Public insurers have limited resources and are faced with a trade-off between individual and collective welfare. This results in rationing, which means some services are available while others are not; and translates into limitations in access to health care. Health care is a need everyone will face at some point, therefore understanding resource allocation to optimize access is vital. This dissertation focuses on contributing evidence in support of policies to achieve this. Costa Rica is a developing country with a public insurer providing universal health care coverage. The country's health outcomes are closer to developed nations than to its socioeconomic peers; all of which make it a valuable research setting. These insights contribute to the ongoing discussion of how to organize health care systems with limited resources while having the goal of maximizing coverage and access to health care. The first chapter provides the first empirical evidence on litigation (where an individual sues an insurer to obtain coverage) as a safety valve to avoid rationing. I construct a novel dataset from all Costa Rican litigation requesting access to drugs, and for all cases involving cancer I use clinical guidelines to determine expected benefits. Using a probit model to predict successful litigation I find that higher benefit drugs have a higher likelihood of approval, but the benefit-cost ratio has no effect. An event study shows no evidence that drugs gaining coverage affects trends in drug requests. The second chapter further explores litigation using prevalence and mortality to characterize cancer cases. No pattern is evidenced for either measure which supports litigation as a complimentary mechanism for access that is sensitive to individual heterogeneity. The third chapter examines how preventive and primary care services affect elderly health by exploiting a natural experiment where the treatment was to increase access by adding providers. I measure the treatment's effect on health care utilization, nutritional outcomes, mental health, chronic diseases, and disability status using a difference-in-differences model. With treatment, all measures of healthcare utilization increased significantly. Findings on health outcomes are small and mixed, showing health possibly improving in some measurements and worsening in others.