Model-based evaluation of cooling-off policies

This paper studies the ex-ante prediction and ex-post evaluation of the effects of cooling-off policies when consumers may exhibit a projection bias. We set up a theoretical model in which a firm optimally reacts to consumers' preferences and the regulatory framework and show that neither the adoption of a mandatory cooling-off period nor a return policy is generically superior or consumer welfare improving. We then analyze how market-level data can help to evaluate the policies ex post using baseline statistics. This exploits the firm's optimal response to a policy. With a return policy, data on quantities, return frequencies, and market size are sufficient to always assess the directional change in consumer welfare, while aggregate quantities alone are sufficient with a mandatory cooling off period. We discuss robustness of the model predictions and ex-post assessment to a variety of modifications, and discuss the benefits of the approach for policy design.