SSRN-id1407747

经济学之反垄断(英文文献)

Horizontal Mergers, Collusion, and Stockholder Wealth

B. Espen Eckbo?

University of British Columbia, Vancouver, BC V6T 1Y8, Canada

November 1981

Forthcoming in the Journal of Financial Economics

JEL classifications: G34, G38, L12, L41 Keywords: Merger, collusion, market power, antitrust, efficiency, rivals, industry rents

Abstract

This paper tests the hypothesis that horizontal mergers generate positive abnormal returns to

stockholders of the bidder and target firms because they increase the probability of successful

collusion among rival producers. Under the collusion hypothesis, rivals of the merging firms

benefit from the merger since successful collusion limits output and raises product prices and/or

lowers factor prices. This proposition is tested on a large sample of horizontal mergers in

mining and manufacturing industries, including mergers challenged by the government with

violating antitrust laws, and a ―control‖ sample of vertical mergers taking place in the same

industries. While we find that the antitrust law enforcement agencies systematically select

relatively profitable mergers for prosecution, there is little evidence indicating that the mergers

would have had collusive, anticompetitive effects.

This paper, which is based on my University of Rochester dissertation, has benefitted greatly from the ―visible hands‖ of my thesis committee: Gregg A. Jarrell, Michael C. Jensen and G. William Schwert. I have also received helpful comments from Fisher Black, Claudio Loderer, Avner Kalay, Wayne Mikkelson, Rex Thompson, Jerry Warner, the participants of the finance and industrial economics workshops at New York University, University of British Columbia, and University of Chicago, and the referee, Eugene Fama. The financial support of the Norwegian School of Economics and Business Administration, and the Center for Research in Government Policy and Business at the University of Rochester, is gratefully acknowledged. Electronic copy available at: http://wendang.chazidian.com/abstract=1407747

经济学之反垄断(英文文献)

1 Introduction

The merger literature contains a substantial amount of evidence indicating that stockholders of merging firms earn positive abnormal returns from merger activity.1 A standard interpretation of this evidence is that control over the target firms resources enables the successful bidder to initiate a revaluation of its own (as well as the targets) shares by implementing a higher-valued operating strategy. Following this view, the stockholder gains reflect an increase in the expected spread between the merging firms future revenues and costs. However, the more difficult issue of whether the gains predominantly originate in cost-side effects (―productive efficiency‖ theories) or in revenue-side effects (―market power‖ theories) has remained unanswered. Indeed, despite the widespread public concern with allegedly anti-competitive consequences of mergers, reliable evidence on the importance of market power theories in the context of this particular corporate activity is almost non-existent.

This paper examines a necessary condition for the proposition that horizontal mergers have collusive, anticompetitive effects. In our context, the central characteristic of the collusion theory is its implication for mergerinduced changes in relative product (and factor) prices. However, noting that changes in product prices induce changes in the market value of firms trading at these prices, we instead focus on the abnormal stock returns to the merging firms and their horizontal rivals. There are several reasons why this focus is of particular interest. For example, the potential social welfare loss from post-merger collusion on price can be entirely offset by an increase in competition on non-price variables, such as product quality and service. While product prices are not necessarily sensitive to changes in non-price competition, under the efficient markets/rational expectations hypothesis stock prices reflect the combined effect of all changes in the firms expected future cash flows. At a minimum, it is this combined effect which should govern a decision to challenge the merger under antitrust laws. Furthermore, while it is difficult to obtain a precise estimate of the time of the mergers impact on product prices, it is well established that stock prices react quickly to public merger announcements. Moreover, while there exists no generally acceptable theory generating ―normal‖ or expected values of product prices, the finance literature provides a model for equilibrium expected stock returns. Finally, the availability of stock price data A brief discussion of some of this evidence, and how it relates to the findings of this paper, is given in Section 4.

1

Electronic copy available at: http://wendang.chazidian.com/abstract=1407747

经济学之反垄断(英文文献)

encourages the use of sample sizes which would be infeasible if we were to rely on product prices, or other firm or industry specific characteristics.

With a sample of 259 horizontal and vertical mergers in mining and manufacturing industries, of which 76 were challenged by the government under claims they ―monopolized‖ product markets, we find that the collusion hypothesis is generally rejected by the data. Rather, the evidence is consistent with the proposition that antitrust policy over the past two decades in part has protected relatively high-cost from relatively low-cost producers by restricting the opportunity to implement lower-cost production techniques by means of merger. This conclusion, which is also to some extent supported by Stillman (1983), casts serious doubt on the validity of a ―consumer protection‖ rationale for this form of government intervention in the market for corporate control.

The paper is organized as follows: Section 2 discusses the testable implications of the collusion hypothesis. Section 3 describes the procedure used to select the merger sample and the portfolio of horizontal rivals for each merger. Section 4 contains the empirical results, and relates the evidence to the findings of previous work. Section 5 concludes the paper.

2 Testable implications of the collusion hypothesis

Merger-related anticipated changes in product or factor prices translate into merger-related abnor- mal performance by the direct competitors of the bidder and target firms. Below, the implications of the collusion hypothesis are stated in terms of this abnormal performance in response to two consecutive public announcements, each significantly changing the probability that the merger will take place. The first is the merger proposal announcement, the second is the announcement that a ―Section 7‖ complaint against the merger has been (or will be) filed by the Federal Trade Commission or the Antitrust Division of the Justice Department.2

2.1 Implications for the performance of the rivals

The traditional collusion argument presumes the incentive to coordinate the production rates of the individual rivals within an industry is a function of the costs of monitoring the collusive agreement. Section 7 of the Clayton Act prohibits one corporation from acquiring the stock or assets of another ―if the effect of such acquisition may be substantially to lessen competition or tend to create a monopoly‘. The merger proposal and the antitrust complaint announcement dates are taken from the Wall Street Journal. In the remainder of the paper, the phrase ―antitrust challenge‘ refers to the antitrust complaint. 2

2

经济学之反垄断(英文文献)

Using Stigler (1964) theory of oligopoly, a horizontal merger can reduce the monitoring costs by reducing the number of independent producers in the industry. The fewer the members of the industry the more ―visible‖ are each producers actions, and the higher is the probability of detecting members who try to cheat on the cartel by increasing output. The higher this probability, the lower the expected gains from cheating, and the more stable (and profitable) is the cartel in the short run.3

Since effective collusion generates monopoly (or monopsony) rents, the collusion hypothesis

implies that the horizontal rivals of the merging firms should earn positive abnormal returns around the merger proposal announcement. The same conclusion holds for rivals expected to remain outside the collusive agreement, in particular since these firms will not bear the costs of restricting output (they are ―free-riding‖ on the higher product price). Conversely, the rivals should earn negative abnormal returns in response to the news of a subsequent antitrust complaint, provided the complaint is expected to significantly increase the costs of collusion (for example, by prohibiting the merger from taking place). Such a complaint announcement will reverse the expectations of increased monopoly (monopsony) rents caused by the earlier merger proposal.

2.2 Productive efficiency and implications for regulation

The above implications of the collusion hypothesis are necessary but not sufficient to conclude a given merger is truly anticompetitive. As indicated in Table 1, a pattern of abnormal returns to the merging firms and their rivals which is consistent with collusion can also be consistent with productive efficiency. The latter hypothesis represents a class of theories predicting an increase in the market value of the merging firms due to the implementation of a more cost-efficient produc- tion/investment policy after the merger is consummated.4 In general, the efficiency hypothesis does not restrict the sign of the abnormal returns to the rivals. To see why, note that with productive Of course, in the absence of government supported entry barriers (such as patents, licences, tariffs, etc.), the collusion argument assumes the degree of resource specialization in the industry is sufficient to slow down the entry process. See, for example, Stigler (1950) for a discussion of the minimum necessary conditions for merger-for- monopoly (or oligopoly) to take place. Note also that the collusion hypothesis does not necessarily presume a complete cartelization of the industry. A subset of firms may find it optimal to form a cartel agreement after the merger has been completed and produce a marginal output (or input) restriction on their own, a scenario which is analytically equivalent to the classical ―dominant firm‖ or ―price umbrella‖ model. 4The productive efficiency hypothesis covers a wide range of possible specific reasons for merger, among others, realization of technological complementarities, replacement of inefficient management teams, utilization of unused corporate tax credits, and avoiding bankruptcy costs. A general review of traditional non-monopolistic hypotheses of merger motivation can be found in Steiner (1975). 3

3

经济学之反垄断(英文文献)

efficiency each of the two merger-related announcements can have a product/factor price effect and a possibly offsetting information effect. That is, the intensified competition in product and factor markets (the merging firms are being replaced by a more competitive corporate entity) tends to result in lower product prices and higher factor prices. This price effect causes a negative change in the market value of the rivals at the time of the proposal announcement, and a positive (reversed) effect at the time of the antitrust complaint. On the other hand, since the production technologies of close competitors are (by definition) closely related, the news of a proposed efficient merger can also signal opportunities for the rivals to increase their productivity.5 Similarly, the news of the an- titrust complaint can signal a significant restriction in the future merger opportunities of the rivals (cf. ―landmark‖ cases). For each of the two announcements the total wealth impact on the rivals is the sum of the product/factor price effect and the (possibly offsetting) information effect, leaving no necessary restriction on the sign of the rivals abnormal returns under the efficiency hypothesis6

For the purpose of drawing normative conclusions concerning merger regulation, a further limi- tation of the tests should be emphasized. The collusion and efficiency hypotheses are not mutually exclusive, which means the observed security value changes resulting from a given merger can rep- resent the sum of simultaneous positive and negative effects due to collusion and efficiency. In principle, the dollar value of the efficiency gains realized within the merging firms can outweigh the negative social welfare effects of collusion. Therefore, even a pattern of abnormal returns to the rivals which is truly consistent with the collusion hypothesis is not sufficient evidence to conclude that public regulation of the merger will increase social welfare. On the other hand, and this con- stitutes the basic motivation of the paper, evidence which according to Table 1 is inconsistent with collusion, but consistent with value-maximizing behavior on the part of the merging firms, implies that blocking the merger will reduce social welfare. Essentially, such evidence would indicate that For example, the proposal announcement may disseminate information which enables the rivals to imitate the technological innovation motivating the acquisition. If such innovation activity requires merger, then the stock prices of the rivals will be bid up in anticipation of the expected gains from the future merger activity. Interestingly, Jarrell and Bradley (1980) presents evidence consistent with the proposition that the introduction of public disclosure laws has resulted in extensive dissemination of technological information associated with tender offers, thereby significantly reducing the private gains from company takeovers. Note also that if the technological innovation is scale increasing, then imitation by the rivals will further reduce the product price (and increase prices of specialized inputs). In fact, merger waves may be a race by imitators to lower their costs in response to this continuing price decrease. 6In principle, one could discriminate between the collusion and efficiency theories by examining the abnormal returns to the merging firms corporate customers and suppliers of inputs. For example, relative to the proposal announcement, corporate customers and suppliers should lose under the collusion hypothesis and gain under the efficiency hypothesis. However, tests based on this notion are difficult since it is necessary to identify customers and suppliers who cannot switch their purchases/sales to other industries at a low cost. 4