The Behavioralist as Policy Designer: The Need to Test Multiple Treatments to Meet Multiple Targets

We explore Tinbergen's fundamental insight that policymakers need at least as many policy instruments as targets. We extend this idea using a large natural field experiment in water resource management. We use social comparisons and loss-framed messages to help achieve two goals of our partner utility: getting consumers to purchase drought-resistant plants and reducing water use. Our results show that seemingly related behavioral instruments can affect different household decisions. By themselves, social comparisons and loss framing have no significant impact on the number of rebate requests; when combined, however, they lead to a 36% increase in requests. Only loss framing leads to a significant increase in the purchase of drought-resistant plants, and only the social comparison reduces water consumption. These results highlight the importance of testing different combinations of instruments, particularly when policymakers have multiple goals and the relationship between instruments and goals is uncertain.