This study evaluates the effects of specialized educational programs for linguistic minorities, focusing on bilingual indigenous schools in Mexico, on academic performance in math and Spanish and on bullying. These schools aim to provide instruction in native languages and create safe spaces for minorities who often face discrimination; however, they encounter significant implementation challenges. I develop and estimate a structural model of parents choosing primary schools for their children that incorporates bullying as a key social interaction depending on the school ethnic composition and heterogeneous human capital formation technologies. I find that bullying consistently negatively impacts academic performance (by 0.15-0.38 standard deviations) and that increasing the proportion of indigenous students in a school from 0% to 100% reduces bullying for indigenous students by 7.3 percentage points, without affecting non-indigenous students. However, I find that teachers in indigenous schools are on average less effective than teachers in regular schools in promoting academic achievement. Their effectiveness is mediated by their indigenous language proficiency. Parental school choice is influenced both by academic aspirations and bullying concerns. I use the estimated model to evaluate counterfactual policies. Enhancing resources in indigenous schools improves academic scores and reduces bullying by encouraging student shifts from regular to indigenous schools. Policies targeting ethnic-based bullying improve academic performance directly and indirectly by fostering a conducive learning environment and encouraging students to choose more productive regular schools. Eliminating indigenous schools enhance academic achievement and reduce bullying for indigenous students attending regular schools, thanks to the influx of indigenous peers. Therefore, indigenous schools need additional resources to address minorities' education needs; otherwise, they risk being counterproductive.

The limited market size of many small emerging economies is a key constraint to the growth of innovative small and medium enterprises. Exporting offers a potential solution, but firms may struggle to locate and appeal to foreign buyers. We conducted a six-country randomized experiment with 225 firms in the Western Balkans to test the effectiveness of 30 h of live group-based training and 5 h of one-on-one remote consulting in overcoming these constraints. Treated firms used techniques such as search engine optimization and improved Facebook content to increase their digital presence and better reach foreign customers. A year later, we find positive and significant impacts on the number of customers, and a significant intensive margin increase in export sales. Qualitative interviews suggest this improvement came from a combination of sector-specific advice on market expansion, and through an encouragement effect which gave entrepreneurs the confidence to try new sales strategies.

The effect of demographic diversity on teamwork is often presented as a double-edged sword: more diverse teams benefit from more creativity and knowledge sharing, but they face higher communication and coordination costs. This paper tests this consensus view in the context of higher education. We exploit an experimental setting in a large undergraduate class, where students are randomly assigned to small homework groups, with varying levels of diversity in terms of race, gender, and place of birth. We find that more diverse groups perform better when the assigned task is creative and complex, and worse when the task is analytical, which confirms the consensus view that diversity's positive impact on team performance hinges on gains from creativity. We then address the effect of team diversity on coordination and communication, building an index of teamwork quality based on collaboration between members, the balance of member contributions, and the absence of conflicts. We find that diversity has a U-shaped effect on teamwork quality, regardless of the type of task performed. This result suggests that faultines – dividing lines that split a group into subgroups based on demographic characteristics – can cause inter-subgroup cohesion to break down, while very homogeneous or very heterogeneous groups collaborate better. The results help provide a framework to understand optimal team composition for institutions and organizations.

As countries expand English instruction to support global economic integration, concerns have emerged that such policies may erode indigenous languages and cultural identity. This paper examines this possibility using a natural experiment in Mexico, where six states introduced English programs in public primary schools during the 1990s. Leveraging a staggered difference-in-differences design with linked school and population censuses, we find that exposure to English instruction instead increased the likelihood of understanding and speaking an indigenous language by 1.2 and 0.7 percentage points, respectively, substantial effects given baseline rates of only 2.2% and 1.4%. Mechanism analysis reveals that these effects reflect ethno-linguistic reinforcement rather than cognitive spillovers or compositional changes, as evidenced by increased self-identification as indigenous (1.7 percentage points from a baseline of 8.3%). Our findings suggest that multilingual education can simultaneously promote economic integration while reinforcing cultural distinctiveness.

We study Mexico’s Bachelor’s Degree in Primary Education with an Intercultural Bilingual Focus (LEPIB), introduced in 2004 to increase the supply of formally qualified teachers in indigenous bilingual primary schools. We exploit its staggered rollout across states in a difference-in-differences event-study design using school census and national test data. LEPIB increased the number of bachelor’s-degree teachers in indigenous schools by 0.3 per school by 2015 (about two additional qualified teachers per municipality), with no detectable change in teacher-pupil ratios, nor in wages, indicating a shift in teacher composition rather than class size. This increase in qualifications translated into achievement gains of 0.14 SD in Mathematics and 0.17 SD in Spanish. Using a unique dataset of survey responses from students and parents, we find that higher teacher qualifications improved perceived classroom practices, especially teacher support, alongside suggestive evidence of declines in parental engagement consistent with household substitution. Using population census, we find longer-run effects for exposed cohorts: higher employment among indigenous women (2.9-3.3 pp), lower fertility on the intensive margin (0.17-0.18 fewer children conditional on having any), and higher Spanish bilingualism among indigenous-language speakers (1.8-3.2 pp); these effects are concentrated in linguistically homogeneous communities.

Does harmonising school curricula across ethnic lines improve learning and social cohesion in deeply divided post-conflict societies, or does it impose identity-related costs on minority students? I study this question in Bosnia and Herzegovina, where cantons have begun adopting the 2018 Common Core Curriculum Defined on Learning Outcomes (CCC) at different times. The 2018 CCC is a standards-based reform: rather than reallocating instructional hours, it specifies what students should know and be able to do across seven of eight subject areas, including the historically divisive humanities, with mathematics and science outcomes aligned to TIMSS topics. Exploiting the staggered cantonal rollout, I apply a difference-in-differences design to two waves of TIMSS (2019 and 2023). CCC adoption raises mathematics achievement by 0.27--0.36~SD and science by 0.29--0.38~SD, robust across TWFE, Callaway--Sant'Anna, and Changes-in-Changes estimators, and larger still in panel schools (0.48--0.54~SD). Oster bounds ( 15--19), nine null placebo outcomes, and pre-treatment checks using four waves of the Life in Transition Survey and PISA 2018 support a causal interpretation. Quantile effects are significant throughout the distribution and largest at the bottom (+0.24~SD at the 10th percentile). Effects on school climate are sharply asymmetric across ethnic lines. Bosniak students, the majority in adopting cantons, report liking school more (+0.24~SD, p = 0.01), feeling safer (+0.49~SD, p < 0.001), perceiving teachers as fairer (+0.19~SD, p = 0.04), and score +0.31~SD (p = 0.004) higher on a 5-item climate index. For Croat students, the community-level minority in cantons that are 68-78\% Bosniak, the same reform produces negative effects on the most identity-relevant items: belonging declines by 0.54~SD (p < 0.001) and perceived teacher fairness by 0.45~SD (p = 0.04); other climate items are null. On Fridays (Jumu'ah, the Muslim day of communal prayer observed in the surrounding community), Croat students in adopting cantons report additional climate declines of 0.56 to 0.92~SD across every item, while Bosniak Friday effects are small and mostly insignificant and non-Friday weekday placebos are zero or positive. Cognitive achievement is unaffected by Friday testing for either group, indicating that the mechanism operates on identity-relevant subjective perceptions rather than test performance. Teacher-reported data trace the mechanism. Total math and science instructional minutes are unchanged, consistent with a standards rather than inputs reform, but within-subject composition shifts toward the domains the new learning outcomes emphasise (geometry +17.9~pp, physical science +16.2~pp, earth science +19.6~pp in panel schools). Treated-canton science teachers receive professional development on roughly 18 percentage points more topics on average (p = 0.020), with the effect stable across the full sample, the clean Callaway--Sant'Anna comparison, and the within-school panel restriction, confirming that the reform reached classrooms through retraining rather than reallocation. School-level ethnic composition is unchanged within panel schools (p = 0.97); near-term segregation effects are absent, though longer-term school-choice responses remain possible. Taken together, the results show that curricular harmonisation can raise achievement substantially and across the distribution, but that its effects on belonging and institutional trust can move in opposite directions for majority and minority students: a tension that shared-content reforms in divided societies cannot resolve without complementary measures to preserve distinctively minority experience.

Between 2010 and 2014, five Indonesian provinces required public schools at every grade level to teach a specific regional language (Javanese, Sundanese, Balinese, or Lampung) as a compulsory subject. Unlike most mother-tongue education policies worldwide, which target only early primary grades, the Indonesian mandates apply throughout secondary school. Using all five waves of the Indonesia Family Life Survey (1993–2014) in an event-study difference-in-differences design, I provide the first causal estimates of these mandates' effects on cognitive scores, bilingualism, and Indonesian literacy. A first stage triangulated across student, teacher, and principal reports shows that the mandate added about two hours per week to the school timetable rather than reallocating existing time. Whom the mandate harms depends on whether the mandated language matches the student's home language. I measure ethnic match at three nested scales: the province (aggregate implementation intensity), the community (local linguistic environment), and the student's own family. Different outcomes are dominated by different scales. Bilingualism reflects the community-wide linguistic environment, cognitive scores fall roughly uniformly, and Indonesian literacy is the only outcome where all three scales matter. The Indonesian-literacy result is the most striking. Estimated alone, the effect of family match on Indonesian literacy is essentially zero. Once I control for province and community match, the family effect flips sharply positive (+2.4 pp), while the province and community effects turn negative (−2.6 and −1.7 pp). The aggregate channel captures classrooms shifting toward the regional language and crowding out Indonesian exposure. The family channel captures L1-to-L2 transfer for students whose families already speak the mandated language, the mechanism predicted by Cummins (1979)'s linguistic interdependence hypothesis. The two forces nearly cancel in aggregate, producing a misleading null. Standardized Bahasa Indonesia test scores fall about 13% of the mean in the most mismatched province (Lampung), confirming the aggregate channel with an externally graded measure. A methodological lesson follows for the mother-tongue education literature: aggregate-level nulls on second-language outcomes may mask two large opposite-signed causal channels operating at different scales.

Financial stress is widespread among college students, yet little is known about how financial hardship shapes peer interactions in collaborative settings. We study this question using a field experiment in which students are randomly assigned to small groups for collaborative homework in a large undergraduate economics course, where approximately 40% of students in our sample report daily financial stress. We document three findings. First, financially stressed students collaborate significantly less with their teammates. Second, this collaboration deficit is partially closed when financially stressed students are paired with other financially stressed peers. Third, and most importantly, this re-engagement represents what we term a “low-effort coordination equilibrium”: financially stressed students paired together report higher experiential collaboration, yet they learn less from each other and produce lower-quality group output—without reporting more conflict or free-riding, consistent with mutual agreement on reduced effort. The penalty is specific to collaborative work; individual exam performance is unaffected. While financially stressed students are substantially more likely to hold jobs, replacing peer financial stress with peer employment status in the interaction specification produces no coordination failure, and when both interactions are included simultaneously, the financial stress channel fully absorbs the employment channel— suggesting that financial hardship carries a psychological burden beyond time constraints alone. These findings reveal a previously unexplored social channel through which financial stress may perpetuate disadvantage: while a large literature has documented the individual cognitive costs of financial scarcity, we show that financial stress also acts as a social multiplier, undermining coordination in team settings and generating aggregate costs larger than individual-level estimates indicate.

This paper investigates the impact of workplace harassment on various occupational outcomes, utilizing a novel empirical approach that combines reduced-form strategies with a structural equilibrium framework. Leveraging a newly available dataset from the collaboration between the Census Bureau and the Equal Employment Opportunity Commission (EEOC), this study provides an unprecedented analysis of harassment claims across firms of varying sizes and industries. By exploiting sharp thresholds in Title VII coverage and the diffusion of the #MeToo movement, we employ regression discontinuity designs and event studies to estimate the causal effects of vicarious liability on firm demographics, wage gaps, and employee turnover. We then exploit the findings in a structural model that estimates a job market equilibrium. The newly integrated Census-EEOC data product, which has not been used in prior research offers new insights into the economic implications of workplace harassment.