Merger waves: a model of endogenous mergers.

We develop a model of endogenous mergers to study their dynamic process. Firms choose whether, when, and with whom to merge. Two necessary conditions are identified for mergers to occur: firm heterogeneity and negative demand shocks. We show that mergers are strategic complements and therefore tend to occur in waves. Moreover, some mergers occur for strategic reasons in order to precipitate further mergers.

1. Introduction

* Mergers have become increasingly widespread in recent years. According to the UN's World Investment Report (UN, 2000), worldwide mergers and acquisitions (M&A) grew at an annual rate of 42 percent over the period 1980-1999 to reach $2.3 trillion in 1999. More than 24,000 M&A took place during that period, and the value of M&A relative to world GDP rose from 0.3 percent in 1980 to 2 percent in 1990 and to 8 percent in 1999. Standard & Poor's has predicted that consolidation through M&A would reduce the number of auto companies from 40 in 1998 to about 20 in the 21st century. (1)

Economic studies of horizontal mergers have focused mainly on two questions: why firms merge, and how they merge. To answer the first question, researchers have typically assumed a single merger, which is decided by a number of designated firms collectively. Nonparticipants are supposed to remain independent. (2) Because the merger structure (who merges with whom and who remains independent) is exogenously imposed on the firms, mergers modelled in such a way are called exogenous mergers. To address the second question of how firms merge, the exogenous merger structure must be abandoned and the merger process must be modelled explicitly. In particular, firms must make their merger decisions individually. Mergers that result from such a process are called endogenous mergers. Once firms are allowed to make individual choices, the resulting industry dynamics are greatly enriched, as multiple mergers may occur and firms may merge in response to some other mergers.

In this study, we present a model of endogenous mergers and study how firms merge in a dynamic process. Firms play a two-stage game. In the first stage, mergers occur sequentially in an endogenized order. Each merger is between one proposing firm that is drawn randomly and one target firm that is chosen by the proposer. After a merger is completed, another randomly drawn firm may propose another merger. This process continues until no further mergers occur, which ends the first stage. In the second stage, the surviving firms engage in Cournot competition and receive their payoffs.

We identify two necessary conditions for mergers to occur. Mergers occur only if firms have different marginal costs and the industry has experienced a shock that reduces demand. In Cournot competition, the profitability of any given merger depends on the interaction between two forces: the merging firms internalize the competition between themselves, which benefits them, and the nonmerging firms free-ride on the reduced competition by competing more aggressively, which hurts the merging firms. If firms are homogeneous with constant marginal costs, Salant, Switzer and Reynolds (1983) have shown that the second force dominates and, therefore, a two-firm merger will never occur. If firms are heterogeneous with different marginal costs, as in our model, merging firms will improve their production efficiency through technology transfer and therefore receive some extra benefit. For a given cost differential between the merging firms, the benefit is relatively large if the market size is small. A reduction of the market size, resulting from a negative demand shock, may therefore turn an unprofitable merger into a profitable one and cause the merger to occur.

The extensive-form game that we use allows us to characterize the equilibrium merger strategies, which usually lead to a unique path of mergers. For example, in a four-firm industry, we show that a negative demand shock will lead to a merger between the two firms with intermediate efficiency, followed by another merger between the most- and least-efficient firms. Indeed, such a pattern can be found in the real world. In 1986, the four largest brands in the U.S. carbonated soft drink industry were Coca-Cola, Pepsi, Seven-up and Dr Pepper, with respective retail sales shares of 37.4%, 28.9%, 5.7% and 4.6% (White, 1989). In January of that year, Pepsi announced its intention to purchase Seven-up. Three weeks later, Coca-Cola announced its intention to purchase Dr Pepper. Although neither merger materialized due to antitrust objections, (3) the sequence of announced mergers matches the predictions from our analysis.

We find that mergers are strategic complements in the sense that firms' incentives to merge increase when some other firms also merge. This is because other mergers reduce the number of free riders for a given merger, making the merger more profitable. The complementarity between mergers implies that forward-looking firms may engage in mergers strategically. That is, firms may carry out an otherwise unprofitable merger in order to facilitate some further mergers that might otherwise not occur. We demonstrate the presence of strategic mergers in many cases. Because mergers are strategic complements, they tend to occur together, leading to a merger wave. Hence, our study offers an explanation for the well-documented observation that mergers tend to occur in waves, one of the "most consistent empirical features of merger activity over the last century" (Andrade, Mitchell, and Stafford, 2001, p. 104).

The predictions of our model are supported by empirical and anecdotal evidence. In a cross-industry empirical study of takeover activity in the 1980s, Mitchell and Mulherin (1996, p. 197) related many mergers to negative industry shocks: "A shock-driven decline in demand can ... pressure firms to merge.... " Dutz (1989) presented evidence of mergers in the steel industry as it faced declining demand. In the popular media, declining demand is often given as one of the major reasons behind some industries' merger waves. Furthermore, industrial analysts often view certain mergers as a response to other mergers in the same industry. For example, in the oil and petroleum industry, oil prices plummeted in the late 1990s due to decreased demand and overproduction. Then, the biggest firms in the industry began seeking large-scale consolidations. British Petroleum and Amoco were two of the first firms to pursue such a move in August 1998. A few months later, Exxon and Mobil began their $88 billion merger, the largest in U.S. corporate history. A few weeks later, France's Total SA and Belgium's Petrofina SA joined the consolidation frenzy. In March of 1999, BP Amoco and Atlantic Richfield unveiled their $25 billion merger plan (Hill, 1999). Commenting on the merger between Conoco and Philips Petroleum in 2001, an article in The Economist (November 22, 2001, p. 60) remarked "it is surely no coincidence that the previous wave of mergers swelled just as oil prices collapsed to around $10 a barrel." (4)

The growing literature on endogenous mergers is still small. Previous contributions have been made by Kamien and Zang (1990), Barros (1998), Gowrisankaran (1999), Fauli-Oller (2000), and Gowrisankaran and Holmes (2004). (5) Due to the complexity of the problem, the merger process in most models is endogenized only partially. For example, researchers have restricted the number of firms to three (Barros, 1998) or four (Fauli-Oller, 2000), or assumed that mergers occur in a predetermined order (Gowrisankaran, 1999; Fauli-Oller, 2000). (6) Our study, by contrast, attempts to model endogenous mergers more completely. We assume an arbitrary number of firms that may differ in their marginal costs, and we endogenize the order of mergers. Although Kamien and Zang (1990) modelled the merger process in a general way, because their firms are identical, they missed the more interesting question of which mergers will occur. They concluded only that full monopolization through mergers will not happen.

Although merger waves prevail in reality, few existing models are sufficiently rich to explain them. By modelling mergers in a simultaneous game, Kamien and Zang excluded the possibility of strategic mergers and merger waves. While Gowrisankaran (1999) assumed that mergers occur sequentially, his analysis focused on industry dynamics through merger, investment, entry and exit with random returns, rather than on the strategic interaction between mergers. To our knowledge, Fauli-Oller (2000) has provided the only theoretical framework for the study of strategic mergers, (7) but his merger game (two efficient firms take turns to bid for two inefficient firms) is not completely endogenous. We analyze strategic mergers when the merger structure and sequence are both endogenous, thus providing a more realistic setting for the study of merger waves.

The article is organized as follows. After setting up the model in Section 2, we discuss firm strategies and the profitability of mergers in Section 3. These results are needed in later analysis to derive the equilibria. In Section 4, a four-firm industry is thoroughly analyzed as a special case. We show how to derive the equilibrium. All of the major results in the study, including negative demand shocks as causes of mergers, strategic mergers and merger waves, are present in this special case. Finally, Section 5 presents the results for the general case and discusses their implications. Section 6 concludes. All proofs are provided in the Appendix.

2. The model