TreasureX Researchnber.wpBetter Experiments for Measuring Spillover Effects
Kirill Borusyak, Peter Hull, Evan M. Munro · August 12, 2026
When researchers run experiments, they usually focus on the direct effect of a treatment on the people who receive it. But in many real-world situations, one person's treatment can affect others who didn't directly get it. This is called a spillover effect. For example, if a new teaching method is given to some students, their classmates might also learn more. Measuring these spillovers is important for policymakers and businesses, but designing experiments to do so is tricky.
This paper proposes a new framework for designing and analyzing experiments that aim to estimate spillover effects. The authors show that the optimal design balances two things: making the spillover signal strong (so you can detect it) and spreading the variation across the network (so you can measure it accurately). They also develop a new estimator that uses a 'recentered instrumental variable' approach to get the most out of the data.
The method is tested on two real-world applications from development economics. The results show that their approach can reduce standard errors by 50–100%, effectively doubling the sample size without extra cost. This means more reliable conclusions from the same budget.
For investors, this research is important because it can lead to better evaluations of policies and programs, especially in areas like education, health, and technology adoption. More accurate estimates can help decide where to invest, whether in public programs or private ventures.
This article is based on research by the National Bureau of Economic Research. Read the full paper at https://www.nber.org/papers/w35570#fromrss.
Educational content only, generated with AI assistance from public research abstracts. Not investment advice.