Optimal Allocation of Seats in the Presence of Peer Effects

We model optimal treatment assignment in programs with a limited number of seats and study how the presence of peer effects impact the optimal allocation rule. We then use data from an RCT to show evidence that there are large peer effects in the context of job training for disadvantaged adults in the United States. Finally, we combine the model and the empirics to show that the program would have had a much greater impact if the assignment choices had accounted for the peer effects.