Abstract
In this paper, we propose a structural framework to study multi-channel demand. Our model explains a comprehensive set of demand outcomes as a function of prices and retail store proximity, including the frequency with which consumers shop, how much they spend per purchase occasion, whether they buy from the online (web) or retail channel, and how they allocate expenditures among multiple product categories. We allow channels to convey different amounts of information about product categories, which in turn affects a consumer's expected utility from purchasing in a particular channel. For example, physical inspection of goods in the retail channel can provide information about product fit and feel that is difficult to assess in the online channel. Another distinguishing feature of our model is that shopping trip expenditures are endogenously determined in the first phase of a multi-stage budgeting process, where consumers allocate their income by trading off utility for the outside option and the expected utility from optimal channel and category expenditure choices with the focal brand. A key methodological contribution of the paper is to advance a highly efficient algorithm to compute optimal expenditures, which facilitates joint estimation of the model parameters by maximum simulated likelihood.
We estimate the model using the purchase histories of approximately 10,000 randomly selected customers from a firm that uses both online and retail channels to sell directly to consumers. The firm doubled its retail footprint over our two year observation window, providing a rich source of customer-specific variation in retail store proximity that we leverage to identify the demand effects of interest. We find evidence of channel complementarity through increased overall shopping frequency as the distance to retail outlets decreases, accompanied by increased substitution from online to retail formats. Our estimates imply a 10% reduction in retail store distance increases existing customer annual revenues 0.53%, by increasing retail revenues 1.96% and decreasing online revenues by 1.43%. In a series of counterfactual experiments, we demonstrate how our model can be used as a decision tool for managers to identify promising locations for new physical stores and to explore channel-based price discrimination policies.