Electronic Thesis and Dissertation Repository

Thesis Format



Doctor of Philosophy




Navarro, Salvador

2nd Supervisor

Stinebrickner, Todd

3rd Supervisor

Mehta, Nirav


In the first chapter, we study the effects of changing the priority structure in the centralized high school admission system in Mexico City. Academically elite schools experience excess demand, while admission priorities are based on a standardized admission exam. The system ignores other skill measures such as grade point average (GPA), which may better capture non-cognitive skills that are important for later education and life-cycle outcomes. Using a Regression Discontinuity Design (RDD), we first show that marginal admission to an elite school decreases the graduation probability for students with below-median GPA and increases it for students with above-median GPA. Guided by this evidence, we then study the effects of a counterfactual admission policy wherein elite schools use a priority index that flexibly combines information on both the admission exam and the middle school overall GPA. Our counterfactual results show that more females and lower-income students would be admitted to elite schools, and the graduation rate at elite schools would increase by six percentage points. Overall, our findings show that including the information contained in GPA to define a priority structure improves equity of access, decreases mismatch, and increases graduation.

The second chapter uses five years (2005-2009) of administrative data on the centralized high school admission system in Mexico City to study whether the academic effects of being marginally admitted to an elite science school depend on the year of admission. I show that the effect on mathematics test scores at the end of high school decreases each year, starting positive and statistically significant in 2005 and ending close to zero and not significant by 2009. I propose two mechanisms to explain this trend. The first is related to changes over time in the composition of marginally admitted and rejected students combined with heterogeneity in the effect of marginal admission. The second considers changes over time in the production functions of elite and non-elite schools. Together, these results highlight the limited external validity of estimates obtained at a single point in time as they may be systematically influenced by time-varying changes in the educational context.

The third chapter studies students’ choice between academic and non-academic schools when they are uncertain about their academic skills. We implement a Randomized Control Trial (RCT) and find that providing students in Mexico City with more accurate information about their academic skills creates a better alignment between students’ skills and the type of schools they attend. This better alignment increases on-time graduation.

Summary for Lay Audience

The first chapter of my dissertation is motivated by the treatment of many centralized education markets as one-sided matching problems. In these markets, schools are seen as objects to be consumed by students, and the central planner creates a priority structure that schools should follow in admitting students. Since the priority structure is ultimately a planner's choice, current systems can improve it, especially if the one in place increases inequality and affects educational outcomes.

The motivation for the second chapter of my dissertation comes from the observation that education markets are not static. Schools and students within a centralized market may change over time. In this context, estimates of the effect of elite schools on test scores for a given period may depend on the aggregate state of the education market. If this is the case, we would expect to observe time-varying effects.

The third chapter of my dissertation focuses on school choice under uncertainty. A particular type of uncertainty students face when choosing schools is that they may not know their academic ability. Suppose students over or underestimate their academic ability when selecting schools, and there are match effects between schools and students. In that case, information could improve students' choices and help create better matches.