Andres Barrios Fernandez

This paper provides causal evidence that close neighbors significantly influence potential applicants’ decision to attend university. I create a unique dataset combining detailed geographic information and individual educational records in Chile, and exploit the quasi-random variation generated by student loans eligibility rules. I find that potential applicants are significantly more likely to attend and complete university when their closest neighbor—-defined as the closest individual applying to university one year before—-becomes eligible for a student loan and enrolls in university. This increase in enrollment is mediated by an increase in the probability of taking the admission exam and applying to university. The closest neighbor typically lives 0.09 km away, and neighbors’ influence decays with distance. My results highlight the importance of social influences for university enrollment decisions and suggest that financial aid and university access policies may have important spillover effects.

Family and social networks are widely believed to influence important life decisions but causal identification of those effects is notoriously challenging. Using data from Chile, Croatia, Sweden, and the United States, we study within-family spillovers in college and major choice across a variety of national contexts. Exploiting college-specific admissions thresholds that directly affect older but not younger siblings’ college options, we show that in all four countries a meaningful portion of younger siblings follow their older sibling to the same college or college-major combination. Older siblings are followed regardless of whether their target and counterfactual options have large, small or even negative differences in quality. Spillover effects disappear, however, if the older sibling drops out of college, suggesting that older siblings’ college experiences matter. That siblings influence important human capital investment decisions across such varied contexts suggests that our findings are not an artifact of particular institutional detail but instead a more generalizable description of human behavior. Causal links between the postsecondary paths of close peers may partly explain persistent college enrollment inequalities between social groups and suggests that interventions to improve college access may have multiplier effects.

This paper investigates whether the effects of a reform that substantially increased daily instruction time in Chilean primary schools vary depending on school institutions. Focusing on incumbent students and exploiting an IV strategy, we find that longer daily schedules increase reading scores at the end of fourth grade and that the benefits are greater for pupils who began primary education in no-fee charter schools rather than in public schools. We provide evidence that these two types of publicly subsidized establishments, which cater to similar students but differ in their degree of autonomy, expand the teaching input in different ways: in order to provide the additional instruction time, no-fee charter schools rely more on hiring new teachers and less on increasing teachers’ working hours than public schools do.

This paper uses rich administrative data from Chile to provide causal evidence that the local institutions of the neighborhood to which inmates return after prison matter. Specifically, we show that the opening of an Evangelical church reduces twelve-month reincarceration rates among property crime offenders by 11 percentage points, an effect that represents a drop of 18% in the probability of returning to prison for this group of individuals. We discuss three classes of mechanisms that could drive these effects: promotion of Evangelism, provision of social support, and increased difficulty to commit crimes. Our analyses suggest that the social support that Evangelical churches offer to their communities—i.e., charitable activities and alcohol and drug abuse rehabilitation—is an important driver of their effects on recidivism. Evangelical churches also seem to make it more difficult to commit crimes by reducing the number of potential criminal partners in the neighborhood. Finally, we show that NGO openings also reduce recidivism. Organizations helping their beneficiaries to improve their earnings potential or to overcome alcohol and drug abuse problems reduce reincarceration rates by 11 and 10 percentage points respectively. These results suggest that interventions that give recently released inmates access to local support networks could play an important role in encouraging desistance from crime.

The identification of peer effects is challenging. There are many factors not related to social influences that could explain correlations among peers. This article discusses the main challenges for the identification of peer effects, describes some of the empirical strategies commonly used to overcome these challenges, and summarizes the main findings of the literature on peer effects in education. Peers have been shown to affect many important outcomes, including academic performance and educational trajectories. Confirming the existence of peer effects is important from a policy perspective. Both the cost-benefit analysis and the design of policies are likely to be affected by the existence of social spillovers. However, making general policy recommendations from the current evidence is not easy. The size of the peer effects documented in the literature varies substantially across settings and depending on how peers are defined and characterized. Understanding what is behind this heterogeneity is thus key to extract more general policy lessons. Access to better data and the ability to map social networks will likely facilitate investigating which peers and which characteristics matter the most in different contexts. Conducting more research on the mechanisms behind peer effects is also important. Understanding these drivers is key to take advantage of social spillovers in the design of new educational programs, to identify competing policies, and to gain a deeper understanding of the nature and relevance of different forms of social interactions for the youth.

This paper uses rich administrative data from Chile to estimate teacher-value added (TVA) on test scores and on an index of educational attainment. We allow each teacher to have a different TVA for male and female students and show that differences in TVA explain an important part of the gender gaps we observe in test scores and in postsecondary education trajectories. We next exploit rich information on teaching practices and show that at least in terms of the practices we observe there do not seem to be important differences in what makes teachers effective for male and female students. We do find, however, significant associations between certain practices and teacher effectiveness, which suggest that some teaching strategies benefit all students independently of their gender. Finally, we also show that on average female teachers are more effective at teaching female students and that math teachers tend to be biased in favor of male students. Interestingly, teachers with smaller gender biases seem to be more effective for both, male and female students.

Do elite colleges help talented students join the social elite, or help incumbent elites retain their positions? We combine intergenerationally-linked data from Chile with a regression discontinuity design to show that, looking across generations, elite colleges do both. Lower-status individuals who gain admission to elite college programs transform their children’s social environment. Children become more likely to attend high status private schools and colleges, and to live near and befriend high-status peers. In contrast, academic achievement is unaffected. Simulations combining descriptive and quasi-experimental findings show that elite colleges tighten the link between social and human capital while decreasing intergenerational social mobility.

This paper uses a large-scale RCT to evaluate in a unified setting---i.e., Chile---a low-touch and a high-touch intervention designed to help high school senior students to make informed choices about their postsecondary education trajectories. In line with previous research, we find that providing information alone improves students' understanding of the higher education system but does not make a difference in their probability of applying to or enrolling in college. In contrast, providing information and mentoring increases students' probability of registering and taking the college admission exam by 12.8 percentage points, of applying for funding by 10.3 percentage points, and of enrolling in higher education by 8 percentage points. The increase in higher education enrollment is similarly explained by an increase in attendance to universities and to vocational higher education institutions. The design of the RCT also allows us to study spillovers of the mentoring program on the classmates and friends of treated students. We find evidence of strong social spillovers. Despite not finding evidence of social learning among classmates of treated students---i.e., they do not improve their understanding of the higher education system---they become 5.1 percentage points more likely to register and 5 percentage points more likely to actually take the college admission exam. Nevertheless, they do not become more likely to apply for funding or to university, and as a result, we do not find evidence of them becoming more likely to enroll in higher education. In contrast, close friends of treated students do improve their understanding of the higher education system and become 5 percentage points more likely to apply to university and 4 percentage points more likely to enroll in higher education. These results suggest that social spillovers can multiply the effect of policies designed to expand access to higher education and that they can be used to improve the cost efficiency of college-going interventions.

Women make up only 35% of global STEM graduates, a share unchanged for a decade. Using administrative data from ten centralized university admissions systems, we provide the first cross-national decomposition of the STEM gender gap into a pipeline gap (access and preparedness) and a choice gap (application decisions). The pipeline gap varies widely--from female disadvantage in Uganda to advantage in Sweden--yet the choice gap is strikingly consistent: even among top scorers, women are 25 percentage points less likely than men to apply to STEM. This stability across diverse contexts points to structural forces beyond local conditions.

This paper provides causal evidence that geographic specialization can significantly enhance police effectiveness. Using rich administrative and survey data from Chile, we examine a major reform that subdivided police operational areas--e.g., municipalities--into smaller zones known as quadrants. On average, each municipality was divided into seven quadrants, with officers permanently assigned to these territories to allow them to develop a deep understanding of their structure, crime patterns, and communities. By exploiting the staggered implementation of the reform across municipalities, we show that this reorganization enhanced police effectiveness along multiple dimensions. Among surveyed households, twelve-month victimization rates declined by 10 percentage points (36%). In line with this result, administrative records from the police reveal a 14% reduction in reported crime. The reform also enhanced public confidence: the share of households reporting high trust in police rose by 12 percentage points (30%), while those perceiving increased criminal activity fell by 15 percentage points (36%). Consequently, the share of households investing in private security measures decreased by 7.7 percentage points (37%). Evidence suggests these improvements stem from geographic specialization, as households in treated municipalities report both greater police presence and better police performance across multiple dimensions associated with a better knowledge of the quadrants and their communities.

The rapid decline of Nokia mobile phone operations in 2009–2012 left many high-skilled workers looking for new career paths. We use rich matched employer-employee data covering all Nokia workers and other individuals in Finland to study how this sudden labor market shock affected displaced workers. We find that workers displaced from Nokia experienced large and long- lasting losses in the labor market. They suffered large drops in earnings and were more likely to be unemployed than similar workers displaced from other firms even three years after the mass layoffs took place. These losses, however, were attenuated by an important increase in entrepreneurship. We find that a distinctively large share of the high-skilled Nokia workers established a new business after being displaced (9% compared to 3% for displaced workers from other firms). This effect was amplified by generous start-up grants provided by Nokia since 2011 as a part of Bridge, their global support program for displaced workers. The larger number of entrepreneurs does not seem to have resulted in lower entrepreneurial quality. The firms founded by former Nokia employees perform similarly to those founded by similar workers displaced from other firms or to all those established in Finland during the decline of Nokia. This result suggests that encouraging high-skilled displaced workers to become entrepreneurs can reduce the costs of mass layoffs, as it increases the number of established firms without significantly affecting their performance.

This paper provides causal evidence that admission to an elite university significantly increases the probability of becoming an influential politician. We combine rich administrative data from Chile with a regression discontinuity design to show that admission to elite college programs increases the probability of becoming an influential bureoucrat, but not the probability of being elected for Congress. These elite college program effects are particularly pronounced for men, and are similar for students of different social backgrounds. Elite college programs also seem to shape political alignment.

This paper exploits intergenerational linked data from Chile to investigate how parents’ higher education trajectories impact the colleges and majors to which their children apply. We overcome endogeneity concerns by exploiting sharp admission cutoffs generated by centralized admissions in a regression discontinuity design framework. We find that children are significantly more likely to apply to a university and to a specific university-major combination if one of their parents was admitted to it in the past. While the selectivity and earnings associated to their parents’ university degree do not make a significant difference for daughters, sons only follow their parents if they attended a high-earnings and high-selectivity university degree.