This thesis is composed of three independent chapters in applied microeconomics. The first chapter contributes to the growing literature on parenting in economics, by providing causal evidence of the impact of parenting on child development. The second and third chapters belong to the field of labour economics, the former analysing gender differences in career choices in a unique setting, and the latter causally estimating the impact of local labour market gender composition on workers’ behaviours. Even though the three chapters are distant in terms of topics, they have one important similarity: they all answer societal questions, by relying on observational data and putting great importance on the technics used to identify the parameters of interest.
The first chapter studies the long-term effects of growing up with more or less protective parents. To induce quasi-experimental variation in parental attitudes, we focus on rare but shocking events: nearby child kidnappings. Using geo-localized information from the PSID and a matching strategy (of U.S. counties), we find that the occurrence of a kidnapping causes a decrease in children’s cognitive skills and lowers the probability of finishing high school. Turning to mechanisms, we find no evidence that kidnappings make parents or children more neurotic. However, they change parenting style, limiting the time children spend unsupervised and decreasing parental involvement.
The second chapter studies gender differences in early occupational choices in a con-text in which the traditional explanations for occupational segregation, namely human capital investment and discrimination, are shut down. This setting allows to focus on the role of individual preferences for particular job attributes in explaining gender-based occupational segregation. Using unique data on the mechanism used in France to allocate medical students to residency positions, we find that men and women facing virtually equivalent occupational choice sets make drastically different career choices. We show that this result holds when focusing on unconstrained choices, which suggests that preferences for job attributes play an important role in determining gender-based occupational segregation. We then estimate gender differences in preferences for these job attributes when facing external constraints on their choices, relying on an incentivized preference elicitation mechanism. We do not find evidence suggesting that men and women differ in their propensity to prefer location over occupation.
Motivated by the observation that there exists a gender gap in the amount that free-billing physicians decide to set in France (thereafter exceedances), the third chapter studies the effect of local gender composition on physicians’ tarification behaviour and activity. To isolate the effect of gender composition from potential confounding factors, I use (i) the panel dimension of the data to get rid of unobserved heterogeneity, and (ii) an instrumental variable strategy solving the endogeneity of gender composition to shocks to the local environment. I use data on the universe of French private physicians, and construct local labour markets using information on commuting times. I find that an increase in female composition leads to an increase in fees, exceedances and number of procedures performed, on average. When looking at men and women separately, I find that while the effect on fees and number of procedures is similar for both genders, feminization increases men’s exceedances, and marginally reduces women’s. I also find that the effect on men’s exceedances is concentrated towards the middle of the female composition distribution, while the negative effect on women is driven by local labour markets on which the share of women is already high.