Research

Abstract: This paper investigates how adolescent exposure to peers from larger or smaller families influences women’s fertility decisions later in life. We distinguish between the extensive margin (whether to become a mother) and the intensive margin (how many children to have), using quasi-random variation in peers’ number of siblings across school-grade cohorts as an identification strategy. Our findings reveal that while exposure to peers with more siblings increases fertility at the intensive margin, it decreases the likelihood of becoming a mother at the extensive margin. We interpret this asymmetry through the lens of theory of economic identity: conforming to social norms entails costs, which are higher when deciding to enter motherhood than when choosing to expand an existing family. Further analysis reveals that adolescent social environments condition peer influence at the intensive margin. Specifically, maternal relationships, school connectedness, parental engagement, and exposure to working mothers shape how young women perceive the costs of conforming to fertility norms. Women with weaker parental and school ties are more susceptible to peer effects, while those exposed to working mothers among their peers perceive lower opportunity costs and are more likely to conform to high-fertility norms. These results underscore the differentiated nature of peer influence across fertility decisions and the critical role of adolescent contexts in shaping long-term demographic behavior. 

Abstract: This paper investigates maternal gender preferences in the U.S. I propose a new way of inferring gender preferences by looking at how mothers allocate disappointment between children within families. This novel approach informs about preferences even in cases where the first-born gender strategy does not yield conclusive evidence, unveiling a more subtle manifestation of gender bias. For identification, I leverage sex variation among dizygotic twins pairs and compare the probability of receiving the same maternal disappointment within pairs, between mixed-sex and same-sex twins pairs. To rule out the role of expectations, I adjust for differences in academic performance (GPA), substance use, and maternal trust, among other variables. The findings reveal that, net of expectations, mothers of mixed-sex pairs are more likely to rank daughters higher than sons in disappointment than mothers of same-sex pairs. The results provide robust evidence of subtle gender bias in maternal disappointment in favor of daughters, even in families without observable structural disadvantages, underscoring the role of implicit preferences in intra-household resource allocation and emotional investment.

Abstract: We provide novel causal evidence that exposure to a high-school alcohol culture increases women’s risk of intimate partner violence (IPV) victimization in early adulthood. Using quasi-random variation in peer composition across grades within schools and peers’ drinks-per-week polygenic indexes as a measure of genetic predisposition to alcohol consumption, we find that a one-standard-deviation increase in peers’ average alcohol predisposition raises women’s IPV victimization index by 0.045 standard deviations—approximately three-fifths of the effect of a one-standard-deviation decrease in parental socioeconomic status. Effects are driven by female peers and operate through two distinct mechanisms: exposure to a peer alcohol culture increases women’s own heavy drinking and smoking in early adulthood, and reshapes their social networks toward partners and friends who abuse alcohol. These findings identify high school as a critical intervention window and suggest that alcohol demand-reduction policies generate social multiplier effects on IPV victimization.

Abstract: We estimate peer effects on academic performance using an instrumental variables strategy that avoids reliance on friendship nominations. Students are grouped by grade and extracurricular activities, and peer performance is instrumented with the exogenous characteristics of students from other grades who participate in shared activities the individual does not attend but whose same-grade peers do. This design allows estimation of the full linear-in-means model, addressing measurement error in self-reported friendships and avoiding the need to assume exogenous network formation. A one-letter-grade increase in peer performance raises individual grades by 0.66 letters. Results are consistent with mentoring mechanisms.