Essays on Operations Management

This dissertation consists of three related but self-contained papers in operations management. Motivated by our discussion with Healthcare practitioners, in Chapter 1, we study the impact of the Hospital Readmission Reduction Program (HRRP) from an operational and economic perspective. The HRRP, a part of the US Patient Protection and Affordable Care Act, requires the Centers for Medicare and Medicaid Services (CMS) to penalize hospitals with excess readmissions. We take an economic and operational (patient flow) perspective to analyze the effectiveness of this policy in encouraging hospitals to reduce readmissions. We develop a game-theoretic model that captures the competition among hospitals inherent in HRRP's benchmarking mechanism. We show that this competition can be counter-productive: it increases the number of non-incentivized hospitals, which prefer paying penalties over reducing readmissions in any equilibrium. We calibrate our model with a dataset of more than 3,000 hospitals in the United States and show that under the current policy, and for a large set of parameters, 4% to 13% of the hospitals remain non-incentivized to reduce readmissions. We also validate our model against the actual performance of hospitals in the three years since the introduction of the policy. We draw several policy recommendations to improve this policy's outcome. For example, localizing the benchmarking process --- comparing hospitals against similar peers --- improves the performance of the policy.;The second paper, in Chapter 2, studies the optimal service differentiation policy for service organizations in the presence of social networks. In our framework, customers' beliefs of the service quality evolve over time according to their own experiences and the reported experiences from their friends in the network. We characterize the conditions under which such belief system converges and the corresponding optimal service differentiation policy. Our main results can be summarized as follows: First, contrary to the existing literature, we show that, when customers directly report their experiences, the importance of a customer only depends on her marginal economic value, her influence on her friends, and her friends' marginal economic values. In other words, the optimal policy only needs first-order friendship information instead of the whole network structure. Second, we demonstrate that the value of knowing the social network structure critically depends on the correlation between customers' economic and social values. The social network value is higher if the correlation is lower. Third, we use a novel data set with more than 1,600 service providers to show empirically that for many service providers, there are negative correlations between the economic and social values of their customers. We also provide empirical evidence to show that the price ranges of services affect their correlations, with service providers targeting high-end markets having more negative correlations. While the second paper is concerned with the theoretic value of social network information in service management, the third paper, in Chapter 3, studies empirically how service providers can construct social interactions among customers and quantify the causal impact of that interaction on the perceived service quality. In particular, we focus on education and analyze whether encouraging social interaction among students improves learning outcomes in Massive Open Online Courses (MOOCs), which are a new service delivery channel with universal access at reduced, if not zero, cost. In this paper, we analyze two randomized experiments in a MOOC with 24,225 students from 183 countries. The first experiment studies large-group interaction by encouraging a random subset of students to visit the course discussion board. The majority of students treated by this experiment had higher social engagement, higher quiz completion rates, and higher course grades. Using this treatment as an instrumental variable, we estimate that one additional board visit causally increases the probability that a student finishes the quiz in the subsequent week by up to 3.5%. The second experiment studies small-group interaction by encouraging a random subset of students to conduct one-on-one discussions. Students who followed through and actually conducted pairwise discussions did increase their quiz completion rates and quiz scores by 10% in the subsequent week. Combining results from these two experiments, we provide recommendations for designing social interaction mechanisms to improve service quality.