In this dissertation, we take a broad view of higher education and examine U.S. educational attainment using models from the macroeconomics and structural labor economics toolbox. This enables us to disentangle the effects of higher education policy on aggregate human capital accumulation that arise due to effects that would be difficult to see using a simple linear model. In the first chapter, we address recent policy proposals that discuss eliminating tuition at public 2-year colleges in order to raise college enrollment among students with low parental income. This paper calculates the elasticity of college enrollment to public 2-year college tuition subsidies and estimates the resulting long-term labor market outcomes for those affected. To this end, we construct a life-cycle heterogeneous agents model with a rich set of college choices and estimate its parameters with panel data tracking educational decisions and post-college earnings. We estimate that there is a sharp increase in aggregate educational attainment as a result of the subsidy. However, we also find that there is an overall decrease in mean life-cycle earnings due to students downgrading from 4-year colleges to 2-year colleges. In the second chapter, we address the recent trends in the increase in baccalaureate time to degree (TTD). Since the 1970s, time to baccalaureate degree has increased, even conditioning on signals of student ability. Previous studies have linked this phenomenon to decreases in appropriations for public post-secondary education that have occurred in the past 30 years. This study proposes an alternate explanation, using a model of learning one's own ability in college. Agents are uncertain about their own ability and conduct Bayesian updating using their grades as they progress in college. With an increase in the college premium, agents are less likely to drop-out of college with poor grades, due to an increase in the option value of staying in college. We find that an increase in the college premium with the same magnitude as observed in the past 30 years explains more than half of the observed increase in TTD.