Essays in development economics

This dissertation is composed of three chapters. The first two study topics related to higher education. Chapter one analyzes the relative advantages of different entry systems to university, and chapter two investigates the effects on academic outcomes of attending a higher quality campus at the college level. The third chapter examines household size choice.The second chapter, "The Effect of College Quality on Academic Performance. Evidence from Mexico", is a joint project with Rodrigo Barros. We estimate the effects of attending a higher quality campus at the college level on academic outcomes using the case of UNAM. Incoming students from high schools run by the University are allocated to majors and campuses in UNAM using a mechanism that is based on a score. The score depends on the student's academic performance in high school, and the mechanism gives priority to students with higher scores. The allocation process generates a series of discontinuities where students with similar transition scores are assigned to different campuses due to capacity constraints. This allows us to use a regression discontinuity methodology. We find that attending the main campus rather than a secondary one significantly improves peer characteristics. That is, classmates of students in the main campus come from a better socioeconomic environment and have a better academic background. However, attending a higher quality campus does not significantly affect graduation probabilities or GPA, it increases the number of failed and dropped courses, and deteriorates the ranking by GPA of a student in her class. A possible interpretation is that other variables, like the student's ranking in the class, counteract the positive peer characteristics.Finally, the third chapter, "Families as Roommates: Changes in U.S. Household Size", is a joint project with Todd Schoellman and Michele Tertilt. This study is motivated by the empirical fact that the average American today lives in a household of three people, compared to six in 1850. We propose a mechanism based on the demand for privacy as an explanation for part of this reduction in household size. We hypothesize that as people get richer, they consume a lower proportion of public goods, which can be shared among all household members, relative to private goods, which cannot be shared. This reduces the economies of scale of living with other people, and endogenously decreases the optimal household size. We present a model that formalizes this idea and calibrate it to fit facts about the relationship between household size, consumption patterns, and income in a cross-section at the end of the 20th Century. We project the model back to 1850 by modifying income to its actual level at that point it time. The results indicate that our proposed mechanism can account for 37% of the difference in the number of adults in the household between 1850 and 2000, and for 16% of the decline in children.In the first chapter, "Entry Systems to University: Evidence on the Trade-offs of Early Selection versus Open Admission", I analyze the relative advantages between the two main entry systems to university in the world. In one, only those students who attended specific high schools are eligible to apply to college (early selection). In the second, all high school graduates can apply (open admission). I use the case of UNAM, a large public university in Mexico, where the two systems exist simultaneously to study the advantages of each of the two channels. UNAM runs a system of high schools (early selection), but it also admits students from other schools (open admission). I estimate whether attending a UNAM high school increases the probability of being a successful student in university. I identify this causal effect using a student strike that displaced students from one channel into the other. Results indicate that attending a UNAM high school reduces the probability of dropping out between the first and second year of college from 9 to 7 percent. Additionally, I show that some students from non-UNAM high schools have higher expected graduation probabilities than some students from UNAM high schools. Thus, the University benefits from admitting students who were not tracked into their schools. Finally, I use an analytical framework to test whether UNAM is selecting students from the two channels in a way that maximizes expected graduation rates. Results suggest that selection is optimal, given the particular constraints of the admissions process.