Modeling retail performance using consumer panel data: a Shanghai case study

Policy-makers and retailers are keenly watching and debating developments in Chinese retailing. As a contribution to the debate we present a method for investigating retail performance. Our focus is on sales performance and the components of performance in the context of store-type patronage (e.g., the number of customers patronizing hypermarkets, supermarkets, small grocers, convenience stores, wet markets and street vendors to buy frequently-bought goods). We report an analysis of consumer panel data collected from 471 Shanghai households during 2004. For mineral water (a Western product) and dried noodles (a more traditional Chinese product) we show there are regular patterns of store-type patronage, most of which are predictable from the NBD-Dirichlet model. These predictions enable retail performance to be assessed. Deviations are apparent too (e.g., that patronage of wet markets and small grocers for dried noodle purchasing is somewhat anomalous, whereas this is not evident for mineral water purchasing).