ادغام عمودی و انگیزه های تبانی: تجزیه و تحلیل تجربی
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|1075||2000||26 صفحه PDF||سفارش دهید||1 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : International Journal of Industrial Organization, Volume 18, Issue 3, April 2000, Pages 471–496
We consider vertically related industries with multiple downstream markets; firms make simultaneous output choices in a repeated game. Upstream duopolists merge with producers in one of the downstream markets that also is a duopoly. Experimental duopoly markets are constructed to assess the effects of vertical integration upon outputs and profits. We find that integration raises outputs in both downstream and upstream markets, although only the upstream effect is statistically significant. Integrated profits are lower and consumer welfare is higher. The integrated markets tend to equilibrate more quickly.
Consider an intermediate good that is sold as an input in multiple downstream markets. The production of this good is dominated by a few sellers that simultaneously choose outputs. One of the downstream markets is oligopolized, and upstream firms integrate into this downstream market. This market structure may describe, for example, large gasoline refiners merging with airline companies, movie studios acquiring theaters in major metropolitan areas, or manufacturers of a product integrating downward and becoming the major retailers in a certain geographic market. There exist a variety of reasons for outputs and prices to change in these markets as a consequence of such integration. We describe some of them. Before merging, it is well known that this vertical relation suffers from the `double marginalization' problem.1 Downstream firms set a price above their marginal cost, which depends on the upstream price of the intermediate good. Profit maximizing upstream firms have already set a price for the input above their marginal cost; thus the input price is marked-up twice. By eliminating the dual markups, vertical integration should raise outputs and lower prices in both markets. In addition, integration can eliminate uncertainty in the vertical relation. If the upstream market is in a state of disequilibrium, or maintains only temporary equilibria, and downstream producers are risk averse, they would be inclined to produce less and correspondingly demand less of the input in a vertically separated structure (Carlton, 1979 and Perry, 1982). The often cited reason for airlines merging with petroleum refiners is that airlines wanted more stable deliveries and prices of jet fuel (Businessweek, November 17, 1980). In the 1948 Paramount Pictures case (334 U.S. 131 (1948)), movie studios claimed they were merging with theaters in order to guarantee outlets for their films. Since the integrated firm has information unavailable to the nonintegrated firms, a more precise prediction of upstream and downstream prices and outputs can be formed after a merger. Integration is therefore capable of mitigating production uncertainties; outputs also should increase for this reason. 2 On the other hand, a fully integrated market structure might facilitate greater levels of cooperation than a nonintegrated market structure, which might lower outputs. There is the persistent notion that if firms interact repeatedly over an indefinite number of periods, more collusive outcomes that raise prices and restrict outputs at both levels of production are possible as a result of the mergers (see Scherer, 1980, pp. 303–312). Enhanced cooperation, however, need not imply output reductions in all markets. The multimarket contact induced by vertical integration allows firms to use output increases in one market to facilitate output reductions in another in order to increase total profits (Bernheim and Whinston, 1990).3 Collusive incentives can therefore counteract incentives to increase outputs from eliminating double marginalization or reducing uncertainty. In a repeated game setting, the net effect of the vertical integration becomes an empirical issue. This paper studies the impact of mergers in vertically related markets. To keep the theory and empirics manageable we assume that before integration there are only two firms in an upstream X market, and two firms in a downstream Y market. An additional competitive Z market also uses the upstream product, so that the market demand for the upstream product can not be exactly inferred from downstream Y output. Moreover, the Y market is a small component of total demand for X, so that Y market firms know they have very little impact on upstream production decisions. We assume there are no vertical efficiencies in the technology and that costs are symmetric across rival firms. Upstream producers have no marginal costs, while downstream marginal costs are proportional to the input price. Given this stylized vertical relation, we are interested in the market effects when the two firms in X merge with the two firms in Y. To investigate how vertical integration changes outputs and prices we collect data from laboratory markets. Subjects acting as upstream or downstream producers choose outputs in a repeated game. In a control treatment, individuals participate in either an upstream duopoly market or a downstream duopoly market, but not both; we term this structure vertically linked. In a second treatment, corresponding to vertical integration, subject pairs make output choices in both the upstream and downstream market. To make our games mathematically equivalent to infinitely repeated games with discounting, we invoke a random stopping rule after 35 choice periods ( Fudenberg and Tirole, 1989, Gibbons, 1992 and Rasmusen, 1994). The use of a random stopping rule to mimic an infinite horizon super game is relatively common in experimental designs ( Plott, 1989). The termination probability we use, 1/5, is small enough to support a wide range of collusive super game strategies, but large enough so that different sessions would be expected to have similar length and with similar average earnings. 4 Comparing behavior in these two treatments, we find substantially larger levels of output in the upstream market and slightly greater outputs in the downstream oligopoly market under the vertically integrated structure. While the second of these effects is not statistically significant, the first is. If output in the X market rises significantly and output in Y only rises slightly, it can inferred that output in the Z market increases significantly. Overall, the hypothesis that vertically linked markets operate at levels identical to those in vertically integrated markets is rejected with great confidence. Moreover, integration lowers industry profits, and by an amount substantially larger than would be implied by the Cournot model. We also find that downstream markets tend to equilibrate more rapidly under integration compared to the linked control markets.5 This last observation is consistent with the perspective that integration allows firms to reduce upstream price uncertainty which facilitates convergence to downstream equilibrium. Taken as a whole, our results support the hypothesis that vertical integration increases total surplus. Indeed, we find that consumer surplus increases about 12% and total surplus rises 3% after integration. While it may be tempting to draw broad conclusions from these results, one must bear in mind that we are analyzing one specific parameterization in which data are gathered from experimental markets. Our discussion depends on a large alternative downstream market, and abstracts from a variety of institutional features that can influence the effect of vertical integration on output levels. For instance, we ignore effects related to technology and we do not delve into the vast literature on transactions costs and principal-agent problems. Nevertheless, we believe our results shed some light on the potential effects of vertical integration. The remainder of the paper is organized into five sections. In Section 2we describe a simple model of two vertically related markets, which forms the basis of our experimental design. Section 3provides a description of our experimental procedure and some summary results from the experiments. Econometric analysis of the data is discussed in Section 4, where we demonstrate the main findings of the paper. Concluding remarks are offered in Section 5.
نتیجه گیری انگلیسی
Our experimental analysis of vertically related markets yields a variety of results, some of which are unexpected. Subjects chose substantially larger outputs in upstream markets when placed in a vertically integrated structure. This is consistent with theory since integration allows firms to eliminate double marginalization. However, the increase in upstream output exceeds the theoretically anticipated change, by an amount that is statistically significant. At the same time, industry profits fell under integration. This is consistent with the noncooperative (Cournot) model, but again the change exceeds the theoretically predicted effect. Despite these lower profits, total surplus was larger in the vertically integrated markets, because of the marked increase in consumer surplus. We also find that downstream markets tend to stabilize more quickly in the integrated design than in the linked design. This may be the result of reduced uncertainty concerning upstream outputs, and the attendant reduction of uncertainty with respect to downstream costs. To the extent that agents are risk averse, this reduction in uncertainty provides an additional benefit to integration, which reinforces the increase in total surplus. Altogether, then, it appears that vertical integration had a procompetitive effect in our experimental markets. For roughly three decades following the Alcoa decision (United States v. Aluminum Co. of America, 148 F.2d 416 (2d Cir.1945), public policy was not sympathetic to vertical integration.18 More recently, antitrust authorities have been less sceptical of mergers. Indeed, the 1982 and 1992 revisions of the Department of Justice's merger guidelines created a policy environment in which mergers are less likely to be challenged, unless there are obvious anticompetitive consequences. In particular, vertical mergers are subject to less scrutiny than in earlier days. Despite this change in policy perspective, there are still some concerns that vertical mergers may facilitate collusion.19 This paper demonstrates that integration under highly concentrated conditions need not lead to higher prices, and as a consequence can benefit the consumer.