My research is concerned with decision making in the presence of peer spillovers. Peers’ behavior can change each other’s incentives or levy externalities on one another, thereby complicating decision making as well as the estimation of primitives from observations of decisions. In my dissertation, I address instances of the resulting challenges in topics ranging from public health and development economics to the study of making friends and contacts. The first paper explores the optimal distribution of a limited number of preventative treatments (e.g., vaccines) for deadly infectious diseases such as malaria or Ebola among individuals on a network. While the literature in economics has considered the optimal distributions of treatments in the presence of heterogenous treatment effects, the main contribution of my paper lies in accommodating for spillovers in treatments. I extend the empirical welfare maximization (EWM) protocol to do so. EMW estimates an optimal rationing rule using data from a randomized control trial (RCT). I start by explicitly modeling disease propagation on contact networks. In two separate models with distinct assumptions that are separately compatible with malaria and Ebola, I place restrictions on how others’ treatments affect one’s outcome as well as assumptions on the shapes of networks considered. I show that as the size of the experiment grows, EWM assigns the treatments to the same nodes on the network as would a planner with complete information on which nodes had the greatest impacts on the population as a whole. I also show that there can exist no statistical procedure with a faster rate of convergence. The second paper considers the role of incentives in Joint Liability Lending (JLL) microfinance programs. JLL is a common loan contract offered, particularly in South Asia. The basic premise is that participating households organize into groups to receive a joint liability loan. In such a contract, the obligation of repaying the loan falls collectively on the group and not on individual members. If the group fails to repay, the lender refuses to offer any member of the group a loan thereafter. Using a game theoretic model, I am able to demonstrate an atypical inefficiency arising in equilibrium where peer pressure leads to excessive repayment. I demonstrate that this excessive repayment, often heralded as a measure of success of the program, inefficiently shifts the burden of risk on to the participants and away from the lender. I then demonstrate that this results in participants using the loan for inefficiently safe investments rather than investments which would increase their productive capacity. Thus, I argue that the informal incentives in the form of peer pressure that JLL leverages to induce high repayments also lead to incentives that do not allow borrowers to escape poverty. In the final paper, joint with Francesca Molinari, we consider identification and estimation of models of network formation. We construct a sharp identified region for parameters of the network formation game under the assumptions that individuals’ preference are only over local network topologies and not the entire network. In particular, we assume that every individual can have a fixed finite number of links and has preferences from links between nodes up to a fixed finite distance on the network. This allows for testing important drivers of social link formation. During an epidemic, it is very desirable to understand how the network of social interactions evolve as measure like social distancing and quarantining are put into effect. A pre-requisite to doing so is the identification and estimation of network formation games. Hence this paper bridges a gap in the literature on identifying models of network formation.