This dissertation focuses on the supply side of healthcare, using the specific case of kidney donation and transplantation to address three research questions. In the first chapter, I test whether kidney transplant candidates with better access to publicly available kidneys from deceased donors are less likely to opt for living kidney donation, or the "private supply." Identification comes from a discontinuous increase in the probability of receiving a publicly available kidney generated by the results of a blood test. I find that patients with better access to the public supply of kidneys are less likely to opt for living donation. The results indicate that policies aimed at increasing the number deceased donors will result in less than full crowd-out of living donation. The second chapter examines the well-documented association between procedural volume and patient outcomes in the context of kidney transplantation. In particular, I test whether a volume effect exists in kidney transplantation, which would be consistent with the "practice makes perfect" hypothesis. Identification of the volume effect comes from plausibly exogenous supply shocks of kidneys within a year at a transplant center. The empirical results suggest that much of the observed volume-outcome relationship in kidney transplantation is due to between hospital differences in unobserved characteristics that are correlated with both transplant volume and patient outcomes. Concentration of transplants at higher volume transplant centers may reduce rates of short term patient mortality, but these gains would need to be carefully weighed against any reductions in patient access to care that would result from regionalization of care. The last chapter examines whether transplant centers experience "forgetting" during temporal breaks between kidney transplants. Identification relies on the randomness of arrivals of transplantable kidneys at transplant centers, which would create plausibly exogenous variation in the size of temporal breaks between transplants at a given transplant center. In addition, I test whether the level of experience immediately before a temporal break mitigates any deleterious effects that arise from the break. The estimated results suggest that there is little relationship between temporal breaks and transplant center productivity, as measured by the outcomes of patients transplanted immediately following a break in transplant activity.