An empirical investigation of the welfare effects of banning wholesale price discrimination.

Economic theory does not provide sharp predictions on the welfare effects of banning wholesale price discrimination: if downstream cost differences exist, then discrimination shifts production inefficiently, toward high-cost retailers, so a ban increases welfare; if differences in price elasticity of demand across retailers exist, discrimination may increase welfare if quantity sold increases, so a ban reduces welfare. Using retail prices and quantities of coffee brands sold by German retailers, I estimate a model of demand and supply and separate cost and demand differences. Simulating a ban on wholesale price discrimination has positive welfare effects in this market, and less if downstream cost differences shrink, or with less competition.

1. Introduction

Wholesale price discrimination means that an upstream firm sets different prices for the same product for various downstream retailers. Wholesale price discrimination is commonly practiced in many markets. Examples include markets such as petroleum distribution, steel, heavy trucking, tobacco, and pharmaceuticals. In several countries, milk and other dairy products are sold using government-administered or -sanctioned discriminatory pricing schemes. Wholesale price discrimination practices are also used for services such as loans, insurance, and advertising. (1) Competition authorities have been concerned with wholesale price discrimination. The European Union Treaty's Article 82 (c) prohibits practices where a dominant firm "would place trading partners at a competitive disadvantage." The U.S. Robinson-Patman Act forbids "discriminat[ing] in price between different purchasers" where the effect "may be to lessen competition" unless the price differences are based on costs, or price differences were needed to meet competition. (2) As I discuss below, theory does not provide clear prediction about the welfare effects of wholesale price discrimination. The goal of this article is to empirically assess the welfare effects of eliminating the possibility of wholesale price discrimination.

Wholesalers price discriminate for several reasons, such as to take advantage of differences in downstream demands and differences in downstream costs, forces which have opposing welfare effects. To exploit differences in retail demands, it is optimal for the upstream firm to discriminate by setting a lower wholesale price in the more price-sensitive downstream markets and a higher price in the less price-sensitive ones (Schmalensee, 1981; Varian, 1985; Ireland, 1992). Bork (1978) posits that, as a result of such discrimination, total welfare will increase if new markets are served due to wholesale price discrimination. This effect is similar to the welfare test of a total increase in quantity in the case of third-degree price discrimination in final goods markets.

If downstream firms' costs differ, the upstream firm sets a higher wholesale price to the more efficient, lower-cost, retailer (Katz, 1987; DeGraba, 1990; Yoshida, 2000) and price discrimination will lower total welfare. By wholesale price discriminating, the manufacturer shifts quantity from the more efficient retailer to the less efficient retailer. Moreover, in this context, the usual result that an increase in quantity sold increases welfare is no longer true. In the presence of downstream cost differences, an increase in quantity sold may actually be a symptom of a welfare decrease due to wholesale price discrimination because the "wrong" retailer is selling more" (Yoshida, 2000). (3)

Because the welfare effects of price discrimination cannot be determined theoretically, I estimate the effect empirically. In particular, I estimate the effects of uniform wholesale price legislation, which bans wholesale price discrimination, in a German grocery retail market where manufacturers wholesale price discriminate among retailers. As a first step, I estimate a flexible demand system using detailed price and quantity data. Using the demand estimates, I am able to investigate whether differences exist in demand for a brand sold at different retailers in this market. Given the demand estimates, and a supply model of linear pricing of manufacturers and retailers, I compute price-cost margins for retailers and manufacturers. (4) By subtracting the estimated retail and manufacturer margins from observed prices, I am able to recover the sum of retail and manufacturer costs for each brand sold at each retailer as a residual. To recover retail cost differences in this market, I assume that a brand sold at two different retailers has the same manufacturer costs. Subtracting the differences in estimated retail and manufacturer margins from price differences of the same brand sold at two different retailers, I estimate there to be retail cost differences, that is a force toward welfare to improve with the ban on wholesale price discrimination. Next, I estimate the welfare effects of a ban on wholesale price discrimination by computing the counterfactual Nash equilibrium when the manufacturers are not allowed to wholesale price discriminate and the estimates suggest that banning price discrimination has positive welfare effects in the market. Finally, I show through counterfactual simulations that the estimated positive welfare effects are smaller if downstream cost differences are smaller, or if there is less competition in the market.

This article is related to the literature cited above that studies the welfare consequences of banning price discrimination in intermediate goods markets. In terms of the empirical strategy, I follow the recent literature by modelling vertical relationships (see, e.g., Bonnet, Dubois, and Simioni, 2006; Brenkers and Verboven, 2006; Goldberg and Verboven, 2001; Hellerstein, 2008; Sudhir, 2001; Mortimer, 2008; Villas-Boas and Zhao, 2005; Villas-Boas, 2007a; Villas-Boas and Hellerstein, 2006). Two related studies combine the same scanner data with additional data sources to empirically examine the determinants of retail and manufacturer margins in the German coffee market. Draganska and Klapper (2007) relate estimated manufacturer conduct parameters, in a reduced-form setting, to exogenous factors related to retail competitive environment. Draganska, Klapper, and Villas-Boas (2008) estimate margins in a structural model of multiple Bertrand-Nash competing retailers. Assuming a certain retail model as given, they estimate a simultaneous bargaining model between each manufacturer and each retailer for each product. In doing so, their goal is to relate bargaining power parameters in the model, in a reduced-form setting, to potential determinants. Although the present article uses the same data as Draganska, Klapper, and Villas-Boas (2008), the contribution of this article is to provide the first empirical investigation of the economic forces at play behind wholesale price discrimination.