This paper identifies language as a far-reaching determinant of the math gender gap and analyzes its underlying mechanisms. Using an epidemiological approach on a global sample of immigrant students, we show that girls speaking a sex-based language at home have worse performance in math, accounting for 0.3 to 0.56 standard deviations of the gender gap. This relationship is primarily driven by linguistic influences on student traits and choices, including reduced motivation and intention to study math, lower self-accessed math competency, and a reduced likelihood of choosing math-intensive academic tracks. It is not explained by lower intrinsic interest in math or higher levels of math anxiety. We also examine whether sex-based language is associated with gender disparities in parental attitudes regarding the importance of math, parental support and expectations for math related careers, and whether parents serve as potential role models through their occupations. Interestingly, these parental channels do not appear to drive the linguistic effects on the math gender gap. The findings are robust to empirical strategies addressing potential omitted variable bias.

The 2022 Programme for International Student Assessment (PISA) reveals an alarming fact. Among around 690,000 15-year-old students across 81 countries, girls have lower self- efficacy in programming, debugging and algorithm design than boys. We study whether language explains these discrepancies. Using an epidemiological approach, we focus on immigrant students and find that sex-based grammatical gender, a marker and a catalyst of traditional gender norms, is associated with a widening of the gender gap by 0.164 standard deviations in programming self-efficacy, 0.217 standard deviations in debugging self-efficacy and 0.149 standard deviations in algorithm design self-efficacy. Examining potential correlated factors, we find that immigrant girls speaking a sex-based language have access to the same digital resources as boys, but are exposed to parental attitudes that assign lower economic value to mathematics. These girls experience higher distress from online content and cyber bullying. Furthermore, even when attending the same school as boys, these girls report a lower sense of safety and belonging and experience more bullying at school, although the teacher support and school risks are similar.

This paper unbundles the role of the Confucian clan through a systems approach, focusing on its long-lasting effects on women by shaping a traditional gender institution in China. Specifically, we ask whether the Confucian clan, as a risk-sharing institution, perpetuates itself by reinforcing a patrilineal system through establishing gender-based norms and choice sets and confining women to the private sphere (inside the household), thereby compelling women to disproportionately bear the social costs. Using nationally representative survey data and an instrumental variable approach, we find that the Confucian clan explains a significant portion of traditional gender norms, and shapes a distinct choice set for men and women, including earlier marriage, son preference, gender disparities in schooling, and a traditional gender-based division of labor in both households and workplaces. Furthermore, we find that when the Dibao program, a minimum livelihood guarantee program in China, replaces the Confucian clan's role in risk sharing, gender norms become more liberal and son preference diminishes.

Affirmative action aims to support underrepresented groups but can also shape how exogamous parents identify their children. Leveraging affirmative action in higher education jointly with the 1999 higher education expansion in China, this paper examines the unintended demographic consequences of affirmative action. A difference-in-differences analysis shows that the expanded higher education increased the likelihood of minority identification among children of Han fathers and minority mothers by 4.0 percentage points (equivalent to 19,000 minority newborns every year), with no impacts for children of minority fathers and Han mothers. Event studies reveal a pronounced impact for boys, suggesting that identity-concerned parents respond differently to economic incentives for sons and daughters. Within-household and province-level analyses further confirm the impact of ethnic affirmative action on ethnic identity choice.