the plot's owner/manager as the covariate of interest and interprets the associated coefficient estimate as the gender gap in agricultural productivity. Unlike this classic approach in the analysis of productivity differentials, my approach lays emphasis on the overlapping and interaction effects of manifold aspects of female's empowerment in agriculture, including female plot's ownership, female plot's management and female output's control. I find significant productivity gaps, which the classic empirical approach does not bring out in the same context. As compared to plots (solely) owned, managed and controlled by male, (i) plots merely owned by female and (ii) those owned & managed (but not controlled) by female are less productive, but those owned, managed & controlled by female are not. Furthermore, the latter are the more productive among plots at least owned by female. All these productivity gaps are predominantly explained by the structural effect, that is differences in productivity returns to observable production factors. Our findings are robust along a number of dimensions and suggest that female's management and control rights are of prime importance. Therefore, female plot's owners should be entitled the rights to manage their plot and, subsequently and most importantly, the rights to control the (agricultural) output of their work, for their productivity to be enhanced and the gender gap in agriculture to be closed. In Chapter 3, entitled “Say On Income and Children's Outcomes: Evidence from Nigeria”, I delve into the effect of female bargaining power on child education and labor outcomes in Nigeria. Female bargaining power is proxied by “female say on labor income”, rather than by her income per se. This is motivated by the fact the female labor force participation might be low in some contexts, while control over income is by all means what matters the most. The empirical methodology accounts for a number of empirical issues, including endogeneity and sample selection issues of female say on labor income, the multi-equation and mixed process features of the child outcomes, as well as the fact that hours of work are left-censored. My findings are consistent with the overall idea that female say on income leads to better child outcomes, rather than female income earning per se. Nevertheless, the type of income under female control, child gender and child outcome matter. Chapter 4, entitled “Violence Exposure and Welfare Over Time: Evidence From The Burundi Civil War” and co-authored with Marion Mercier and Philip Verwimp, investigates the relationship between exposure to conflict and poverty dynamics over time. We use a three-wave panel data from Burundi, which tracked individuals and reported local-level violence exposure in 1998, 2007 and 2012. Firstly, the data reveal that headcount poverty has not changed since 1998 while we observe multiple transitions into and out of poverty. Moreover, households exposed to the war exhibit a lower level of welfare than non-exposed households, with the difference between the two groups predicted to remain significant at least until 2017, i.e. twelve years after the conflict termination. The correlation between violence exposure and deprivation over time is confirmed in a household-level panel setting. Secondly, our empirical investigation shows how violence exposure over different time spans interacts with households' subsequent welfare. Our analysis of the determinants of households' likelihood to switch poverty status (i.e. to fall into poverty or escape poverty) combined with quintile regressions suggest that, (i) exposure during the first phase of the conflict has affected the entire distribution, and (ii) exposure during the second phase of the conflict has mostly affected the upper tail of the distribution: initially non-poor households have a higher propensity to fall into poverty while initially poor households see their propensity to pull through only slightly decrease with recent exposure to violence. Although not directly testable with the data at hand, these results are consistent with the changing nature of violence in the course of the Burundi civil war, from relatively more labor-destructive to relatively more capital-destructive. / Doctorat en Sciences économiques et de gestion / info:eu-repo/semantics/nonPublished