Retailers' choice of product variety and exclusive dealing under asymmetric information.

This article considers vertical relations between an upstream manufacturer and a downstream retailer that can independently obtain a low-quality, discount substitute. The analysis reveals that under full information, the retailer offers both varieties if and only if it is optimal to do so under vertical integration. However, when the retailer is privately informed about demand, it offers both varieties even if under vertical integration it is profitable to offer only the manufacturer's product. If the manufacturer can impose exclusive dealing, then under asymmetric information it will do so and foreclose the low-quality substitute even if under vertical integration it is profitable to offer both varieties.

1. Introduction

* Downstream retailers sometimes enhance their product variety by offering low-quality, discount substitutes for the products produced by upscale manufacturers. Thus, for example, the market share of private labels that have been introduced by supermarkets and drugstores has been growing rapidly in recent years, and stores selling electronic goods and home appliances often offer reputable brands as well as unfamiliar, low-priced substitutes. At the same time, upstream manufacturers sometimes limit the variety that their retailers can offer by imposing exclusive dealing arrangements prohibiting the retailer from selling products that compete with those of the manufacturer.

This article addresses three questions. First, what are retailers' incentives to enhance their variety by offering both qualities instead of just high quality? In particular, are these incentives different for vertically integrated and separated industries? This question is of special concern in the context of private labels, because it might be expected that, with their superior production capabilities, upstream manufacturers will be able to produce high-quality products at quality-adjusted costs that are lower than those of the private labels, thus making the introduction of private labels unprofitable.

The second question relates to the incentives that an upstream manufacturer may have to impose exclusive dealing on its retailer, which prohibits the sale of brands that are substitutes for the manufacturer's brands. On one hand, a manufacturer may impose exclusive dealing because of welfare-enhancing reasons. For example, exclusive dealing may induce a retailer to focus its promotional activities on the manufacturer's products and thereby improve customers' service. Marvel (1982) argues that exclusive dealing can secure investments made by the manufacturer (in quality assurance and advertising, for instance) by preventing other manufacturers from free-riding on them. However, exclusive dealing may also be anticompetitive when a manufacturer that benefits from a leading position in the market imposes exclusive dealing for the sole purpose of foreclosing competing brands. (1)

This second potential anticompetitive effect of exclusive dealing has been challenged by the well-known Chicago School for two related reasons. (2) First, if offering a second brand increases the retailer's gross profit, then it will also benefit the manufacturer, which can now charge the retailer higher franchise fees. In this case, the manufacturer will not profit from foreclosing the competing brand because doing so will substantially reduce the franchise fee that it can charge the retailer. Therefore, if a manufacturer finds it profitable to foreclose a competing brand, then it has to be that this brand is a poor substitute to begin with. That is, the manufacturer will profit from excluding the competing brand only if it does not provide any additional value to the retailer's gross profit. (3) Second, even if a manufacturer imposes exclusive dealing, it will still need to compensate the retailer for the forgone profits from not offering the competing brand. Thus, it is not clear why exclusive dealing is any better from the manufacturer's viewpoint than offering quantity discounts such that the retailer will independently choose not to sell the competing brand. As Gilbert (2000) points out, the arguments made by the Chicago School parallel a more tolerant approach by U.S. courts toward exclusive dealing. (4) Altogether, these arguments raise the question of whether a manufacturer will ever choose to impose exclusive dealing for the sole purpose of foreclosing a competing brand and, if so, what the effect is of exclusive dealing on the retailer, consumers, and welfare.

The third question relates to the practice of market share contracts, in which a manufacturer provides a discount to a retailer for buying a certain percentage of its units from the manufacturer. For example, in the USA, tobacco wholesalers sued Philip Morris, a leading cigarette manufacturer, for its Wholesale Leaders program, which rewarded distributors based on their sales of Philip Morris cigarettes as a percentage of their total cigarette sales. (5) Brunswick, a leading manufacturer of marine engines, was sued by competing engine manufacturers for offering quantity and market share discounts to boat builders for buying its engines. (6) At first glance, this practice may appear to be a softer version of exclusive dealing, in that the manufacturer is restricting its retailer to commit to a certain percentage of exclusion, instead of the 100% exclusion of exclusive dealing. This raises the question of why manufacturers sometimes use market share contracts instead of quantity discounts or exclusive dealing, and what the effect is of market share discounts on consumers and welfare. (7)

This article studies vertical relations between an upstream manufacturer (M) that produces a high-quality product (H) and a downstream retailer (R), when R can obtain a low-quality substitute (L) at a given cost. For example, the substitute product can be interpreted as a private label or a low-quality product available from a perfectly competitive fringe. I compare three types of contracts: first, a simple nonexclusive contract that only specifies a quantity of H and a total price; second, an exclusive dealing contract that also prohibits R from selling L; third, a market share contract that restricts R to sell a certain quantity of L, which can be higher than zero. The model reveals that the answer to the three questions raised above depends crucially on the extent to which R is privately informed about consumers' willingness to pay for the two brands. Under full information, M can implement the first-best profits by offering the nonexclusive contract. This contract induces R to sell both L and H whenever L is efficient (such that a vertically integrated monopoly chooses to offer both L and H) and only H otherwise. In the latter case, M does not need to impose exclusive dealing or to use a market share contract to obtain exclusivity. The intuition for this result is that M can use the nonexclusive contract to capture R's entire added value from selling H and therefore wishes to maximize R's gross profit. This result implies that under full information, the decision whether to offer low-quality substitutes in the form of private labels, for example, is not affected by the vertical structure. It also supports the argument that exclusive dealing does not offer any advantage in foreclosing a competing brand and shows that this argument applies also to the market share contract.

Then I turn to consider the case where R is privately informed about a parameter, [theta], that measures consumers' willingness to pay for H and L. To induce R to reveal the true [theta], M offers R a menu of contracts in which the total payment and the quantity of H are contingent on the [theta] reported by R. R has the incentive to understate the true [theta] because this lowers the profits that M can extract from R. To minimize this incentive, M distorts the quantity of H downward. If M can only use the nonexclusive contract, then the ability to sell additional units of L provides R with a degree of freedom because the supply of L is independent of R's report on [theta] to M. Consequently, R can understate [theta] and compensate itself for the low quantity of H by selling additional units of L, which M cannot limit. Moreover, R can report a [theta] that misleads M into believing that R intends to sell H alone, while in practice R intends to sell both brands and earn additional profit from selling L. Thus, in the nonexclusive contract, under some conditions on the model's parameters such as the marginal costs, the degree of product differentiation, and the degree of the asymmetric information problem, R may offer both H and L even if L is unprofitable under full information. In this case, although L is a poor substitute for H, the degree of freedom that selling L provides R forces M to increase R's information rents.

This result indicates that under asymmetric information, retailers will expand their product variety by offering brands that are unprofitable under full information, because it enables them to gain informational leverage over manufacturers.