Identification and Estimation of Spillover Effects in Randomized Experiments

Despite the longstanding interest in spillover effects across different fields in economics, the empirical literature has not come to an agreement regarding their identification and estimation in experiments in the presence... [ view full abstract ]

Despite the longstanding interest in spillover effects across different fields in economics, the empirical literature has not come to an agreement regarding their identification and estimation in experiments in the presence of externalities. I provide a causal framework to study spillover effects when units are clustered and their potential outcomes can depend on the treatment assignment of all the units within a cluster. Using this framework, I discuss parameters of interest and provide conditions to identify spillover effects, and show how several commonly studied models can be subsumed under this general framework. I assess the performance of the difference in means, which ignores the presence of externalities, linear-in-means models and two-stage randomization designs. I then evaluate under what conditions these three techniques identify the parameters of interest. Finally, I put forth easily implemented empirical strategies for estimating these spillover effects and give conditions for consistency and asymptotic Normality of the estimators when clusters are allowed to be large relative to the total sample size. I illustrate my proposed methodology using two empirical applications: a randomized conditional cash transfer pilot in Colombia conducted by Barrera-Osorio et al (2011), and an experiment on voting behavior conducted by Ichino and Schündeln (2012).