The Exercise of Market Power in Laboratory Experiments

Douglas D. Davis and Charles A. Holt March 1998 Introduction Many aspects of antitrust policy are influenced by the possibility that sellers in concentrated markets may have the power to raise prices above competitive levels. Of course, anyone can raise price, so the issue is whether a change in structure, e.g. a merger, will allow one or more sellers to raise price profitably. A price increase by one seller diverts sales to others, so a firm is more likely to have market power when its competitors have limited capacity to expand their sales. Thus a merger that reduces competitors' capacity may create market power. In laboratory experiments, any treatment that reassigns "units" of capacity from one seller to another may create or destroy market power, even if the aggregate supply curve is unchanged. For example, giving one seller a relatively large endowment of infra-marginal, low-cost units may make it less risky for this seller to raise price and forego the sale of high cost units. Thus it is not surprising that experiments with apparently similar supply and demand structures may yield different price outcomes. As discussed below, there are a number of other ways in which structural changes can affect the degree of power that sellers enjoy. Market power can also be sensitive to the trading institution, and this is the context that power issues first arose in experimental economics. For example, the public posting of uniform prices in a posted offer auction eliminates the incentive to offer discounts on marginal units that arises in a double auction. In fact, Smith (1981) found that even monopolists are sometimes unable to maintain prices above competitive levels in double auctions, whereas, supra-competitive prices are the norm in postedoffer monopolies. Similarly, Holt, Langan, and Villamil (1986) found that sellers in concentrated double auctions were sometimes able to raise prices, but that prices fell to competitive levels in about half of the sessions, despite the high concentration of seller capacity. Davis and Williams (1991) replicated the Holt, Langan, and Villamil results for double auctions, and then used the same market