Centralized assignment mechanisms are widely used and present in many markets. The empirical evaluation of these markets is an essential and challenging task. For instance, in college admissions, students may report their preferences strategically, making it difficult to evaluate policy changes. Also, it is unclear whether students make mistakes when applying to college and if this has consequences on their welfare. Finally, little is known if the market design can affect students’ downstream outcomes. In my dissertation, I shed light on these issues, by analyzing the Chilean centralized college admissions system. In the first chapter, with I. Rios, we document strong evidence of strategic behavior in students’ applications, even though students face no incentives to misreport their preferences. Taking this into account, we build anew methodology that recovers students’ preferences from observed application lists, even when students face a large number of choices. In the second chapter, with I. Rios, we analyze the effects of centralized assignment mechanisms on downstream outcomes. We evaluate two channels that can explain students’ dynamic choices: (i) students might switch programs or drop out due to initial mismatches, and (ii) students might switch or drop out due to learning about their match-qualities. Based on these facts, we build a structural model of students’ college progression in the presence of a centralized system. We use the estimated model to analyze the impact of changing the market design on the system’s efficiency. We find that policies that elicit information on students’ cardinal preferences and leverage dynamic incentives can significantly improve the system’s efficiency. Finally, in the third chapter, with M. Martinez, C. Neilson, and I. Rios, we analyze the prevalence and relevance of application mistakes in college admissions. Using survey data, we find that a significant fraction of students makes welfare-relevant mistakes due to a lack of information and biased beliefs. We use these insights to design and evaluate an information policy to reduce application mistakes. We find that showing information about admission probabilities has a causal effect on improving students’ outcomes.