A new method allows for causal estimates of treatment spillover effects in an instrumental variable design when spillovers in treatment take-up can be restricted by the analyst.

Menu