Abstract
Efficiencies have long been an important aspect of the antitrust analysis of mergers, but in recent years there have been certain noteworthy changes in the nature and analysis of efficiencies. One change appears to be the greater ability of the antitrust agencies to evaluate conventional efficiencies carefully and critically. This is demonstrated by the analysis of efficiencies associated with a number of recent proposed mergers before the antitrust and regulatory agencies. The second important change concerns the types of efficiencies typically claimed by merging firms. Whereas in earlier years these largely involved scale economies, they now more often focus on dynamic efficiencies, quality benefits, and vertical economies. This article discusses each of these in turn and argues that these newer types of efficiencies pose new and greater challenges for analysis.
I. Introduction
The 2010 Horizontal Merger Guidelines issued by the Federal Trade Commission and the Antitrust Division of the Department of Justice were the culmination of an extensive effort by the antitrust agencies to incorporate the best of modern merger analysis from economics and to provide the best operational guidance to companies, their attorneys, and other outside parties with respect to the application of these analytical techniques. The Guidelines in fact broke new ground relative to the earlier Guidelines issued in 1997 in several important respects: They set out a more eclectic and less formalistic analytical approach than the approach in earlier versions. They emphasized competitive effects over the calculation of market shares and concentration, although they reinstated the presumption that high shares and concentration signaled competitive concerns. They incorporated specific elements of unilateral effects analysis, including the diversion ratio and upward pricing pressure. And they restated a concern over mergers that eliminated potential competition, a proviso that had disappeared with the 1992 Guidelines. 1
But even as much changed in these areas, the 2010 Guidelines suggested only modest change with respect to one major topic—efficiencies. The discussion in the new Guidelines included, with perhaps one exception, no new concepts, no new analytical approaches, no new criteria. That exception involved the integration of efficiencies into the analysis of market power through the use of “upward pricing pressure.” While the formal model is limited to mergers involving differentiated products, this integration is a useful step in developing a more integrated approach. Otherwise, the 2010 Guidelines addressed efficiencies largely with the same language as previously. Few readers of these Guidelines would disagree with one of their architects, who has said, “The 2010 Guidelines make very few changes to the treatment of efficiencies articulated in 1997.” 2
Despite the lack of formal change in the Guidelines, however, there have been some important changes in the merger efficiencies area in the preceding fifteen to twenty years. Three such changes are the focus of this article. First, during this period, the nature of efficiencies claimed by merging parties has evolved, emphasizing new types of cost and consumer benefits. Second, the antitrust agencies have become more skilled in their methods of analysis of certain efficiencies, prompting improved claims by merging parties and generally strengthening evaluation. Third, even as agencies have become more skilled at evaluating many efficiencies, newer types of efficiencies pose greater challenges for agency evaluation. This article will develop these points, beginning with a discussion of traditional efficiencies and their measurement, and then setting out three types of efficiencies that are now more often advanced as justifications for mergers. These involve dynamic considerations, quality benefits, and vertical economies. In each case, the greater challenges of measurement, and even identification, of efficiencies will be highlighted. A final substantive section will discuss how these and other issues converge in the case of mergers in network industries, a setting that has seen an increasing number of significant mergers.
II. Efficiencies, Then and Now
From its origins with Williamson’s analysis to the Merger Guidelines, the benefits of mergers have been construed as some type of unit cost savings, and merger analysis has been geared to evaluating such claims. The central feature of Williamson’s analysis is readily described.
3
As shown in Figure 1, a market is assumed to be in competitive equilibrium at P
1 and Q
1, where P
1 equals marginal production cost C
1. A merger occurs that has two simultaneous effects. The first is that price rises to P
2 as the result of market power, leading to deadweight loss given by the area D. The second assumed effect is that unit costs fall to C
2. The lower production cost is a total cost saving in the amount S. The trade-off created by the merger is then between the social benefit S and the social loss D. In a simple setting,
4
it is readily shown that the cost saving benefits outweigh deadweight loss (that is, S > D) under the following condition: The Williamson trade-off.
where C is total cost, η is demand elasticity, and m is the percent markup. This criterion is easily met: For example, if η = 2 and m = .10, then a 1% cost saving (S/C) suffices to justify the merger. That is, a 10% price increase on consumers rise is offset by a mere 1% cost saving to the firm.
Williamson’s analysis is subject to a number of well-known qualifications and reservations. Among them are the following:
It assumes the entire premerger industry operates perfectly competitively, so that market price equals marginal cost. The latter is taken to be the same for all producers. It assumes the entire industry shifts to the higher postmerger price, that is, as if all firms merged. Alternatively, perhaps, the merger could be assumed to result in a uniformly higher price and uniformly lower cost for all firms, merging and nonmerging. Those scenarios are unlikely, and the model is silent as to the nature of market equilibrium under various assumptions about the costs and prices of other firms in the industry. It assumes that the merger is strictly necessary to achieve the efficiencies, whereas in many cases, alternatives such as contracting between the firms can achieve some or all of the cost savings. Determining the actual merger-specific cost savings therefore requires a comparison between two scenarios, neither of which is presently observed. It assumes that efficiencies are achieved concurrently with the price increase, although in reality most efficiencies are not realized until after the firms integrate operations, whereas price increases are easier to implement quickly. Analysis should therefore conduct the appropriate time discounting of these two effects. It assumes that the policy objective is maximization of total surplus, thereby allowing a dollar-for-dollar trade-off between savings S and losses D. Most countries' competition policies emphasize the effects of mergers on consumers, requiring some or all of the cost savings to be passed on to consumers, rather than remaining entirely with the firm.
These and other factors make clear that the simple description and trade-off of Williamson’s model are unrealistic. That said, many of these real-world qualifications can be accommodated within this framework by respecifying the elements of the basic model, for example, by making explicit assumptions about rivals’ behavior, adjusting for rising marginal costs, or accounting for the timing of effects. In any event, even if not straightforward in practice, this model highlights the importance of efficiencies that may be achieved through merger. This view has long been firmly embedded in the Merger Guidelines. Starting with the 1992 version, the Guidelines state that “[T]he primary benefit of mergers to the economy is their efficiency-enhancing potential, which can increase the competitiveness of firms and result in lower prices to consumers.” 5 This particular statement is a reminder that most mergers in fact do not create or enhance market power, so their net effect is simply whatever degree of cost savings that they achieve. 6 From a policy perspective, the vast majority of mergers require no review whatsoever.
But for those mergers that create or enhance market power, the 1992 Merger Guidelines directed attention to “economies of scale, better integration of production facilities, plant specialization, lower transportation costs, and similar efficiencies relating to specific manufacturing, servicing, or distribution operations of the merging firms.” 7 These Williamson-type efficiencies were those typically credited by the agencies at the time. Horizontal competitors with similar products would combine operations and seek cost savings from larger scale, production complementarities, avoidance of duplication, improvements in logistics, and so forth. These traditional efficiencies were characteristic of an economy still centered on manufacturing, where economies arose from consolidation of production and closely related activities. 8
To be sure, these types of traditional economies continue to be important in many present-day mergers and, as such, remain an integral part of modern merger analyses. Specialization of facilities, avoidance of duplication, and better utilization of production capacity, for example, have been central arguments in the merger of Evanston and Highland Park Hospital, in Heinz’s attempt to acquire Beech Nut’s baby food business, and in the Sirius and XM Satellite Radio merger. 9
As has been noted, conventional efficiencies are focused on greater realization of economies of scale in production. But since most mergers do not conveniently occur in industries where studies have been conducted, the more important development is the growing expertise of the antitrust agencies in methods of estimating many traditional efficiencies from mergers. Staffs know better what to look for in the evidence presented by merging parties, how to probe the evidence for weaknesses, how to frame follow-up questions, and ultimately what claims to accept and which to discount or reject altogether. The result is that at present the types of claims and evidence that previously might have sufficed for policy determinations—assertions by company executives together with a smattering of data—no longer suffice. This in turn has elevated the evidence presented by merging parties on these issues and resulted in a considerably more sophisticated debate about scale and similar efficiencies.
Current methods and standards are nicely illustrated by the evidence presented and the analysis performed in the proposed merger of AT&T and T-Mobile. This merger would have combined two of the four national wireless carriers, raising fairly straightforward concerns about the likely price effects. 10 But the parties claimed substantial cost savings from utilizing each other’s existing capacity rather than having to expand their own respective capacities in order to serve growing demand. Their claims were evaluated by the Antitrust Division of the Justice Department and also by the Federal Communications Commission. The latter agency must approve any transfer of licenses, as would occur in this merger, and is guided by a statutory requirement that such an action must be “in the public interest.” One component of the public interest standard is concern over competition, and in this the agencies were said to cooperate in their analyses. The staff analysis conducted at the Federal Communications Commission (FCC) was placed on the record and gives considerable insight into how the efficiency issues were analyzed. The Justice Department came to the same ultimately negative conclusion about efficiencies as did the FCC but without disclosing the analytical basis for its conclusions. Given the cooperation between the agencies—facilitated by some personnel crossovers—the FCC’s detailed analysis can be taken as a good indication of how traditional efficiency claims are evaluated not only in a regulatory agency but also in the relevant competition agency.
As outlined by participants in the FCC analysis, 11 AT&T and T-Mobile argued that both their networks would soon become capacity constrained and that the cost of expanding each of their respective network infrastructures would be very expensive. The merger was said to alleviate this problem in two ways. It would allow AT&T to integrate some of T-Mobile’s currently underused cell sites into its own network and, in addition, would free up some spectrum by combining the two carriers’ legacy services. In support of these claims, AT&T built an engineering model to project the cost of expanding service capacity on its own network, and then compared that to the cost of meeting demand increases with the integrated system that would result from the merger, for fifteen major cities.
The demand side involved projections of overall growth in demand together with a model of traffic distribution to identify which cell sites were likely to become congested. For each of these, calculations were made for the incremental costs of alleviating that congestion first by deploying wider-area cell towers at existing sites or, if necessitated by substantial congestion, by the further and more expensive strategy of adding more antennae near the congested cell. These incremental costs were determined to be significantly higher when each carrier avoided congestion by itself than by doing so on an integrated basis. This constituted AT&T’s core argument for the merger.
In an earlier era, such a demonstration might have carried the day for the merging parties. But drawing on its engineering and economic expertise in wireless technology, the FCC conducted its own analysis of efficiencies and ultimately dismissed the parties’ claims. The FCC did not dispute that a single network might be more efficient, but it determined that the parties’ engineering model vastly overstated the benefits for a simple reason: that model assumed that the remedy for cell congestion was to place new towers throughout the city, rather than to concentrate them at the congested site. This algorithm for addressing congestion dramatically raised the cost of addressing congestion and more often required invoking the more expensive alternative strategies for single-firm operation, thus making the latter appear considerably more costly. When the FCC reran the engineering model with a different method of cell placement, most of the claimed benefits disappeared.
This example illustrates the increased capability of the reviewing agency to assess the validity of claimed efficiencies. In this case, of course, an expert regulatory agency (the FCC) was involved; but for many other industries, equivalent expertise resides within the antitrust agencies. The Department of Justice (DOJ), for example, has extensive experience with airline and electricity mergers, while the Federal Trade Commission (FTC) has long done work in the hospital and petroleum sectors, among others. To be sure, this example also illustrates the increased sophistication of parties’ submissions with respect to efficiencies, but of course, due to their advocacy orientation, that underscores the need for careful, critical review of those submissions.
Thus, the antitrust agencies would seem to have gained considerable expertise in evaluating conventional efficiencies such as changes in marginal costs due to increases in scale. In this, of course, they are in a race with the capabilities of merging parties to make credible claims of efficiencies, resulting in undue deference to such claims. 12 Of greater concern is the fact that the types of efficiencies commonly argued by merging firms are changing and that the agencies may once again lag behind firms’ capabilities and once again give unwarranted deference to such claims. This will be discussed in the following sections.
III. The Shifting Nature of Efficiency Claims
As noted, the changing nature of the U.S. economy in recent decades has in turn altered the nature of mergers. As a fraction of U.S. GDP, manufacturing has been in a long-term decline and has been overtaken by other sectors, most of which involve services and information technology. The locus of major mergers has followed suit, with important mergers now concentrated in sectors such as telecommunications, video entertainment, air transport, software, search, and health care. 13 Among many such examples are Comcast–NBCU; Ticketmaster–Live Nation; Oracle–PeopleSoft; Google–ITA; and all four recent airline mergers: Delta–Northwest, United–Continental, Southwest–AirTran, and USAirways–American. In these sectors, claimed efficiencies less often focus on the lowering of unit cost by increasing production volumes of similar products. Rather, efficiency benefits are now more frequently said to derive from other sources, three of which are discussed in this section. These are dynamic considerations, quality benefits, and vertical economies. In what follows, I will define and describe each of these types of efficiencies and examine the special problems of identification and measurement that they pose.
A. Dynamic Efficiencies
Dynamic considerations in merger analysis refer to technological change, that is, the rate at which the technology relevant to an industry’s product and process advances over time. This perspective stands in contrast with the static Williamson-type framework that focuses on the effects of a merger on the quantity and cost of the firms’ current output. In industries where innovation is the key competitive variable—industries such as pharmaceuticals, software, and search—dynamic considerations take on equal or greater importance compared to static cost and price issues.
A merger can affect the rate of technological change in two distinct ways. The first is the familiar scenario by which consolidating two firms may lower costs of operation. Here the relevant costs are those for achieving the same “output” objective, that is, the same degree of technological innovativeness. Efficiencies in the innovation process can arise from economies of scale, avoidance of duplication, and so forth, but now with respect to innovation rather than some final output. In principle, this effect would eventually manifest itself as the lowering of the total cost of achieving the same extent of technological output. 14
The second mechanism is more distinctive to the innovation process and perhaps ultimately the more important possible benefit from a merger. This involves greater productivity from the combination of the two firms’ separate innovation efforts, that is, more innovation output per dollar of input. Such greater productivity could result from combining innovation assets in ways that exploit complementaries or economies of scope, for example, one firm’s expertise in basic research now joined with another’s strength in trial and testing. This effect would be reflected in more innovation from the same total innovative input and effort (e.g., R&D budget).
Before discussing these issues further, it should be noted that there is another argument concerning dynamic considerations that is sometimes advanced in merger cases but in reality must be distinguished from the present issues. Parties to mergers may claim that the overall pace of technological change in their industry is so rapid as to remedy any transient competitive problems from the merger, or at least so rapid as to make antitrust policy largely irrelevant by the time it takes effect. Such an argument was made, for example, in the context of Comcast’s merger with NBC Universal, and is again being made in the proposed merger of AT&T and DirecTV. 15 Note that this argument does not involve technological change or other efficiencies by the merging parties or from the merger itself; rather, simply, that the dynamic market environment will trump any antitrust action or remedy. Since this argument does not involve a merger-related efficiency, it will not be discussed further here.
Returning to the main issues, in practice, the innovation cost savings and the innovation output increases from a merger may become intermingled. The Merger Guidelines comment on both mechanisms, albeit in quite different ways. The Guidelines first state that “[w]hen evaluating the effects of a merger on innovation, the Agencies consider the ability of the merged firm to conduct research or development more effectively. Such efficiencies may spur innovation but not affect short-term pricing.” The Guidelines also acknowledge the potential for cost efficiencies in the innovation process, but with the following cautionary message: “Research and development cost savings may be substantial and yet not be cognizable efficiencies because they are difficult to verify or result from anticompetitive reductions in innovative activities.” Thus, merging firms are free to make such claims, but they will be viewed skeptically.
The exact balance between input and output measures of the effects on innovation is not clarified in the Guidelines. Indeed, the Guidelines say that “[a]n important question for any merger is whether it is likely to diminish innovation competition by encouraging the merged firm to curtail its innovative efforts below the level that would prevail in the absence of the merger.” This passage suggests innovation, but then speaks in terms of “effort” and finally, as a reminder of the purpose of the exercise, “competition.” The most that can be said, seemingly, is that all of these considerations matter.
In their actual analysis of the effects of a merger on innovation, the agencies often define an “innovation market” within which market shares and concentration of innovative effort can be calculated and competition and innovation output assessed. Of course, these concepts are easier to state than to work with: the definition of an innovation market, the identity of all participants, and the meaning of market share in this context all defy easy explanation and measurement. The well-known difficulties include the lack of reliable method for measuring the ultimate objective of technological progress, which is the rate at which the stock of knowledge increases. The closest observable approximation may be something like a count of patents, but their varying significance has led many researchers to opt for R&D expenditures instead. The latter, however, is a yet more imperfect proxy, albeit more measurable. 16
Beyond that, the very relationship between competition and R&D or patents is a matter of some debate. Many researchers have found that R&D intensity rises with market concentration up to some point and then declines. This has led to the inference that some intermediate levels of concentration might be justified for its favorable effect on innovation, in contrast, of course, to static market analysis, which unambiguously favors less concentration. But others have questioned whether this relationship is in fact causal. 17
In any event, this general empirical guidance does not answer the specific policy question raised by a prospective merger of two technology-oriented firms. That requires an analysis not just of economies of scale in R&D but also the specific research capabilities of the merging firms, the nature of their intellectual property, the impediments that might be surmounted by merger, the prospects for integrating operations, and the competitors who may be working on the same matters. Past experience suggests that combining staffs, eliminating duplicative functions and personnel, and introducing new management can affect research productivity in many ways, not all of them favorable.
Interestingly, there is a certain amount of evidence on the effects of mergers on innovative activity. Studies of the actual effects of mergers on R&D expenditures suggests that curtailment of effort is indeed the norm. 18 But each case must be evaluated on its own merits, and neither the antitrust agencies nor the firms themselves can be very certain of the consequences of consolidating R&D functions. As a result, policy treads rather cautiously when dynamic considerations lie at the heart of a merger. 19
B. Quality Improvements
In addition to claims of dynamic efficiencies, parties to many recent mergers have contended that the primary benefit of their consolidation involves improvements in the quality of output. These claims of quality improvements from mergers take two quite different forms. The first is illustrated by the hospital sector, where a number of recent mergers have been said to create opportunities to transfer best practices between facilities. In this case, the quality improvement benefits accrue to the patients (consumers) of the acquired hospital, elevating service quality there to the standard of the acquiring hospital. 20
The quality-transfer scenario should be distinguished from the second type of quality improvement scenario—the claim that quality rises directly as a result of integrating the operations of two firms. Such benefits have often been claimed for airline mergers. Delta and Northwest Airlines, for example, argued that their merger would increase the frequency of service between cities and would also replace so-called interline service (requiring a change of carriers) with single-carrier service on some routes. 21 In contrast to the quality transfer scenario, integration-related benefits more likely flow to customers of both premerger firms, all of whom now presumably have access to the higher-quality service (e.g., single-line airline service).
These quality improvement arguments raise two questions: causal relationship to the merger and measurement. As with other efficiencies, quality benefits are not attributed to a merger if they can be achieved in whole or substantial part through some alternative that raises no competitive concerns. In the case of quality transfer, one obvious alternative may be contracting, that is, some arrangement between the independent parties that provides the lower-quality provider, for compensation, insight into the techniques employed by the higher-quality provider. While not uncommon in many sectors, it is also clear that such arrangements may have limitations in terms of their adequacy as a device for transferring operational know-how.
For realization of integration-related efficiencies, outright mergers might seem more important than for quality-transfer benefits. Nonetheless, there may be alternatives that preserve independence of the providers. In the case of airlines, for example, at least some integration-related benefits—for example, (nearly) seamless transfer of passengers—arguably can be achieved by code sharing. Code sharing is a form of contracting that has generally been shown not to raise the same degree of competitive concerns. 22
Economic studies prove the obvious point that quality improvements are beneficial to and valued by consumers. 23 Just as with cost efficiencies, such improvements constitute socially beneficial outcomes of mergers and should be viewed as another means by which the market process can deliver benefits. That said, the conventional Williamson-type apparatus for identifying and measuring such benefits is not designed to analyze quality improvements. Rather than a lowering of the cost curve from C1 to C2 as shown in Figure 1, a quality improvement effectively increases the value of the product or service to consumers and thereby shifts the demand curve upwards. Figure 2 describes this as the displacement of demand from D1 to D2, with the gap between the demand curves an indication of the product’s now-higher valuation by consumers. Despite this apparent close analogy between cost and demand shifts, measuring the rise in consumer valuation from a merger poses considerably more challenging problems.
Benefit of quality improvement.
The key question is how to quantify the consumer benefits from this upward (rightward) shift of demand. An outline of how this might be done has been provided in the context of a recent airline merger. As described by Isreal et al., Delta and Northwest Airlines projected $1 billion in cost synergies and consumer benefits from consolidation. Most of this total would take the form of improved service quality from “(1) reduced numbers of connections on itineraries, (2) more connecting itineraries on a single carrier, i.e. without interlining, (3) reduced average travel times due to better connections or less circuitous routes, (4) richer and better coordinated schedules of flights on given routes, and (5) more efficient fleet utilization that improves connections, increases the numbers of passengers flying on larger aircraft, and/or allows more passengers to be accommodated on their preferred flights.” 24
The quantification of such consumer benefits requires, first, determination of the magnitude of the demand shift, and then, second, valuation of the surplus associated with that shift. The first step is in principle a straightforward empirical exercise of estimating the demand responsiveness to quality change. In the case of the airline industry, the carriers themselves routinely perform this exercise using proprietary Quality of Service (QSI) models. These models estimate the relationship between service characteristics and traffic on a route in order to project an airline's need for ground personnel, aircraft, and other inputs as traffic changes. Importantly, the QSI models take price and total traffic on a route as given, varying only service attributes as the factor shifting market shares among carriers.
The effect of one carrier’s service improvements (due to, say, a merger) is described in Figure 3. With original demand D 1, the carrier serves Q 1 passengers at price P 1. Holding nominal price constant at P 1, the QSI model predicts that certain specific service improvements (e.g., frequency) would increase quantity to Q 2. The total consumer benefit from this quality improvement is equal to area A, since this represents the additional value placed on the higher-quality service by consumers collectively. Area A is not readily measured, but it can be shown under certain conditions that area A is equal in magnitude to another value that can be calculated more directly. The logic is as follows: in the example, quality rises enough to increase quantity from Q 1 to Q 2. That same quantity increase could have been caused by a price decrease from initial P 1 down to P 2 on Figure 3. Geometry tells us that the area of consumer benefit from that price reduction—area B—is identical to area A, but the magnitude B is much more easily determined: it requires knowing only the demand elasticity in that range. Then the benefit to consumers is simply the difference between the two prices multiplied by the relevant quantity.
Valuing quality benefits.
This demonstration appears to be a neat solution to the problem of measuring consumer surplus from quality improvements, and it has been used for exactly that purpose. Yet this technique is subject to some important caveats, two of which are especially notable. First, if the demand shift from the quality improvement is not parallel, and even if it appears parallel but the curves are nonlinear, the equivalence between the quality increase and the price decrease no longer holds. This is easily seen in Figure 4, which shows the case where the quality increase is more valued by marginal customers than by inframarginal customers, but the range of possibilities is limitless and the information requirements about consumers’ valuations in order to perform the benefits calculation are insurmountable. This approach is no longer an easy alternative.
Alternative measure of quality benefits.
Second, the equivalence between a price decrease from P 1 to P 2 and a quality improvement holds only if the latter is fully passed through to consumers. The reason is that a price decrease creates a benefit of that dollar magnitude for consumers, but there is no comparable assurance that the demand shift due to quality improvements is similarly passed through to consumers. To the extent that the merged firm captures part of any quality improvement in the form of a higher price to consumers, actual consumer benefit is correspondingly less.
As this example illustrates, there have been some useful developments in measurement methodology for quality claims. And since improvements in products or services represent increasingly common claims by merging parties, those developments are most welcome. That said, the issues are more challenging and the new techniques less satisfactory relative to the conventional case where unit costs of manufacturing fall as a result of a merger. For all these reasons, the antitrust agencies face considerable challenges in their efforts to measure and assess the frequent claims of efficiency benefits of quality improvements from mergers.
C. Vertical Economies
Antitrust policy toward vertical mergers has changed dramatically over time, to the point that few if any cases are brought against mergers strictly for their vertical properties. The reasons are firmly rooted in industrial organization economics, which now teaches that vertical integration is usually motivated by various types of cost savings. 25 These cost savings may take several forms: The information and transactions costs of using the market can be avoided. Supply can be ensured and stages of production can be better coordinated by vertical integration. And integration can eliminate the problem of “double marginalization” wherein an independent downstream stage can mark up the upstream markup and create a distortion to the firm as well as the market. All of these and more can result in cost savings to the merging firm and very possibly to consumers as well.
These types of vertical economies are routinely claimed by firms engaging in vertical or mixed horizontal-and-vertical mergers. Such mergers appear increasingly common, especially in high-tech sectors where complex multistage companies merge. Recent examples include the mergers between Time Warner and Turner Broadcasting, between Comcast and NBC Universal, and between Ticketmaster and LiveNation. The first two of these combined a large cable distribution system with a program provider, and the parties to both mergers claimed substantial efficiencies from better coordination, lower transactions cost, and avoidance of double marginalization. The Ticketmaster–LiveNation merger involved both horizontal and vertical issues, since Live Nation had recently entered the ticketing business long dominated by Ticketmaster. But the parties stressed the efficiencies from incorporating Live Nation’s large presence in the promotion, venue management, and artist representation stages, into Ticketmaster. 26
The difficulties associated with measuring vertical economies are immediately evident from a recognition of their origins. As previously noted, such economies derive from savings in information and transactions costs, from greater assurance of supply, from better coordination between stages of production, and from avoidance of double marginalization. With the possible exception of double marginalization, all of these pose substantial problems of identification and measurement. It is easy, for example, to envision information and transactions cost savings from vertical integration, and such savings are often claimed. But it is quite difficult to pinpoint those costs in actual business transactions and even more difficult to quantify them with any precision or confidence. It is similarly difficult to trace and measure the costs of imperfect coordination between firms at different stages of production or of possible supply disruptions between stages when they are performed independently. Yet these are precisely the cost savings from vertical integration at issue in many mergers, and neither the merging parties nor the reviewing agencies would seem fully equipped to assess them.
Economics offers modest insights into the broader question of the circumstances under which vertical integration is likely to result in cost savings. Case study evidence beginning with Adam Smith’s pin factory to studies of steel production cast light on aspects of the production process that are more likely associated with benefits—for example, process technology, complex products, among others. And a few broad studies offer some confirmation. The classic analysis of vertical integration in autos by Monteverdi and Teece, for example, found that auto companies were more likely to choose in-house production of complex parts and components than those more commodity-like in nature. 27 Such studies provide support for the general theory but little operational guidance in specific cases.
A different approach has been taken by studies that examine the relative cost outcomes from varying degrees of vertical integration in the electric power industry. 28 These studies compare the cost of electricity sold to final consumers by utilities that perform all three industry functions—generation, transmission, and distribution—relative to others that purchase most or all of the electric power that they distribute to final customers. The usual finding is that integrated utilities are more cost-efficient. While this methodology might seem to be a promising approach to determining vertical economies in other industries, the circumstances of the electricity sector are unusual; notably, it is subject to regulation, so that data on costs and other variables are publicly available, a wide range of integration is observed, and there are numerous firms.
As noted above, claims of avoidance of double marginalization are in principle more straightforward to demonstrate and to assess, since their essential elements are simply the markups at both stages of production. Yet firms are cautious about advancing such claims since they concede the existence of market power at both stages, an admission that can complicate other claims in their affirmative case to the agencies. Nonetheless, given the importance of efficiencies to otherwise problematic mergers, such claims have been advanced in a number of prominent cases.
A particularly insightful examination of vertical economies was conducted in the Comcast–NBCU merger. This consolidation was reviewed for its competitive implications both by the Antitrust Division of the DOJ and by the FCC. Their respective analyses offer an unusual glimpse into not only the evaluation of vertical efficiencies but also striking differences between the two agencies’ reviews. Baker reports that the FCC analyzed the claims of various cost savings and “credited these possibilities, but not to the extent claimed by Comcast. The FCC found them to be plausible in principle, but in some respects speculative, overstated, or unsubstantiated.” 29 By contrast, remarkably, the Antitrust Division was said to have examined the very same claims and, “finding each negligible, not specific to the transaction, or not verifiable,” came to the conclusion that all of them should be dismissed outright—a considerably more skeptical view of the same issue arrived at by an equally expert agency. 30
With particular respect to the argument about double marginalization, and on the very same evidence, the agencies again came to quite different conclusions. The Antitrust Division found that “programming was transferred at a price close to marginal cost … , so there was little double marginalization to eliminate.” 31 The FCC, on the other hand, “accepted that the price of programming exceeded marginal cost” and concluded that there were in fact benefits from eliminating double marginalization, although “substantially smaller than Comcast claimed.” 32
Thus, the review of the Comcast–NBCU merger underscores the difficulty of assessing those claims and the very real possibility that two expert bodies might come to significantly different conclusions about the very same issues on the very same evidence. Such a divergence of opinion rarely if ever becomes public, but this case makes clear the considerable degree of difficulty associated with rendering consistent and convincing judgments about the magnitude—and indeed, even the existence—of vertical economies.
IV. Efficiencies in Networks
Efficiencies from innovation, from quality improvements, and from vertical integration have become quite important in many recent mergers. As illustrated by several of the examples previously discussed, these issues tend to arise and converge in network industries such as communications, entertainment, airlines, and distribution services. The communications sector is characterized by substantial technological change, and parties to mergers in these sectors routinely claim that they will result in greater investment in new technologies. Mergers involving video production and delivery, or music production and performance, often cite closer coordination of stages of production as key benefits. Transportation sector mergers, such as rail and airlines, are justified by the creation of integrated end-to-end service, among other things. And in all of these, the direct consumer benefits of seamless, bundled, and one-stop shopping are often asserted.
Where realized, all of these constitute efficiency and related benefits, of course. And where merger-specific, verifiable, and at least partially passed through to consumers, they represent efficiency defenses to otherwise problematic mergers in network industries as well as elsewhere. But network industry mergers are of special interest from an efficiency perspective since they often raise two additional efficiency-related issues with special force. These issues are out-of-market efficiencies and integration inefficiencies, as will now be discussed.
A. Out of Market Efficiencies
The objective of antitrust policy in the U.S. and most countries is not generally construed as simply maximizing total surplus—the sum of consumer surplus and producer surplus. Rather, priority is given to cost efficiencies that result in at least some gains to consumers: a merger that results in consumer harm is not acceptable even if the gain in producer surplus is larger. A somewhat different issue arises if a merger in fact results in a gain in consumer surplus, but within that overall increase, some consumers lose while others gain to a greater degree. In such a case, is it only the overall consumer surplus change that matters—in which case the merger presumably can proceed—or should some additional consideration be given to identifiable segments of consumers—“antitrust markets”—that are adversely affected? And if the latter, might the merger be prohibited despite the fact that consumers as a whole benefit?
These issues may be clarified by an example from the airline industry. A major airline is a vast multiproduct firm, serving thousands of individual routes that constitute separate antitrust markets since they are not demand substitutes. But different airlines face each other in diverse ways on their myriad routes. On some routes, they may be direct horizontal competitors. Simultaneously elsewhere, one may be a potential entrant on a route served by the other. On yet additional routes, they may provide end-to-end service, connecting in a vertical relationship. And on others, of course, they may not overlap or intersect in any way whatsoever. It is therefore entirely possible that a given merger may reduce competition on some routes (the “overlap” routes) but improve service or reduce costs on others where the carriers provide end-to-end service, and have no effect whatsoever on yet additional routes.
Of course, the fact that mergers occur between companies that face each other in diverse ways and create diverse competitive outcomes is not unusual. One company may produce products A, B, and C; and the second C, D, and E. If combined ownership of product C were competitively problematic, the standard policy and practice would be to permit the merger conditional on divestiture of one of the two companies’ operation of product C to an independent entity. If done correctly, this divestiture should preserve the competition otherwise lost from merger. This scenario is entirely familiar and does not, by itself, create a policy problem. Rather, the policy problem results from a different factor, specifically, the extreme economies of scope in the production of outputs for these demand-defined markets.
On each airline route, for example, a carrier operates a set of assets—aircraft, personnel, baggage services, and marketing—that also represent inputs into production on other routes. As a result, few routes can be served at any reasonable cost on a standalone basis. Hence production can be thought of as occurring at the network level. On the other hand, consumers of airline service on a particular route do not care how its operation fits into the rest of the carrier’s network structure. Demand, in short, arises at the route level.
It is this fundamental mismatch of product demand and product supply—route versus network—that lies at the heart of the policy problem. To simplify a bit, competition problems arise at the route level, but due to economies of scope, they can only be solved at the network level. At the route level, it is impossible to divest one carrier’s service on overlap routes to some third party since the assets required to produce service on that route are so fully shared with other services (routes). Attempting to preserve or create independent service on the route in question would be prohibitively expensive, and no firm would consider entering into airline service in such a selective standalone manner.
As a result, a competitively problematic merger of the sort just described would seem to pose an all-or-nothing choice to the competition agency: either prohibit the merger in its entirety since there is no practical remedy, or permit it in its entirety, conceding competitive harms to certain customer classes. These issues are not altogether novel, as this airline example makes clear. 33 Yet they seem to have taken on special urgency and frequency in recent years, as more mergers occur in network industries. 34 Efficiencies in many transportation, telecom, and utility mergers are not necessarily realized in the same demand-defined market as where harms arise, and practical solutions to the problem are not obvious.
The growing importance of network mergers and out-of-market efficiencies focuses greater attention to the question of remedies and, specifically, to the question of whether there are any possible alternatives to this all-or-nothing dilemma worth considering. In that spirit, we note the following possible policies to address localized competitive concerns within a broad merger:
Joint ventures of the relevant assets. Joint ventures can in principle allow for the realization of economies involving certain assets while preserving independence of all other assets and operations of the parties. The degree of independence depends on the governance arrangements between the parties, but certainly can be less anticompetitive than full merger.
36
In the case of airlines, code-sharing arrangements on particular routes. Code-shares allow for one carrier to offer apparent service on a route by selling tickets to service on the second carrier’s scheduled operations. Code-shares may succeed in confining carriers’ coordination to a limited number of specified operations.
37
Regulated colocation, whereby specific assets are made available to both parties, who otherwise remain separate. Colocation was employed, for example, to allow the postdivestiture long-distance AT&T to share access to certain key assets with the Bell Operating Companies that were being split off.
38
Trackage rights in railroads. These take various forms, but generally speaking trackage rights allow a “tenant” railroad to run its equipment over the rails and other infrastructure of a second carrier on routes otherwise served only by the second carrier. Their effectiveness depends importantly on the contractual arrangement between the parties. Trackage rights have been used in a number of major rail mergers.
39
So-called “competitive rules joint ventures,” an arrangement that creates a joint operating and governance structure over one parties’ operating assets.
40
Properly devised, these allow for two parties to share crucial assets and with appropriate contractual arrangements can in principle provide adequate compensation for current production and for investment in future capacity, while preserving the independence of the parties. Such an arrangement was employed to resolve concerns with the consolidation of Alcan with Arco.
41
These examples illustrate the various creative options that can be devised to address competition concerns in a network industry where a merger is proposed. While promising, two features of these deserve mention. First, several require appropriate contractual arrangements between the parties, a task that may itself raise substantial difficulties. Second, for alternatives such as these to be adopted, they need not duplicate the full efficiency benefits from merger. Rather, all that is required for one of them to represent superior policy is that the gain in efficiency benefits it allows exceeds any loss of competition by a greater amount than the net effect of full merger.
Any of these techniques may bring competitive relief to affected markets and avoid the unappealing alternatives of simply permitting mergers with localized competitive harms or prohibiting them in their entirety. 46 In this manner, out-of-market efficiencies may be preserved while addressing the specific harms that would otherwise result.
B. Inefficiencies of Integration
Ironically, some of the same characteristics of networks that make them difficult to disentangle for purposes of antitrust remedies also render them difficult to combine when network mergers are allowed to proceed. That is, the extreme degree of integration of the production process by which multiple markets (defined by demand distinctiveness) are efficiently served by a single producer complicates the process of integrating new services and operations. The reasons are analogous: each single market has substantial spillover effects on other markets, so that altering—integrating—operations anywhere creates a host of effects in related markets. In airlines again, merging operations on a route alters upstream feed traffic from many routes as well as downstream traffic to numerous destinations. Aircraft, crew, and ground infrastructure on all these related routes must all be changed accordingly.
Beyond operational implications, airline mergers have implications for several other facets of the business. These include frequent flier programs, reservation systems, and labor forces and work rules. From much experience, it may be said that these defy easy change. Changes to frequent flier programs tend to frustrate passengers, especially those business travelers who generate the most mileage credits. Based as they are on different hardware and software, reservation systems have often proved difficult to integrate. Some carriers have postponed full integration for two or more years in order to minimize the difficulties. 47 In addition, different work, pay, and seniority rules for each carrier have often set the stage for labor strife upon integration. The USAir–Piedmont merger was two years in negotiation due to difficult labor issues. 48 More recently and perhaps most famously, US Airways and America West spent ten years seeking, unsuccessfully and sometimes bitterly, to integrate their union seniority lists, a task superseded by US Airways’ merger with American Airlines. 49
The known and direct costs of integration can be considerable. Moss has recently compiled important evidence on the projected and actual integration costs for four recent airline mergers. 50 The projected costs are not inconsiderable, ranging from $500 million to $1.2 billion, and taking up to three or four years. Moreover, there is some evidence that actual integration costs may be substantially greater than projected levels. The one completed airline merger at the time of Moss’s research—Delta and Northwest—reported actual costs were $1.5 billion, fully three times their estimate of $500 million for integration.
In addition, these direct costs of integration are only part of the social cost of merger. Difficulties in integration result in congestion, frustration, time losses, and other costs imposed on consumers. These, too, are predictable consequences of efforts to combine networks. While the Merger Guidelines explicitly provide that cognizable efficiencies are “net of the costs produced by the merger or incurred in achieving those efficiencies,” there is little on the public record to indicate whether or how the antitrust agencies trade these off in evaluating the overall efficiency consequences of mergers.
V. Conclusions
In the merger review process, as this article has demonstrated, there has been considerable change in the issue of efficiencies. One change has been a dramatic improvement in the actual techniques of evaluation employed by the agencies. This in turn has prompted better analysis in the submissions of the merging parties and arguably more accurate judgments, measurements, and determinations of efficiencies in the merger review process. 51
The other important change has been the shifting nature of claimed efficiencies, with increased emphasis on dynamic efficiencies, quality benefits, and vertical economies as justifications for mergers. These newer types of efficiencies confront the agencies with novel analytical challenges, reminiscent of the difficulties faced by the agencies many years ago as they sought to evaluate traditional efficiency claims. One might hope that over time the same expertise may be developed for analyzing dynamic, quality, and vertical efficiencies as now exists for claims of more traditional economies.
Footnotes
Author’s Note
This article is an outgrowth of discussions and presentations at the AAI Annual Conference, June 2014, and a prior workshop at Northeastern University, October 2013. It has benefited from comments and suggestions from many individuals, especially Bert Foer, Rick Brunell, Jim Donahue, Greg Gundlach, Kevin Hearle, Diana Moss, Phil Nelson, and two anonymous referees. Opinions and remaining errors are the sole responsibility of the author.
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author received no financial support for the research, authorship, and/or publication of this article.