Policies that guarantee admission to the flagship university for the highest performing students from every high school in the state have the potential to make enrollment more proportionally representative of the state's high schools. However, existing evidence shows that these policies have only a modest impact on high school to flagship enrollment patterns. It is unclear whether these policies fail to induce new applications, fail to alter admissions decisions, or both. This paper studies the effects of Wisconsin's new top percent policy at the state flagship, UW–Madison, where students in the top five percent of their high school class are now guaranteed admission. Using a difference-in-differences design comparing Wisconsin and Minnesota high school seniors, I find that the policy increased the probability a student from Wisconsin applied to UW–Madison by only 0.4 percentage points, equivalent to roughly 240 additional applications annually. Students also respond to the policy's early action requirement by shifting when they submit applications. I then estimate how likely guarantee-eligible applicants are to have been admitted under the university's standard admissions process and predict that 97 percent would have likely been admitted without the guarantee. I find little change in the distribution of UW–Madison's in-state freshmen enrolling from the state's various high schools. My findings provide evidence that top percent policies may not meaningfully change enrollment patterns when they have limited impact on students' application choices and when guarantee-eligible applicants would have likely been admitted without the policy. This highlights a limitation of admission guarantees alone and suggests that efforts to change enrollment patterns may also need to target students’ application choices.

Unstably-housed students are substantially more likely to be suspended from school than their stably-housed peers, but less is known about whether disciplinary incidents are related to specific housing shocks or transitions these students face. I use linked administrative data from Wisconsin’s Homeless Management Information System and public schools to study whether unstably-housed students are more likely to be suspended from school during weeks in which they enroll in homeless shelters or services. The data identify the specific timing of both shelter or service enrollment and school suspension, allowing me to compare suspension risk during and outside periods of housing-related shocks or transitions. I find that unstably-housed students are 0.3 to 0.5 percentage points more likely to be suspended during weeks in which they enroll in homeless shelters or services, relative to an average weekly suspension probability of 1.1 percent. Results are robust to individual fixed effects, alternative sample restrictions, and placebo tests using randomly assigned enrollment weeks. These findings provide evidence that incidents resulting in school discipline are particularly likely during periods of acute housing instability and suggest that unstably-housed students could benefit from additional school-based support during these periods.

Where students attend college may shape their earnings not only through the characteristics of that institution, but also through the labor markets surrounding that college. This paper examines whether the strength of a college's local labor market affects graduates' post-college earnings. Because students' location preferences may jointly determine where they attend college and where they subsequently work, I use college application portfolios to identify and account for revealed location preferences. I study this question using two data sources: the nationally drawn Educational Longitudinal Study of 2002 and administrative data from the Texas Education Research Center. Among students willing to attend college outside their home state, I find that attending college in a labor market where bachelor's degree holders earn $10,000 more annually increases graduates' early-career earnings by $2,200-2,400. Mediation analysis suggests that this operates through graduates' subsequent work locations: attending college in a stronger labor market increases the likelihood of working in a stronger labor market. In contrast, for students choosing among public universities within Texas and remaining in the state for work, I find little evidence that colleges' local labor market strength affects earnings one or five years after graduation. Together, these findings suggest that college location can affect graduates' earnings by impacting the region in which a college graduate enters the labor market, but that differences in a college's labor market strength within the same region do not impact graduates' earnings.