Why Do Customers Start Returning More?

The emergence of online shopping led customers to take into account the presence of returns when making their purchase decisions. In this paper, I present a theoretical framework that models purchases and returns as two integrated parts of the same search-like process and sheds light on their dynamics. The model of customer learning about the return process explains population-level dynamics of the return rate: the unknown value of each customer's hassle cost of the return is drawn from a common distribution at the moment of the first return incidence. After that, the customer may adjust her return behavior, or stop purchasing from the retailer entirely. For a large family of distributions, this results in an aggregate-level increase in the return rate. The model of learning about the brand explores the individual-level dynamics. In this model, the items sold share a common uncertain attribute - brand quality, and the customer chooses the optimal sequence of purchasing, taking into account the presence of returns and the fact that some of the information of the attributes will be revealed in the process. I solve this model numerically and find that for higher-end brands with high and predictable quality, and low and unpredictable fit the individual return rate increases over time. I test differentiating between higher- and lower-end brands predictions of this framework on the data available from a large Russian online retailer. The group-level and individual-level dynamics are consistent with the models' predictions. Additional evidence on purchase and return rate dynamics also supports the hypothesized mechanisms.