Policy design in experiments with unknown interference
This addresses the challenge of policy optimization under interference for researchers and practitioners in experimental design, offering a novel approach but with incremental elements in methodology.
The paper tackles the problem of designing experiments to estimate and test policies when units interact within clusters in unknown ways, introducing single-wave and multiple-wave experiments that provide strong theoretical guarantees and are implemented in a large-scale field experiment.
This paper studies experimental designs for estimation and inference on policies with spillover effects. Units are organized into a finite number of large clusters and interact in unknown ways within each cluster. First, we introduce a single-wave experiment that, by varying the randomization across cluster pairs, estimates the marginal effect of a change in treatment probabilities, taking spillover effects into account. Using the marginal effect, we propose a test for policy optimality. Second, we design a multiple-wave experiment to estimate welfare-maximizing treatment rules. We provide strong theoretical guarantees and an implementation in a large-scale field experiment.