عملکرد ادغام تحت دستاوردهای بهره وری نامشخص

In view of the uncertainty over the ability of merging firms to achieve efficiency gains, we model the post-merger situation as a Cournot oligopoly wherein the outsiders face uncertainty about the merged entity's final cost. At the Bayesian equilibrium, a bilateral merger is profitable provided the non-merged firms sufficiently believe that the merger will generate large enough efficiency gains, even if ex post none actually materialize. The effects of the merger on market performance are shown to follow similar threshold rules. The findings are broadly consistent with stylized facts. An extensive welfare analysis is conducted, bringing out the key role of efficiency gains and the different implications of consumer and social welfare standards.

Mergers and acquisitions constitute a major feature of the economic landscape of most industrialized countries. To provide an idea of the resources involved, over the period 1981–1998, there were nearly 70,000 merger announcements worldwide, each worth at least 1 million U.S. dollars, of which nearly 45,000 were actually implemented. The average deal was valued at 220 million (base year 1995) U.S. dollars (Gugler et al., 2003, henceforth GMYZ). Mergers have been an important source of increase in market concentration, particularly outside the U.S. (Schmalensee, 1989). An extensive empirical and theoretical literature has explored the motives behind mergers and their impact on business activity. While both approaches have yielded useful insights, allowing industrial economists to reach a consensus on some aspects of merger performance, important discrepancies exist between key theoretical findings and stylized facts from empirical and event studies. By their very nature, mergers pose a complex conceptual challenge, wherein structure and conduct are inextricably intertwined. The pioneering work of Salant, Switzer and Reynolds (1983), henceforth SSR, showed that in the context of a symmetric Cournot oligopoly with linear demand and costs, for a merger to be profitable, it should comprise a pre-merger market share of at least 80%. This result forms the so-called “merger paradox”. Allowing the merging firms to exploit production synergies in some way, thereby lowering their post-merger costs, leads to a wider scope for profitable mergers (Perry and Porter, 1985, Farrell and Shapiro, 1990 and McAfee and Williams, 1992). A similar result holds under sufficiently concave demand (Fauli-Oller, 1997). By contrast, postulating Bertrand competition with differentiated products, Deneckere and Davidson (1985) establish that every merger would be profitable.1 While some degree of controversy, mostly of a quantitative sort, persists, the empirical literature has delineated some important stylized facts. On the key issue of profitability, in the largest cross-national study to date, GMYZ reports that nearly 60% of all horizontal mergers were profitable, with this proportion being higher in services than in manufacturing. As for sales (or revenues), it is essentially the other way around, with nearly 60% of merged firms experiencing a drop in sales. A similar negative effect is also reported for the post-merger market shares of the merged firms (Mueller, 1985). On the other hand, two other broad-based studies concluded that the profitability of acquired firms declined after the merger for U.K. firms (Meeks, 1977) and for U.S. firms (Ravenscraft and Scherer, 1987).2 The overall conclusion one can draw from this rather mixed picture is that while horizontal mergers have a limited negative impact on sales and market share, they do not appear to have, on average, a clear-cut effect on profitability. It is widely held that mergers typically lead to price increases.3 For example, Kim and Singal (1993) find a 10% increase for airline mergers.4 Regarding the effects on share prices, initially the target firm's shareholders earn a substantial premium of about 30% on the merger while those of the acquirer tend to have more variable fortunes, with an average on the low side (Mueller, 1985). For the merged firm, the overall initial effect is a substantial rise in share value, which however turns into a subsequent fall in value a few years after the merger. For firms outside a merger, the evidence does not seem conclusive for recent times, but Banerjee and Eckard (1998) report significant losses of about 10% for the merger wave at the turn of the 19th century in the United States. As to the crucial issue of whether mergers generate efficiency gains, the evidence is not direct as such gains are difficult to estimate, but rather deductive.5 While many studies, including Ravenscraft and Scherer (1987), report little support for a positive relationship, GMYZ concludes that 29% of all mergers engendered efficiency gains, as suggested by observed increases in both profits and sales. Naturally, it is very difficult to disentangle the efficiency gain and the market power effects due to a merger. On the other hand, there appears to be a consensus reached on the basis of case studies and casual observation that while some mergers were successful in securing substantial efficiency gains,6 there is great variability on this issue. In view of the lack of congruence between theoretical and empirical findings,7 the primary challenge of theoretical work on mergers is to come up with alternative models of merger behavior that would close this gap. This paper constitutes an attempt in this direction within the framework of static analysis. The novel ingredient is that all the firms in the industry face uncertainty as to the efficiency gains, in terms of variable costs, that the merged firm could achieve. The efficiency gains may correspond, for example, to the claim made by the merging firms to the antitrust agency, possibly appropriately discounted by the rival firms, or to a past average achieved by comparable mergers in related industries. Pre-merger competition is modelled as a standard Cournot oligopoly with identical firms while short-run post-merger competition involves a Bayesian Cournot model, with the merged firm alone being informed about its true cost. As simplifying assumptions, we take demand and costs to be linear, and the uncertainty to be binomial. This simple formulation seems appropriate in view of the stylized facts on mergers. Indeed, for the merger to obtain antitrust approval in most countries, the candidate firms have to convincingly document scope for significant efficiency gains, via the exploitation of organizational and production synergies. In most cases, the approval of a merger presumes that the antitrust authority has been swayed by the firms' claims of lurking efficiency gains. Likewise, the initial surge in share prices provides some support for the presumption that the merger is likely to lead to strong efficiency gains, as an increase in market power alone would be unlikely to yield the concomitant increase in expected profits. Another point is that the firms in the industry frequently react with apprehension to a merger announcement by two of their rivals. These typical facts lend credence to the postulate that all concerned parties generally hold beliefs about the prospect of efficiency gains that are naturally captured by a Bayesian model. Indeed, the revised Section 4 of the Horizontal Merger Guidelines issued by the U.S. Department of Justice and the Federal Trade Commission in 1997 states that “efficiencies are difficult to verify and quantify, in part because much of the information relating to efficiencies is uniquely in the possession of the merging firms. Moreover, efficiencies projected reasonably and in good faith by the merging firms may not be realized”.8 Further discussion in support of our Bayesian setting is given in Section 5.2. Recent studies have also proposed settings where uncertainty plays a key role. Chone and Linnemer (2008) analyse a very general model with multi-product firms and uncertainty over efficiency gains of the merged firms. In contrast to the present paper, this uncertainty fully resolves before post-merger market competition takes place, but nonetheless affects the ex ante social welfare of the merger for antitrust authorities, and hence their approval decision. Depending on the nature of competition and on the demand specification, social welfare may be convex (e.g. for linear demand) or concave in these gains, so that uncertainty may enhance or lower the approval chances. In a setting with uncertainty over demand or costs, Banal-Estanol (2007) shows that there are added incentives for mergers arising from merged firms sharing their private signals.9 One of our main results states that if the non-merged firms believe with a sufficiently high probability that the merged firm will experience a high enough efficiency gain, the merger will be profitable, even if one takes the worst-case scenario for the merged firm, wherein it ends up not experiencing any efficiency gain at all. Similar threshold rules are shown to govern the effects of a merger on the merged firm's and outsiders' outputs as well as on industry price, using worst case, best case and expected term scenarios. In all theoretical models with complete information and no efficiency gains, whether based on Cournot or on Bertrand competition, mergers always exert a positive externality on non-merged firms. In a Bayesian formulation, the nature of this externality also follows a threshold rule depending on the same pair of parameter values, so that it may well be negative. Similar remarks may be made about market shares and sales. The set of possible outcomes following a merger is substantially expanded, with one or both the merged firm and the outsiders, or neither of them, being possible beneficiaries. For both consumer surplus and social welfare, the worst-case benchmark yields a negative effect of mergers while the two benchmarks lead to thresholds depending again on the belief and the efficiency gain levels. The threshold rule associated with the ex ante and best case benchmarks confirms the central role played by expected efficiency gains in gauging the welfare effects of mergers, as in common antitrust practice in many countries. Another main conclusion of the paper is that an ex-ante profitable merger is necessarily social-welfare, but not always consumer-welfare, improving. This result provides support for a laisser-faire policy if the decisive criterion rests on social welfare, but not if it rests on consumer welfare. This underscores the importance of the selection of a decision criterion for antitrust approval of a merger. The present set-up also demonstrates that the merging firms have a strong incentive to overstate the extent of their potential efficiency gains ex ante, not only to secure approval of the merger by antitrust authorities, but also to twist the terms of Bayesian Cournot competition in their favor, in a short-run perspective. All in all, our results form a major departure from the complete information equilibrium analysis of the literature starting with SSR. Particularly noteworthy is the fact that the novel features of the present paper hold even in the worst-case ex post outcome of no efficiency gains. In such a case, the only difference between the post-merger markets in this paper and in SSR is the first-to-know advantage of the merged firm inherent in the Bayesian Nash concept. An important consequence of this difference is that, unlike most previous theoretical results, our conclusions are quite consistent with many empirical findings and stylized facts on the effects of mergers on profitability, sales and market shares, both for the merged firm and for the outsiders to the merger. This first-to-know advantage thus emerges as a natural candidate for the fundamental asymmetry that mergers seem to trigger in favor of the merged firm. By its very nature, this new type of asymmetry is transitory, as are most investigated effects of mergers. In this sense, the present theory constitutes a short-run analysis, but the short run is where most of the interest in mergers actually lies. 10 In addition, we can add a plausible dynamic extrapolation of our model to capture the resolution of uncertainty over the merged firm's cost. Our results are consistent with GMYZ's finding that over their five-year data window, from one year to the next, realized profits increased for profitable mergers but decreased for unprofitable mergers (see Section 5 where a dynamic extension of the model is discussed). A similar mechanism may be invoked to account for the initial substantial rise in share values that typically accompanies merger announcements, which often ends up spiraling downwards after one to three years. This paper is organized as follows. After setting up the model in Section 2, the effects of mergers on market performance are presented in Section 3, followed by a detailed welfare analysis in Section 4. Section 5 is devoted to some dynamic extensions. All computations, proofs and quantitative illustrations are gathered in an Appendix A.

This paper argues that many of the circumstances surrounding mergers call for a theoretical model wherein the firms outside the merger face a new type of rival, characterized by unknown unit costs, reflecting their natural initial uncertainty about the ability of the merged firm to achieve any (of the claimed) efficiency gains. This pervasive uncertainty also affects the approval decision of antitrust authorities, and triggers the favorable response by financial markets. Within the obvious confines of a static model, the proposed Bayesian Cournot equilibrium leads to an outcome that is broadly consistent with much of the empirical evidence on the industry effects of mergers, on profits, price and market shares for the merged firm as well as for outsiders, at least in the short run. All in all, the model at hand reflects a simple and natural modification of the standard Cournot approach, based on an inherent informational advantage of the merged firm over outsiders, bringing about a surprising level of congruence with stylized facts. In terms of welfare, mergers lower consumer and social welfare for sure only in the worst-case scenario. In the other two scenarios, welfare depends on the levels of belief and the efficiency gain. An ex-ante profitable merger is necessarily beneficial for expected social welfare but not necessarily for expected consumer welfare. Overall, these results vindicate the central role assigned to efficiency gains in merger policy.