The United States faces a projected physician shortage, with medical school expansion serving as a prominent policy response. How expansions affect local physician supply is poorly understood. Using physician location data from the National Plan and Provider Enumeration System (NPPES) and a stacked difference-in-differences design with matching, I find physician supply increases by 9.0\% over the following decade in the opening county. These gains are driven by established physicians moving into the county, particularly early- and late-career physicians. In contrast, isolated rural counties within 60 miles of a new medical school experience a 10.3\% decline in physicians over the same period. Mortality rises alongside these physician losses, with a 3.6\% increase in all-cause mortality in rural spillover counties.

State Loan Repayment Programs (SLRPs) are a prominent policy tool for addressing healthcare provider shortages in underserved areas, yet little is known about how participation shapes providers' subsequent location decisions. I characterize applicants to the Colorado Health Service Corps (CHSC) and document their location decisions using administrative data from Colorado's Primary Care Office matched to longitudinal practice location data from the National Plan and Provider Enumeration System. A key contribution of this paper is the ability to identify both awarded and unawarded applicants, allowing me to compare the location patterns of providers who received a loan repayment offer against those who were willing to serve in a shortage area but were not funded. I find that offered providers are less likely to move during their obligation period relative to the not offered, consistent with the program's service requirement. However, this pattern reverses in the first two years following the obligation period, with offered providers exhibiting notably higher mobility. These findings suggest that the CHSC is able to recruit individuals into underserved areas in Colorado; however, its impact on long-term retention is less clear.