Identification of peer effects is complicated by the fact that the individuals under study may self-select their peers. Random assignment to peer groups has proven useful to sidestep such a concern. In the absence of a formal randomization mechanism it needs to be argued that assignment is `as good as' random. This paper introduces a simple yet powerful test to do so. We provide theoretical results for this test. As a by-product we equally obtain such results for an approach popularized by Guryan, Kroft and Notowidigdo (2009). These results help to explain why this approach suffers from low power, as has been observed elsewhere. Our approach can equally be used to test for the presence of peer effects in the linear-in-means model without modification.
asymptotic power; bias; fixed effects; peer effects; random assignment; test;
- C12: Hypothesis Testing: General
- C21: Cross-Sectional Models • Spatial Models • Treatment Effect Models • Quantile Regressions
TSE Working Paper, n° 21-1270, 10 novembre 2021