Working Papers

The Effects of College Desegregation on Academic Achievement and Students’ Social Interactions: Evidence from Turnstile Data [Draft] [Slides]

How does the desegregation of elite schools impact academic achievement? And does desegregation affect students' interactions with different types of peers within their school? In this paper, I study a natural experiment at an elite university in Colombia where the number of low–income students tripled as a result of the introduction of the financial aid program Ser Pilo Paga. The average increase in the percentage of low–income peers –9.5 percentage points– had modest to null impacts on wealthy students' academic performance. I shed light on the mechanisms behind this lack of peer effects by studying changes in social interactions using data on students' co–movements across campus captured by turnstiles located at all entrances. Desegregation led to increased connections between wealthy and low–income students. At least half of the increase in interactions between wealthy and low–income students, however, is explained by interactions of wealthy students with low–income but high–achieving students. These results suggest students diversify their interactions primarily among students with similar academic achievement levels.

Presented at: Association for Public Policy Analysis and Management Conference 2020, North East Universities Development Consortium 2021, Southern Economic Association 2021, Economics and Education Research Seminar at Columbia University 2021
Trust and Network Formation (joint with: Juan Camilo Cardenas, Danisz Okulicz, Davide Pietrobon and Tomás Rodríguez Barraquer)

We study whether trust towards strangers is a determinant of social networks among an incoming cohort of first-year undergraduate students. We employ an experiment and survey questions to measure students’ trust before they have substantial chances to meet and socialize. After four months, during which the students have many opportunities to interact, we elicit five networks capturing different relationships between them. The students’ initial levels of trust do not significantly predict the relationships they formed after four months. In contrast, time of exposure, similarity in socioeconomic status, and hometown are relevant determinants of relationship formation.

Work in progress

The Returns to College for Low-income students: Evidence from a Student Loan Program in Colombia (joint with Fabio Sanchez)

This paper examines the returns to college investments for low–income youth participating in a student loan program in Colombia. The loan’s policy setup plus the rules determining eligibility enable an instrumental variable research design that isolates exogenous changes in the number of college terms attained. Plus, the loan assignment rule allows examining individuals over all the distribution of academic achievement. First, we find the loan has a significantly positive impact on the probability of college enrollment, completion, and the cumulative number of college terms attained. Plus, the loan has a positive effect on average daily wages of 13 percent. Our estimates of the returns to college investments indicate that the average wage premium to an additional college term is 3.5 percent. Importantly, we find these returns are driven by achieving college completion, as students who do not earn a degree experience a return below 3 percent. Our findings suggest low–income students do experience significantly positive returns to college investments but that policies targeting low–income youth should place a strong emphasis on boosting college completion. This paper overcomes the limitations of prior research that has estimated returns to college education by either using natural experiments or by estimating effects at the margin of policy eligibility based on different rules.

Presented at: Association for Education Finance and Policy (AEFP) 2021


Reports on higher education