تقاضا ی چند دسته ای و قیمت گذاری سوپرمارکت

When a supermarket cuts its price for one product category it may increase the demand for another by drawing more consumers into the store. We call this a multi-category effect. We contrast the prominent role that the multi-category nature of supermarket shopping has enjoyed in competition policy discussion – and in the theoretical literature on supermarket pricing – with its lack of prominence in the empirical IO literature on pricing incentives for products sold in supermarkets. Using a data set of store-category choices from the UK we document empirical features of supermarket shopping and find that these are consistent with the multi-store multi-category framework of many theoretical models of retail pricing. We report on a project in progress that empirically models consumer demand allowing for multiple demand categories and two store shopping, and describe how the model can be used to measure the empirical importance of multi-category effects in supermarket pricing.

The supermarket industry's sales represent about 5.5% of disposable income in the US ($562bn in 2010, Food Marketing Institute), 8% of GDP in the UK (£150bn in 2010, Competition Commission (2008)), and similar shares in other EU countries. Given its importance it is not surprising that competition authorities regularly examine grocery retailing. In recent years competition authorities in the US and EU have considered a range of issues in supermarket competition, notably merger control, market power and below-cost selling. A distinctive feature of supermarket competition is that shoppers use a small number of stores for their weekly shopping and buy a large number of grocery categories—dairy products, alcoholic beverages, meat, bread, fruit, vegetables etc.1 Supermarkets allow shoppers to avoid the costs of visiting separate specialized stores for each category (butcher, baker, and so on). Shoppers will choose a supermarket on the basis of which is best for an entire bundle of products, and this means that the price (and quality) set by a supermarket for one product has an effect on the demand for unrelated products in the same store. We refer to these as “multi-category” or “cross-category” effects.2 A supermarket firm will internalize cross-category effects, unlike specialized stores that sell only one category, with potential consequences for equilibrium prices. We have three objectives in this paper. The first is to highlight an imbalance in the literature on supermarket competition between competition policy and theoretical IO – where the multi-category nature of shopping has had a prominent role – and the empirical IO literature on supermarket pricing where most empirical studies have considered a single product category in isolation. Our second objective is to document some empirical features of shopping that are relevant when considering multi-category effects, using a survey of UK consumer shopping decisions that contains information on the number of stores visited and the number of categories bought. The findings point to the relevance of a multi-category and multi-store-shopping framework for supermarket pricing, as used in several theory models. Our third objective is to report on a project in progress that empirically measures multi-category effects in supermarket competition, using a model of store choice and consumer demand for a range of grocery categories. We sketch some basic elements of the model and contrast it with previous work on pricing of products sold in supermarkets. The paper is organized as follows. In Section 2 we discuss policy issues that motivate our interest in a multi-category multi-store model of demand. In Section 3 we highlight relevant theoretical models of retail competition. In Section 4 we look at the patterns in the data. Section 5 discusses existing empirical models of pricing for grocery products in which multi-category effects are absent, and reports on a new modeling project that aims to incorporate these effects into the analysis. Section 6 concludes with avenues for future work.

A number of theoretical models have analyzed retail pricing incentives in a two-store, two-category framework. These models suggest that cross-category incentives can have a significant impact on equilibrium prices. We have documented some evidence from the UK supermarket industry highlighting the empirical relevance of a multi-store multi-category framework: the median bundle size on a trip to a store is 18 categories, with a lower and upper quartile of 9 and 29 respectively, and the typical shopper visits on average one or two stores for his weekly shopping. The multi-category nature of supermarket competition has played a prominent role in competition policy discussions in the EU and the US. There is however a surprising lack of work in empirical IO that measures the impact of cross-category incentives on equilibrium prices of products sold in supermarkets: most existing studies of pricing incentives for products sold in supermarkets consider only a single product category, and thus do not allow for cross category effects. We report on an empirical project in progress that models consumer demand in a multi-category-two-store framework capable of measuring cross category effects. The next steps in the project are to apply the model to questions of substantive interest. These questions include an assessment of the effect of cross category effects on equilibrium prices of individual supermarket categories (for example the price of milk). The model is also suitable for analyzing questions that have dominated the competition policy debate, namely the incentives for below-cost selling and the effect of inter-format mergers. In a related paper (in progress) Schiraldi et al. (2012) estimate a model of supermarket choice with multi-store shopping, and consider the effect of UK planning regulation interventions that restrict the development of certain store formats (e.g. so-called “big-box” stores). Finally, we note that the model assumes a complete information framework in which consumers are fully informed about prices and products when they make their choice of store. Incomplete information and advertising has played a role in some of the theoretical models of supermarket pricing—for example Lal and Matutes (1994) and Lal and Rao (1997). A promising avenue for future work, outside the scope of this research project, is to incorporate incompletely informed consumers and advertising decisions by firms.