EXPRESS: Optimal Return Policies with Heterogeneous Search Rewards

This paper studies return policy design for an online seller serving consumers who can resolve product-fit uncertainty through costly pre-purchase search, with search rewards varying across consumers. Return policies shape not only post-purchase return behavior but also ex-ante search and purchase decisions, thereby generating endogenous consumer segmentation. One might expect that inexpensive search would reduce the seller's incentive to offer a generous refund. In our baseline model, however, we show that full refund outperforms no refund when search costs are sufficiently low. We trace this result to a fundamental trade-off in implementing a full-refund policy: the search-relief effect versus the return-inducing effect. Full refund substitutes for search by allowing consumers who would otherwise search to purchase directly and evaluate fit after purchase, thereby removing their search-based information advantage and restoring the seller's pricing power. However, it also induces returns from consumers who would otherwise purchase without search and keep the product, sacrificing the revenue generated by their tolerance for misfit. When search costs are sufficiently low, most consumers would otherwise search, and the search-relief effect therefore dominates the return-inducing effect. As search costs fall further, full refund can also achieve a win-win outcome, indicating that pre-purchase search may be socially inefficient. These findings explain when and why sellers may optimally discourage ex-ante search in favor of ex-post product evaluation.