This thesis consists of two chapters and examines questions centered around the educational choices of students in Mexico. When school-age children work, their education competes for their time and effort, which may lead to lower educational attainment and academic achievement. Chapter 1 develops and estimates a model of student achievement in Mexico, in which students make decisions on school enrollment, study effort and labor supply, taking into account local schooling options and wages. These decisions affect their academic achievement in math and Spanish, which is modeled using a value-added framework. The model is a random utility model over discrete school-work alternatives, where study effort is the outcome of an optimization problem under each alternatives. The model is estimated using an administrative test score database on Mexican 6th grade students combined with survey data on students, parents and schools, geocode data on school locations, and wage data from the Mexican census. The empirical results show that if students were prohibited from working while in school, the national dropout rate would increase by approximately 20%, while achievement would increase in math and Spanish. Expanding the conditional cash transfer, either in the magnitude of the benefits or the coverage, in conjunction with prohibiting working while in school is an operational policy that would greatly reduce dropout while maintaining the achievement gains. In Chapter 2, my coauthor Emilio Borghesan and I analyze a large-scale and long-running distance education program in Mexico. We use an empirical framework that combines value-added modeling with a sample selection model to estimate Marginal Treatment Effects (MTEs) for learning in telesecundarias relative to traditional secondary schools. The estimated MTEs reveal that school choice is not random, and that the effect of telesecundaria attendance is positive for nearly everyone. Using performance on nationally standardized exams as a measure of knowledge, we find that the average student experiences a 0.34 standard deviation improvement in math and a 0.21 standard deviation improvement in Spanish after one year of attending a telesecundaria. We conclude by estimating the effects of counterfactual policies that expand telesecundaria availability and find that they generate improvements in academic performance.