Chapter 1 studies how the interaction of student information with constraints dictated by market design determines higher education choices and outcomes. I study strategic application incentives in imperfect implementations of centralized assignment mechanisms in higher education. I ask whether, in markets with both a central match for public colleges and a broader private market, choices on the match are affected by the availability good private outside options. It is unclear whether the common market configuration with outside private options and application size restrictions generates strategic incentives in applications on the public match that is advantageous to students with higher socioeconomic backgrounds. I assemble data from the college match in Albania and utilize a policy change that incorporated all private colleges in the centralized platform to generate insight. The policy does two things: first, it removes private programs as an outside alternative to the match, differentially shifting outside options of students of different backgrounds. Second, it expands the set of choices on the match among which students must choose, tightening the constraint on application sizes for students who can afford private colleges and must substitute away from applying to public college to accommodate good private options in their application. I find that ambitiousness in public college applications declines after policy implementation, and more for private high school students, supporting the hypothesis of non-truthful application behavior that is responsive to market structure. I then build a model of applications and enrollment behavior that accommodates strategic behavior and can disentangle the effects of heterogeneous beliefs, preferences, and outside options on choice to evaluate the distributional consequences of the interaction between non-truthful applications and market partitioning. I find substantial uncertainty about admission thresholds, in part responsible for strategic applications. The application behavior induced by a single match for all schools with list size restrictions worsens outcomes on average, especially for private school students, both for those who would change their applications, and indirectly through crowd-out for those who would not. The main channel is list size restrictions becoming more binding with more options on the match, but some of the effect is due to outside options becoming less good. The relative worsening of outcomes for higher-SES students may be redistributive, but better outcomes can be achieved for both groups under a single match by slightly extending application lists. In Chapter 2, joint with Martha J. Bailey, Vanessa Wanner Lang, Alexa Prettyman, Lea J. Bart, Daniel Eisenberg, Paula Fomby, Jennifer Barber, and Vanessa Dalton, we study the consequences of costliness in fertility regulation in the current US policy environment, which leaves 1.4 million uninsured Title-X clients with substantial cost sharing for contraception and reproductive health. We experimentally vary contraceptive subsidies to women in Planned Parenthood clinics of Michigan and find a substantial response to contraceptive cost, in particular for high-fixed cost methods. Mothers are the most financially constrained, but all groups increase their take-up. Our take-up estimates imply that a U.S. policy eliminating out-of-pocket costs for all Title X patients would reduce pregnancies by 5.3%, birth rates by 3.9%, and abortions by 8.3%. Finally, Chapter 3 studies the role of information and beliefs that supervisors have in determining individuals' education and career trajectories. In it, I investigate whether academic supervisors have differential information quality about their male and female students and ask how supervisors learn about skills of individuals and groups. In particular for students of high talent, early-career signals may have important consequences on future opportunities, career trajectory, and intellectual output. I assemble new archival data on all participants of the Putnam Mathematical Competition over four decades to make progress on the above questions. The setting allows me to observe an objective measure of talent in the scores achieved by each student and the supervisors' prediction about the talent ordering of students they supervise. I find that ex-ante, for women and men with the same ex-post competition scores, supervisors expect women to do worse than their male peers from that college. Women are less likely than men to have been predicted to be in the top three performers of their school even when they obtain a score that places them in the top three. Female supervisors are no better at predicting female achievement than male supervisors. I find evidence that supervisors learn about individual women, but little evidence of supervisor learning about the group. With a model of learning, I investigate whether supervisors are biased in their perceptions of gender talent or hold beliefs consistent with the empirical distribution of performance by gender. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://bibliotheek.ehb.be:2222/en-US/products/dissertations/individuals.shtml.]