The attempted merger between General Electric and Honeywell A case study of transatlantic conflict March 2005 Jeremy Grant (Graduate Institute of International Studies, Geneva) and Professor Damien J. Neven (Graduate Institute of International Studies, Geneva) Financial support from the TMR program on « Competition Policy in international markets » is gratefully acknowledged. The authors would also like to thank all those participants in GE/Honeywell who generously gave of their time to discuss the case. Acknowledgements to be approved). Thanks are also due to Sarah Nash at JP Morgan in New York. Helpful comments were also received from Dr. Thomas Kirchmaier at the LSE and Selman Ansari at Bates, Wells in London. Abstract The thwarted merger of General Electric and Honeywell stands out as, so far, the only merger between US companies to be derailed solely by the European anti-trust authorities, while being cleared by the US Department of Justice (DoJ) and 11 other jurisdictions.
In this paper, the authors examine the European Commission’s decision, and the theories underlying it and compare the Commission’s approach with that followed by the DoJ. They observe that the Commission and the DoJ had a different assessment of broadly similar facts, and attempt to understand the source of the divergence. The authors find that (i) the horizontal effects identified by the European Commission rely on a particular perspective of market definition which is debatable (and leaves some questions unanswered). ii) The anti-competitive effects in the bundling and archimedean leveraging theories are not sufficiently robust so that they could be presumed. Accordingly, their likelihood should be supported by strong evidence but the evidence presented by the Commission was far from compelling. (iii) The deal may have involved significant efficiencies that were overlooked. These observations raise the suspicion that the Commission’s decision may have been affected by bureaucratic capture, such that civil servants did not follow the mandate that had been assigned to them.
We find that the procedure enforced at the time was vulnerable to capture and that the Commission had an incorrect perception of the standard of review that the Court would apply to its decision in the context of an appeal. The accountability to which the Commission felt subject to was thus biased downwards and enlarged the scope for capture. In addition some (admittedly casual) evidence regarding the actual unfolding of the procedure, as well as subsequent reforms of process and procedure undertaken by the Commission, would support the view that significant problems arose in this area. “In the macho world of merger regulation ….. authorities strive to win a tough reputation” 1. Introduction: Project Storm As General Electric’s CEO Jack Welch strode onto the floor of the New York Stock Exchange on October 19th 2000, little did he know that he was about to become embroiled in a series of events that would lead to him bidding for Honeywell, a rival industrial conglomerate, and place him in the eye of a transatlantic storm over competition policy. 2 The deal, codenamed Project Storm, would have been the largest industrial merger in history.
Instead, it became infamous as the first, and so far only, merger between US companies to be derailed solely by the European anti-trust authorities, while being cleared by the US D epartment of Justice (DoJ) and 11 other jurisdictions. This paper will critically examine the European Commission’s decision, and the theories underlying it and compare the Commission’s approach with that followed by the DoJ. We observe that the Commission and the DoJ had a different assessment of broadly similar facts, and attempt to understand the source of the divergence.
In particular, we examine how process and procedure differed between the United States and the European Union, to determine what extent these factors influenced the outcomes of the regulatory processes on either side of the Atlantic. While a significant amount has already been written on the case, systematic comparisons between the approaches followed by Commission and the DoJ have so far remained difficult because much of the DoJ’ s analysis is not in the public domain. In addition, little attention has been given so far to process and procedures.
We attempt to gather systematic evidence on these issues; beyond an extensive survey of public sources currently available, we have extended our research through access to non-public sources including submissions to the regulators and other non- public research done at the time of the transaction. We have also conducted interviews with the regulators on both sides of the Atlantic and lawyers representing parties on both sides of the argument3. Part I briefly reviews the Commission’s decision, discusses possible sources of divergence and dismisses some of them as highly unlikely.
Part II provides a critical review of the Commission’s decision and highlights differences with the DoJ’s analysis. Buck (2004). Welch and Byrne (2001). 3 One of the authors also attended the GE and Honeywell appeal hearings at the European Court of First Instance in Luxembourg on May 25th and 27th 2004. e 2 1 3 Part III discusses the role that process and procedure have played in the emergence of divergent outcomes. Part 1: The European Commissions Case: The Commission’s case was threefold.
GE held a dominant position in the market for Large Jet Aircraft engines (between 43% and 65% depending on how market share was calculated) a situation where firms’ conduct is subject to particular scrutiny under Article 82 of the Treaty of Rome. 4 Meanwhile, Honeywell had a leading position in the avionics and non-avionics aerospace component markets. The European Union Merger Regulation enforced at the time prohibited mergers or acquisitions which “create or strengthen(s) a dominant position as a result of which effective competition would be significantly impeded in the common market”. According to the Commission, effective competition would be impeded mostly because of “conglomerate” effects. These may arise when the merging parties are active in different products markets 6 and when the merger raises the scope for exclusionary practices. The three main arguments in the case were as follows: 1) Bundling -The strength of the combined positions of the merging firms, the Commission feared, would allow GE to engage in exclusionary product bundling with the ultimate effect of foreclosing markets for single product line competitors (particularly Rolls
Royce in aircraft engines and Rockwell Collins in aerospace components). Such foreclosure involves the deterrence of new entrants, the reduction of investment by existing competitors and possibly their exit. Art 82 ECT prohibits the abuse of a dominant position and not the dominant position by itself. The Merger Regulation has since been amended. The criterion for assessing mergers has been modified. Mergers which “would significantly impede effective competition, in the common market or in a substantial part of it, in particular as a result of the creation or strengthening of a dominant position” would now be held unlawful.
The new formulation is meant to include horizontal mergers which have substantial unilateral effects despite the fact that merging partners do not reach the threshold of dominance. Given that the anti-competitive effects identified by the Commission in the GE/HW merger were primarily associated with conglomerate effects, and given that if anything, the criterion for horizontal effects have been strengthened, one can presume that, other things being equal, anti-competitive effects would also have been identified under the new merger regulation.
The new merger regulation, however, emphasizes the role of efficiencies and the question arises whether efficiencies could have trumped anti-competitive effects under the new regime. This is further discussed below. 6 The Commission first utilized the theory of “portfolio effects” in Guinness/Grand Metropolitan (1997). The merged entity was required to divest its rum distribution in Greece, although there was no horizontal overlap as only Grand Metropolitan was a rum distributor in the country.
The reasoning was that the firm had too wide a portfolio of market leading brands, thus giving it extensive opportunities to increase market share via bundling and tying, and “will be able to realize economies of scope and scale in sales and marketing activities”. In Coco Cola/Carlsberg (1998), the Commission required the divestiture of the 3 rd largest cola brand in Denmark and a bottling plant, because of fears of greater market power stemming from the combined companies drinks portfolio and resultant economies of scope and scale, although in this case there was clearly horizontal overlap.
In both cases, the EC viewed enhancement of product portfolios post-merger as extending scope for exclusionary practices, which by their very nature would harm competitors. 5 4 4 To back up its case, the EC put forward an economic model of bundling supplied by Rolls Royce and economic expert Professor Jay Pil Choi (Choi, 2001). 7 The Commission claimed that the merged entity would have incentives to bundle its avionics (e. g. aircraft communication and navigation equipment) and non-avionics products (e. g. heels, lights and landing gear) with engines, at a discount to single product purchases, which competitors could not match. In the short to medium term, this would marginalize competitors by depriving them of revenues which would mean they could not cover their fixed costs. In turn this would effect their spending destined for research and development on the next generation of products meaning they could not compete effectively with respect to future platforms. The Commission presented evidence that Honeywell was already deploying bundling in what it designated as “multi-product” bids.
The EC also received evidence from aerospace component competitors (in particular Rockwell Collins) that they had faced Honeywell component bundling, and were unable to compete. 8 Therefore, if the deal were to proceed, they would have to withdraw from either the avionics or non-avionics businesses to concentrate their resources. 9 2) Archimedean Leveraging – GE would further increase its dominance by using the position of GE Capital Aviation Services (GECAS) in the market for leasing service.
This was something GE had been doing since 1996 with its “GE only” policy of purchasing GE engines, where available, for the planes it leases to clients, and had two effects. Firstly, the Commission believed that substantial engine market share shifting to GE had been driven by this policy on multi-source platforms (i. e aircrafts for which several competing engines are certified, the choice of engine being made by the buyer). Secondly, the Commission believed that this policy could convince manufacturers to certify a GE engine on sole source platforms (where a single engine is certified).
The Commission observed that in the market for Large Regional Jet, the 3 manufacturers choose GE as their single certified engine supplier, and GECAS subsequently ordered a large number of regional aircraft from each. The EC feared that “GE only” would be extended to Honeywell components, hence further transforming a leading market position into a dominant one, again leading to or re-enforcing market foreclosure. It went on to argue that this process would be aided by GE’s financial strength demonstrated by its low market based cost of capital and triple AAA credit rating.
These were all elements of “GE’s toolkit for dominance” (Drauz, 2002). This theory was controversial, not least because GE was being accused of market tipping with a market share of less than 10% in aircraft leasing. However, to bolster its case the Commission relied on an economic model of “Archimedean Leveraging” supplied to it Professor of Economics, Michigan State University. Source: Testimony given at Honeywell appeal hearing, Luxembourg May 25, 2004 – Rockwell’s clients name was kept confidential. 9 Interview with John DeQ.
Briggs, Rockwell’s American lawyer – Washington DC, February 2004. 8 7 5 by United Technologies (UTC), a multi-product competitor of GE. 10 This model purported to demonstrate how a firm with a limited presence (as a buyer) in a downstream, but not adjacent market, could affect competitive outcomes in upstream markets due to the embedded nature of the products concerned. GECAS role as launch customer for new aircraft models would further strengthen this effect as GECAS would “seed” airlines, particularly smaller and regional ones, with GE and Honeywell products.
This might distort their later purchasing decisions in favor of General Electric products from air-framers and other leasing companies, as it lowers airlines costs to maintain “fleet commonality”. Overall, according to opponents of the deal, “the merger would have sandwiched airframers between their largest purchaser of aircraft, their largest supplier of avionics and other equipment, and a major source of finance for the building of new aircraft. At the same time, the combined firm would gain an advantage in the sale to airlines by bundling a broad range of products and seeding them with aircraft” (DeQ.
Briggs and Rosenblatt, 2002)11. 3) Vertical and Horizontal Effects – The first two arguments related to “portfolio” and/or “conglomerate” effects (also known as range effects), areas of anti-trust regulation that remain contentious amo ngst economists and across jurisdictions. For its third line of argument, the EC returned to more traditional anti-trust areas: horizontal overlap in large regional and medium corporate jet engine markets and power systems. It also highlighted vertical issues related to Honeywell as the sole manufacturer and supplier of component engine starters to Rolls Royce. 2 The Commission also claimed that the parties had not presented any efficiencies as a justification for the transaction. In principle, the analysis of efficiencies would allow the Commission to distinguish a merger that strengthens a company’s market position through efficiencies that ultimately benefit consumer welfare, from one whose main purpose is to enable the combined entity to implement anti-competitive conduct including possibly predatory behavior, with the effect of marginalizing and ultimately driving its competitors from the market.
The Commission’s reasoning behind the decision is extensively detailed in its final report which is 150 pages plus in length. By contrast, the US authorities cleared the merger, demanding only limited remedies detailed in a 2 page press release on May 2nd 2001. 13 Interestingly, Rolls Royce and UTC seemed to effectively coordinate their efforts. RR concentrated on bundling, while UTC based it arguments on the vertical integration of GECAS and share shifting. Clearly it was in their interests to not push the bundling argument in isolation as they had been the original purchasers of Honeywell, and were out-bid by GE. 1 For an overview, see diagram in Appendix 1 presented by Rockwell Collins at appeal hearings of the European Court of First Instance in Luxembourg, May 2004. 12 The parties did agree to a number of divestiture remedies in relation to some of these issues. 13 The merged entity would be required to divest Honeywell’s military helicopter engine business (approximately $200mm in revenues) and open up the repair, maintenance and overhaul services for certain Honeywell engines and auxiliary power units, to further competition. (See http://www. usdoj. gov/atr/public/press _releases/2001/8140. htm).
Subsequently, the DoJ did detail other 10 6 Therefore, we are left with the question of how two regulatory authorities could come to such different outcomes. This is an important question as “such a wide divergence … creates a lot of uncertainty – chaos – for business. This uncertainty will lead to behavior that we can well imagine would discourage firms from doing the sort of acquisitions that we would like them to be doing” (See Kogut, 2001 14). In principle, one can identify the following possible sources of divergence. First, a divergence may occur because the authorities are subject to different legal standards.
In particular, these standards may differ because they express different objectives or because the burden and the quantum of proofs do not coincide. For instance, one legal standard may give more importance to the prospect of prohibiting a pro-competitive merger than another, and may accordingly impose more stringent conditions on the evidence that is necessary to prohibit a merger. These objectives and rules governing evidence can be explicitly stated by the statutes or inferred from the relevant case law. Second, the authorities may pursue different objectives from those that they have been assigned.
The view of antitrust authorities as omniscient and benevolent agents seeking the public good has long been dismissed. These authorities are properly seen as a collection of agents pursuing their own objectives subject to an imperfect monitoring by the principal having assigned particular tasks to them. Being exposed to imperfect accountability, agents (civil servants) can take decisions which deviate from what the objectives that have been assigned to them would dictate, but achieve their own goals. This phenomenon is usually referred to as bureaucratic capture.
For instance, a high profile prohibition may advance the career objectives of civil servants. In such an environment, interested parties may also try to influence the agents by offering prospects for further fulfillment of their objectives, which are made contingent on a particular course of action. Of course, the extent to which capture can arise is highly dependent upon the type of accountability that agents are subject to. In this respect, the procedures which govern the investigation as well as the prospects for appeal and the standard of review that will applied by the Courts will play an important role.
Third, different authorities may have a different judgment on a similar set of facts. Merger control involves a prospective exercise which is surrounded by much uncertainty and different agents may have genuinely different views. Fourth, the concentration under review may have different consequences across jurisdictions. This may typically arise if the different jurisdictions belong to different geographically relevant markets, in which the conditions of competition may differ 15. aspects of its reasoning in a report submitted to the OECD on conglomerate mergers – (“Range Effects: The United States Perspective” – http://www. sdoj. gov/atr/public/international/9550. pdf). 14 Antitrust Fall, 2001, Roundtable Discussion, pg 9 15 It may also arise if the jurisdictions belong to the same geographical market, but efficiencies differ across jurisdictions and the authorities pursue a total welfare standard (which takes into account the profit of the firms). 7 At the outset, it would appear this last source of divergence is unlikely to have played a role in the case at hand, as both jurisdictions considered the relevant market as global.
Some observers have also suggested that capture by wider political interests is at the source of divergence. In particular, it has been suggested that the US DoJ did not undertake a thorough investigation of the merger under pressure from GE and the incoming Bush Administration (see Kolasky, 2002 for a response). Meanwhile, the European Commission was accused of acting in the narrow protectionist interests of European companies and their member states (see DeQ. Briggs and Rosenblatt, 2001 for a response). 16 It is not clear however that one should give much credence to these suggestions.
First, regarding the claim that the DoJ did not undertake a proper investigation, it is worth noting that while the major analysis in the US was done in a period after a number of Clinton administration officials had departed the DoJ, 17 and the Bush appointees had not arrived, a rigorous analysis was undertaken by the departments’ career staff in the Office of Operations. The investigative team had in-depth knowledge of the aerospace industry, having previously examined the merger of Honeywell and AlliedSignal.
The DoJ retained an external economist to assess the conglomerate effects doctrine with a track record suggesting that he was unlikely to dismiss such effects off hand. 18 Second, regarding the claim that in challenging the merger, the European Commission was acting in the narrow protectionist interests of European companies and their home states, it is worth noticing that the major complainants in Europe included US firms such as United Technologies and Rockwell Collins. These, as we have already seen, provided significant elements of the Commissions case.
Airbus, by contrast, appears not to have opposed the deal. In addition, it does not seem that the Commission has a track record suggesting that it favors EU firms in merger control,19 and we have not uncovered evidence suggesting that this case is different from others in this respect. 16 For example, then US Treasury Secretary Paul O’Neill stated that the decision was “off the wall” and the Europeans were “meddling in things that one would think were outside their scope of attention” (Carney, 2001).
Meanwhile, Senator Ernest Hollings, Chairman of the Senate Commerce Committee, sent a letter to the EC accusing them of “an apparent double standard by swiftly approving mergers involving European companies and holding up those of US groups. Its apparent EU disapproval gives credence to those who suspect that the EU is using its merger review process as a tool to protect and promote European industry at the expense of its US competitors. ” (DeQ.
Briggs and Rosenblatt, 2001) 17 For example, Joel Klein, Assistant Attorney General for Antitrust had left for a position at the Bertelsmann, while John Nannes the Acting Assistant Attorney General for Antitrust had to recuse himself. (DeQ. Briggs and Rosenblatt, 2001). 18 Interestingly, the external expert was Professor Marius Schwartz who had previously served as the DoJ’s Deputy Assistant Attorney General for Economics and Economics Director of Enforcement during the Clinton Administration.
Schwartz had substantial research experience in vertical restraints and exclusionary practices including having published research with Dr Robert Reynolds, co-author of the Archimedean Leveraging theory. Source: Schwartz CV online at: www. georgetown. edu/faculty/schwarm2 19 A number of econometric studies have analysed the outcome of merger control in the EU, in particular Lindsey and Williams (2003), Bergman et al. (2004) and Duso et al. (2003). These studies examine different samples and use different methods. None of them uncovers evidence of a bias against non EU firms. 8
Differences in legal standards between the US and the EU have been much discussed (see for instance, Fox, 2002), both in terms of the substantive standard and in terms of the standard of proof that agencies are held to. Differences in substantive standards are probably such that the EU can be expected to prohibit mergers when the US may not, in particular because the EU legal framework may give more weight to the fate of competitors. But these differences are usually considered to be marginal. Similarly, differences in judgment will only affect outcomes in relatively marginal cases20.
There is no clear indication either that agencies are held to different standards of proof, at least formally, as both are supposed in a way or another to show that anti-competitive effects would materialize “very probably” 21. Hence, in order to identify the importance that capture (the third source of difference) may have played, a two pronged test can be developed. First, is it that the decision taken by the DoJ was “marginal”? If not, it will provide some (negative) evidence that capture may have played a role (competing hypotheses being inconsistent with this observation).
Second, is it that the Commission’s analysis rests on firm grounds? In particular, is it that the weight the Commission has attached to particular pieces of evidence appears well founded (relative to the DoJ). If not, it will suggest that capture has indeed played a role. Whether the decision taken by the DoJ can be seen as marginal is difficult to assess in a systematic fashion; however, both the interviews with officials and lawyers involved as well as the DoJ’s comments on the case (in particular in the context of the OECD) suggest that the DoJ had a lot of confidence in its decision.
The second question, namely whether the Commission’s analysis rests on firm grounds can be assessed through a detailed analysis of the Commission’s published decision. This is the object of the main analytical section of this paper, which follows. Part 2: Re-Examining the Commission’s Decision: Overview of the Aerospace Market: Jet Aircraft Engines: As we have seen, General Electric is a manufacturer of jet aircraft engines. Its main competitors are Rolls Royce of the UK and United Technologies (UTC), a US industrial conglomerate, through its subsidiary Pratt & Whitney (P&W).
The market is concentrated between these three independent players. Manufacturers sometimes compete to be the sole supplier of engines on each platform, in particular within the Regional and Corporate Jet category (see below). However, for most large commercial Assuming that both jurisdictions were exposed to a similar set of facts. This appears to be the case given that the DoJ and the Commission had extensive contacts and debates from the very beginning of the procedure (Source : interview with DoJ and Dg comp officials). 21 See Kolasky (2002) and Versterdof (2004) on the standard of proof in merger cases.
Of course, the significance of any particular standard of proof is largely determined by the standard of review that will applied by Courts in case of an appeal. As discussed below, there are good reasons to believe that the Commission was operation under an illusion with respect to standard of review that would apply in the event of an appeal. 20 9 jets, the ultimate customer retains a choice of engines. 22 Therefore, the competitive process in this area is two-stage. Firstly, the manufacturers compete to have their rival engines certified on a new plane.
Secondly, they have to further compete to persuade the ultimate purchasers (airlines and leasers) to choose their engines over rivals, for the planes they are purchasing. 23 Platforms can be broken down into 3 main categorizes based on aircraft size, and are used in order to define segments in the engine market (as larger planes require more powerful engines). The DoJ adopted a different approach based on the thrust of the engine itself. This issue will be further discussed below. For the time being, we follow the approach of the Commission. Each category is served by different aircraft manufacturers. See Appendix 2 for a complete breakdown of each category by manufacturer, aircraft model, engine type and thrust). 1) Large Commercial Jets: In its decision, the Commission concentrated on the market for large commercial aircraft which is dominated by the duopoly of Airbus and Boeing. These can be sub-categorized into wide-bodied (twin aisle) and narrow-bodied (single aisle) aircraft, based on the number of passengers and mileage. The ultimate customers for these jets are airlines. Buyers from Airbus and Boeing are either the airlines or leasing companies. ) Regional Jet Aircraft: In the next segment, regional jet aircraft, there were four manufacturers at the time of the decision, Bombardier, Fairchild Dornier, Embraer and British Aerospace. 24 Regional jets can be sub-divided into large and small and are purchased by airlines, and in recent years, increasingly by leasing companies. The Commission found that there was a “clear growing focus of airlines on large regional jets …. demonstrated by the fact that such aircraft constituted 14% of the European fleet in 1992 and 33% in 1998” (pg 8). ) Corporate Jet Aircraft: In this category there are a greater number of manufacturers including Gulfstream (General Dynamics), Raytheon, Dassault and Lear (Bombardier) amongst others. The ultimate customers are corporations and high net worth individuals. GE only manufactures one engine in this market, the CF34. A few platforms, most prominently the Boeing 737, sole source their engines (see Nalebuff, 2003 for a discussion of the rational behind such exclusive agreements). 23 However, the 3 independent engine manufacturers have formed a number of joint ventures.
Most prominent of these is CFMI, a joint venture between GE and SNECMA, a state owned French industrial firm (currently being privatized), which makes the CFM 56 engine. GE also has a joint venture with P&W, the Engine Alliance. The other major joint venture is International Aero Engines (IAE), which was established by P&W, Rolls Royce, MTU and Japanese Aero Engines. However, the Commission regards the 3 major firms as controlling all joint ventures they participate in. 24 It should be noted that Fairchild Dornier subsequently went into bankruptcy and British Aerospace withdrew from this segment, ceasing to manufacture the Avr o regional jet.
Honeywell claimed that the Commission knew this at the time and should not have placed emphasis on this model in finding horizontal overlap in the engine market for large regional jets, as it was the only model for which Honeywell produced an engine. This would have been in line with the exclusion of the Fokker 70 and 100 LRJs which are no longer manufactured, and whose engines are solely supplied by Rolls Royce. Moreover, both GE and the DoJ defined the market based on the thrust power of the engines, which excluded the engines for the Avro, as their thrust was much lower than that of engines or other regional jets. See below for more details. 22 10 The Commission found GE to be dominant in the market for engines for large commercial jets. It calculated market share based on current platforms and firm future orders. It specifically excluded “aircraft that remain in service, but are no longer manufactured” (pg 14). GE objected stating that the current installed base is not relevant in isolation, and would not explain how it overtook the leadership position of Pratt & Whitney in the 1980’s. P&W had an 80% market share in the 1960’s.
In fact, it had a 90% market share as late as 1982 (Source: Back Associates Fleet Database, quoted in Emch (2003)). The turn around in the engine market was based on Boeing’s launch of the 737 with GE/CFMI engines. The 737 went on to become the most successful aircraft in history, selling 4,000 of 13,000 large commercial aircraft over 20 years (Emch, 2003). Although, the first generation of 737s’ were fitted with P&W engines boosting its market share, the second and third generation’s engines were sourced from GE/CFMI.
Therefore, it is clear that both GE and Rolls Royce emerged as competitive responses to P&W’s market dominance in the 1960’s and 70’s. However, GE has failed to achieve anything close to the same position P&W once enjoyed. Moreover, GE/CFMI is a joint venture in which GE only receives half of the revenue stream, unlike P&W which received 100% of all sales. This also suggests that even a 90% market share in an industry with high entry barriers can be challenged (that is, even when competitors would appear to be marginalized).
Moreover, GE stated that the market was a bidding one subject to constant competition, and market shares can be transitory, as the P&W case entailed. The latter argument was ignored by the Commission (the issue is further discussed below). Non-Avionics and Avionics Markets: Non-Avionics Products: These products include the basic components of an aircraft including wheels, brakes, landing gear, auxiliary power units, lighting and environmental controls. Avionics: These products are employed to control the communication and navigation of the aircraft, as well as monitor external flying conditions.
The market divides into two segments, large commercial aircraft (LCA) and regional corporate jets. Customers in the LCA market are the airframe manufacturers and also airlines and leasing companies. Avionics products are federated into a cockpit suite, which means that they can be chosen or changed by airline and leasing customers. This is the basis for a further segmentation: 1) Buyer Furnished Equipment (BFE): These products are chosen by the airlines. Standard avionics products tend to be buyer furnished. Such products are multi-sourced, and the final products are selected from 2 or 3 certified suppliers. ) Supplier Furnished Equipment (SFE): Chosen by the airframe manufacturer. Non-avionics products, (except those like wheels and brakes) 11 are supplier furnished. The competition between rival suppliers takes place at the time a new aircraft platform is being designed and developed. However, for regional and corporate jets, avionics products are sold as part of an integrated cockpit. Therefore, the ultimate customers have no choice over the products. The Commission defined Honeywell as having a leading position in the supply of aerospace components.
In the avionics market, the EC calculates it holds a 50% market share overall, followed by Rockwell Collins with 25%, Thales with 18% and Smi th Industries with 3%. In the non-avionics market, Honeywell competes against Hamilton Sundstrand (a subsidiary of United Technologies), BF Goodrich and SNECMA. Bidding Markets: The question arises whether the markets for aircraft engines and components can be characterized as “bidding market”, where competitors are required to repeatedly bid against each other in order to win orders. In their investigation, the US authorities defined it as such.
According to Shapiro and Patterson (2001) to determine this, the DoJ examined: (1) Whether multiple suppliers consistently entered the bidding contests to supply platforms and subsequently airlines? They found that this was indeed the case (Kolasky, 2002). (2) Did the customers determine that the competitors produced high quality alternative products, which gave them a real choice. Again, the DoJ answered this in the affirmative, after extensive surveys. Neither Boeing nor Airbus opposed the deal. 25 (3) Are competing firms able to preserve their competitive advantages despite setbacks?
The DoJ found that this was indeed the case, and that GE’s competitors were thriving. In the DoJ’s view, a high market share did not impact the ability of others to compete in future, as buyers did not regard it as a proxy for quality. If it was, then GE would never have been able to overtake Pratt & Whitney in market share. The nature of competition is such that emphasis is on future competitions, not ones that have already been decided. (4) Is the bidding process vigorous, with multiple rounds of bidding that force down prices? The DoJ found that this was the case, and significant discounts were necessary to win orders.
The European Commission confirmed this, but defined the discounts as evidence of the scope for anti-competitive practices on GE’s part, rather than evidence of vigorous competition in the marketplace. 26 (5) Where all competitors winning orders on a regular basis? As Kolasky (2002) states, “empirically, what we found when we examined the markets in which GE already competes, was that GE’s engine rivals are both investing just as heavily as GE in developing their next generation of engines and have had no difficulty in raising capital to finance that effort. Overall, Platt Majoras (2001)27 states 25 However, a number of the airframer’s customers did, although Lufthansa was the only airline to air its complaints openly. 26 The Commission states, “These heavy discounting practices actually resulted in moving the break-even point of an engine project further away from the commercial launch of a platform”. 27 Deborah Platt Majoras, Deputy Assistant Attorney General for Anti-trust at the time of the decision, now Chairwomen of the Federal Trade Commission. Quote from speech made to the Anti-trust Section of the Georgia State Bar in 2001. 2 that there was no evidence “supporting the EU finding that Rolls Royce and Pratt & Whitney were no longer in a position to constrain GE’s behaviour”. These findings fundamentally undermine the market foreclosure argument put forward by the EC. Therefore, in its approach, the Commission would appear to have ignored the dynamics of the market itself, and focused purely on market share, and a controversial definition at that (see above). 28 Unbundling Bundling: Having established GE’s market dominance, the Commission then turned to the issue of bundling.
There are two different types of bundling. Firstly, mixed bundling involves the sale of two products tied together as well as the sale of stand alone components. Mixed bundling can be incomplete if only one of the components, typically the tied good (in which the seller does not have a dominant position), is sold independently of the other. This is normally referred to as tying. The second type of bundling is pure bundling, where the 2 products are tied, and neither is available separately. (See Nalebuff, 2003 for more on these definitions).
The type of bundling emphasized by the Commission was mixed bundling29. Bundling: The Theory: In principle, bundling will allow sellers to undertake some profitable price discrimination. This arises because customers’ valuation for the bundle may vary less than their valuation for individual items. As a result, the pricing of a bundle can escape, to some extent, the problem of any firm facing customers with different valuations (a demand curve) such that an increase in price will increase revenues from customers with high valuation, but lose the custom of those with low valuations.
Bundling will increase profit, but it is not clear that welfare will fall and some customers will typically be better off. The Commission has defined bidders markets in previous cases such as Pirelli/BICC (2000) – there it defined it as where “tenders take place infrequently, while the value of each individual contract is usually very significant. Contracts are usually awarded to a single successful bidder (so-called “winner-takes-all” principle).
Strong incentives therefore exist for all for all competitors to bid aggressively for each contract” (quoted from Shapiro and Patterson, 2001). The Commission has argued that in GE Honeywell it did not define the market as a bidders one because of structural features such as high entry barriers, commonality issues and the resource differentials of the players. The EC regards bidding markets to be more important to cases of collective dominance, than simple dominance. (Source: Interviews with the Merger Task Force Team members – Brussels). 9 The Commission has found tying illegal where a firm holds a dominant position in relation to one of the products. See Hilti AG vs. Comm’n, Case C-53/92P, 1994 E. C. R I-667 (CJ), which relates to a German manufacturer of nail guns who attempted to exclude independent producers of nails for it products via tying the sale of its guns to its nails. The Commission has also found (complete) mixed bundling illegal when undertaken by firms who occupy dominant market positions. For example, Hoffman-La Roche v. Comm’n, Case 85/76, 1979 E.
C. R. 461 (CJ) and Michelin NV v. Comm’n, Case 322/81, 1983 E. C. R. 2461 (CJ). In the latter case, the European Court of Justice declared that “no discount should be granted (by a dominant firm) unless linked to a genuine cost reduction in the manufacturers costs”. Mixed bundling was a key element of the EC’s case, particularly in the Statement of Objections (May, 2001). The EC placed some emphasis on pure bundling as theoretical future behavior by the merged parties in relation to new generations of aircraft. 28 13
When bundling involves complements, a second effect comes into play. Complementary products are such that a fall in the price of one product will increase the demand for both; for instance, a decrease in the price of an engine will make planes more affordable and will thus increase demand for avionics as well as engines. When different firms sell the different components, they will not take this into account. For instance, a seller of engines will set its price considering only the effect that a change in price will have on the sale of engines.
When complements are sold by the same firm, the matter is different. When pricing engines, the firm will also consider the effect that a fall in price will have on demand for avionics, besides engines. As a result, prices will be lower30. This is the so called Cournot effect ; profits will increase but, since prices fall, consumers will also be better off . 31 Note that in those circumstances, pricing can be seen as more “efficient”. 32 The Cournot effect is strongest with firms implement pure bundling and faces no competition.
In the presence of competitors and mixed bundling matters are different. In the presence of competition, it is not clear that bundling will be profitable: lower prices for the bundle may trigger a response from competitors such that bundling is not attractive. Mixed bundling in the presence of competition may or may not be attractive depending on particular features of demand and in particular on the extent to which the decrease in the price of the bundle enlarges aggregate demand (as opposed to shifting market share).
However, bundling will also typically be more profitable when it involves a large number of components. This arises because of a Counot effect in “reverse” ; in lowering its price to lure away consumer of the bundle, the manufacturer of any given component will produce a large external benefit to manufacturers of other components. Indeed, as he switches away from the bundle, each consumer will have to choose a variety of components. This external effects will not be internalized and lead to higher prices (a weaker response) by competitors.
Overall, if the profitability of bundling is somewhat uncertain, bundling will generally tend to lower the demand faced by competitors in each market. They can expect to suffer both because of lower prices, and because bundling allows for some price discrimination (which effectively enhances the bundling firms’ ability to extract profits from buyers) 33. The extent to which they will suffer is however dependent upon may parameters and modeling assumptions. Note that in the presence of two monopolies, coordination of prices will suffice.
With competitors in either market, pricing coordination will not suffice as some of the benefit of lower prices in one market will accrue to competitors in the market. Bundling is then necessary to make sure that the stimulation of demand in the second market accrues to the firm selling the two items. 31 This effect was first pointed out by the 19th century French economist Antoine Augustin Cournot (18011877). Cournot analyzed the effect on prices of merging producers of complementary products, for example, copper and zinc used to produce brass in his illustration (see Cournot, 1838).
He found that if two monopoly producers of complementary products coordinated their pricing via a merger, not only would the aggregate price of the products fall, but the combined corporate entity would increase both revenues and profitability. Cournot’s model involved t wo monopolists producing single products and selling to a fragmented customer base producing brass. 32 This issue will be further discussed below. 33 GE went to substantial lengths to show that it was not free to act without regard to competitors or customers. However, ronically if the parties had monopolistic positions, the result on consumer welfare of the merger would have clearly been positive via an increased consumer surplus (see Nalebuff, 2003) – i. e. in the simplistic Cournot model – price for consumers would simply fall and stay at this lower level. 30 14 These effects also rely on the assumption that firms set a single price for their products. Yet, there are no list prices in the aerospace industry. V powerful buyers, Boeing, ery Airbus and the airlines individually negotiate prices. Meanwhile, the vendors themselves have significant information about the buyers’ preferences.
Both of these factors aid price negotiations, which precede the sale of products. In those circumstances, bundling is obviously less attractive. It does not help in terms of reducing the heterogeneity of buyers facing a single price, as individual prices are set for each deal. In addition, it does not help in terms of exploiting the Cournot effect: firms will not obtain additional rents from stimulating demand across products simply because competitors (who also negotiate with the individual buyers) will force the price down to their marginal cost on individual components.
All the firm can obtain is the rent associated with customers’ particular preference for each of its product, whether they are bundled or not. In those circumstances, the transfer of profits from competitors to GE is small, as the buyer already had a preference for GE products which limited rivals sales. Moreover, bundling has the least effect where rivals are the preferred supplier for both engines and avionics products. Where the buyer has a preference for one GE product and the other from a rival, it is unlikely they will choose the bundle, as there is substantial evidence that components are chosen on a “best of breed” basis.
Hence, the consequences of mixed bundling are highly dependent upon particular circumstances, and a stylized economic model can hardly provide conclusive evidence. Both the models presented to the Commission by Frontier Economics and Professor Jay Pil Choi34 on behalf of Rolls Royce and by Professor Barry Nalebuff35 for GE are highly stylized and they have not even been calibrated to actual data. For example, the Rolls Royce model only has 2 competitors and 2 products. Whether there is actually an economic incentive to bundle in the aerospace ndustry can hardly be inferred from these exercises, nor can buyer and competitor responses and the possible losses to competitors. To model bundling realistically, one needs to know the number of items in the bundle, the total market demand elasticity, the preferences of each customer, the supplier’s degree of awareness of these preferences and how much these preferences are likely to shift over time due to exogenous shocks (Nalebuff, 2003). Estimates of these parameters are naturally surrounded with much uncertainty (and all the more so because their development needs to be anticipated).
When the US authorities examined these same issues they found that “such conduct is best addressed if and when it occurs, rather than ex ante, at the time of the merger when we have none of the facts we would need to determine whether the conduct, if it even occurs, would in fact be anticompetitive”(Kolasky, 2002). Bundling: The Practice: This leaves us to examine current practice. The Commission put forward evidence that Honeywell had engaged in bundling, backed up by evidence from Rockwell Collins.
However, in its decision to clear the Honeywell AlliedSignal merger, the Commission’s case team found that bundling was not a widespread practice 34 35 Professor of Economics, Michigan State University. Professor of Economics, Yale School of Management. 15 in the avionics and non-avionics markets. 36 They found that the characteristics of the industry are such that bundling is difficult. This was because: 1) As stated above, there are no list prices for products and transactions are individually negotiated. Buyers choose the best product on technical specifications. 2) The procurement process is long. ) Concentrated and powerful buyers are able to play sellers off against one another. 4) Suppliers regularly re-price all products over a 2 month period. The Commission stated that Honeywell had engaged in a number of multi-product bids which were bundles by another name. However, the bundle discounts were small (Nalebuff, 2003), and customers were able to break-up the packages, picking “best of breed”, whilst still able to retain the bundle discount. Therefore, if bundling was taking place, it clearly was not very effective in displacing buyers away from competitors.
The Commission presented confidential evidence of a number of “multi-product bids” by Honeywell, which they claimed proved that despite the hurdles, bundling is possible. In the original decision (see pages 127 – 130), the Commission gives a number of examples such as a bid for a US airline by AlliedSignal, where extended warranties were offered for purchasing two products together. We are not told by the Commission what AlliedSignal’s competitors were offering, the value of the extended warranties or if the bid was successful (in whole or in part), so it is almost impossible to draw any conclusions from this example.
One other example is offered, where more products were offered in a bundle, although again the value of the discounts is not made clear, nor whether the bundle was purchased in whole or in part. 37 The above examples relate to large commercial jet aircraft. The Commission goes on to present examples of bundling in the area of corporate jets. In one example, the airframer requested a bundle for non-avionic components, having already selected Honeywell as the avionics supplier. Again, the facts are limited, and we are not informed of the outcome.
Subsequent investigation by the authors demonstrates that although, according to the Commission, the manufacturer requested a bundle for non-avionics systems including ECS, APU, electrical systems and wheels/brakes, the firm was not successful in selling the entire bundle. 38 While Honeywell was chosen to supply the ECS and AP U, Hamilton Sunstrand (UTC) supplied the electrical system and ABSC the brakes/wheels. This example would seem to demonstrate the industry practice of unbundling bundles, and selecting components on a “best of breed” basis, whilst retaining substantial discounts. 6 The unit of the Merger Task Force headed by Enrique Gonzalez Diaz which investigated GE/Honeywell was different from the one that dealt with Honeywell/AlliedSignal. In the US the same DoJ team dealt with both. 37 One is tempted to presume that if these “multi-product bids” had been accepted by the airlines, the Commission would make it very clear that this was evidence that bundling was being practiced in the aerospace market. As it stands, these cases are forwarded as evidence that “bundling is feasible” (page 127). 8 Source: Confidential correspondence with the manufacturer – on file with author. 16 Ultimately, the Commission only presents one case, of a mid-size corporate jet, in which bundling (admittedly of engines and avionics) can be said to have taken place successfully. Moreover, the complaints of 2 rival components manufacturers, UTC and Litton, maintained that bundling didn’t take place. 39 The EC also omitted vital evidence from Embraer, other OEMs and 5 airlines stating that bundling didn’t occur. Airbus gave evidence that it breaks down all components and examines each on its merits.
Also Boeing and Bombardier require bids to be itemized. Airbus’s policy is to maintain “dynamic competition”. In fact, its CEO Noel Forgeard stated in a letter to the Commission that the company did not oppose the transaction, and that GE/HW were important, but not dominant, and competitors were gaining ground. He went on to state that Airbus own strategies could deal with any anti-competitive threats should they arise. 40 Ignoring this evidence, the Commission goes on to speculate that the merger would then allow GE to bundle engines with avionics and non-avionics components.
However, where GE is a sole source engine supplier, for example on the Boeing 737, the customers negotiate the purchase of the plane with Boeing, giving GE no room to attempt to bundle other components. 41 Moreover, on multi-source platforms, the engines are chosen a significant period before the avionics and non-avionics components. Again, this makes it impractical to bundle. 42 While GE could offer a subsequent discount on avionics, this is a problematic process, since the subsequent pricing of components is not based on official price lists, but is negotiated.
As Nalebuff (2003, pg 37) states, “all prices are negotiated so a discount off the list price has no bite. One person can’t promise to give the other party a better deal in future negotiations, as there is no baseline against which to measure what makes a better deal. ” However, even if bundling becomes a common future practice, the Commission takes no account of strategic counter moves by competitors. For example, even if competitors were reluctant to offer counter-bundles, possibly due to the negotiating hurdles relating to which artner would cut prices more or unwillingness of current managers to give up power in a merger, it is likely the market would require them too (i. e. negotiations with Boeing and Airbus would effectively concentrate their minds, as may their institutional Source: Oral arguments at 25th May 2004 Honeywell hearing in Luxembourg. In a breach of procedure, the Airbus letter was never revealed to the parties or the Commission’s advisory committee prior to the decision. The summary of the letter above is based on the limited contents revealed during the appeal hearings at the Court of First Instance.
The Commission has argued that the contents of the letter were manipulated by GE. As evidence they point to Forgeard’s admission in the letter that from Airbus position, “remedies could be negotiated”, as heavily indicative of the strength of their case (Source: Interviews with Merger Task Force Team members). 41 As stated above, Boeing pre-negotiated an engine pricing deal with GE to protect its customers (Nalebuff, 2003). 42 The Commission presents some evidence that selection timelines are flexible enough to allow bundling to take place.
However, again the only conclusive example is the mid-size corporate jet, for which Honeywell bundled avionics and engines. 40 39 17 shareholders). 43 The end result of competition between bundles, again assuming bundling is economically viable, would be unambiguously good for customers and consumers. Such an outcome may have driven Airbus endorsement of the deal. It is also important to note that even if competitors would not offer counter bundles, the extension of the product line over which bundling occurs, while increasing the attractiveness of bundling, would also reduce the extent to which competitors would be affected.
Indeed, bundling can then easily lead to an increase in all prices. That may not be attractive from the point of view of consumers but indicates that the Commission’s concern, that competitors would be marginalized, would become irrelevant. This observation further emphasizes that the presumption such that competitors would be marginalized is not robust and accordingly cannot be presumed. After the publication of its Statement of Objections on May 8, 2001, the Commission began to downplay the bundling arguments, instead relying on a roader theory of Archimedean Leveraging. 44 Archimedean Leveraging: The role of GE’s aircraft leasing company, GE Capital Aviation Services (GECAS) is at the heart of the final decision to block the merger. This division of GE engages in a number of transactions with airlines. A large percentage of its business relates to leaseback agreements, whereby airlines sell planes they already own to GECAS and then lease them back. These transactions improve airline balance sheets, and create tax and depreciation benefits.
However, GECAS has no influence over the choice of aircraft in the deals, as these have already been purchased by the airline. Therefore, the Commission concentrated on the speculative purchases of new aircraft by GECAS. This is where the leaser purchases planes for which it does not have a final customer. It then looks to lease these to airlines with short operating leases (usually 8-9 years, Emch (2003)). Such a transaction allows the airline to lower its risk profile and may be efficient because GE is better able to bear the risk.
Ultimately, consumers may benefit. Archimedean Leveraging:The Theory: As stated earlier, the Commission believed that substantial engine market share shifting to GE had been driven by GECAS “GE only” policy in the speculative purchase market. The EC also feared that this policy would be extended to Honeywell components, hence further transforming a leading market position into a dominant one, leading to or re-enforcing market foreclosure. This was a theory put forward by Reynolds and Ordover (2002) on behalf of United Technologies.
This theory of “pivotal” or “Archimedean” leveraging is built around the following elements; assume 43 It’s an interesting question as to whether the Commission should concern itself with the fates of managers who forego pro-competitive, value enhancing mergers at the expense of shareholders, to retain there own positions. Moreover, it is ironic, as Kolasky (2002) points out, that the Commission downplays the effectiveness of joint ventures as a competitive response, whilst describing in detail how the CFMI JV between GE and SNECMA has been so successful. 4 This switch in emphasis was criticized by the Financial Times. “More than once Commissioner Monti allowed his officials radically to shift ground for their objections, creating the unfortunate impression that EU anti-trust policy was being made on the hoof. ” (Hill and Done, 2001). 18 that a firm is present both as a supplier of components to manufacturers (Airbus/Boeing) and as a buyer (GECAS) of final products from these manufacturers. This buyer is, however, not a final user of the products. It packages the aircraft with a financing scheme and sells the packages to final users (airlines).
Final users can alternatively buy the aircraft themselves and obtain financing from different sources (these alternatives being substitutes for one another). Assume further that the manufacturers choose the components (hence, components are supplier furnished equipment – SFE – or embedded) and a single brand of component is selected for each model produced by the manufacturer (choices of components are once and for all and exclusive). Finally, assume that final users are rather indifferent to the choice of the component made by the manufacturers (in other words, there is little product differentiation among them).
Consider the purchasing policy of the integrated firm. Faced with two aircrafts sold at the same price, one with its components and one with other components, it will naturally choose the former, as a purchase of this product will also increase the profit of its affiliate, unless of course, its choice triggers changes among other buyers which reduce demand for its components. More generally, faced with a given price for the aircraft, the integrated company will choose to charge a lower price towards airlines for its bundle as it has a higher margin on each aircraft sold (unless there is a compensating change in the behavior of other buyers). 5 Hence, the integrated firm will have a higher demand for aircraft (at a given price of the aircraft) if they include its component. Assuming that the price of financing airlines can obtain when they organize their own is fixed, aggregate demand increases. Consider now the behavior of the intermediate manufacturers. In the absence of integration between a buyer and an upstream component supplier, it would select components solely in terms of prices. Faced with an integrated firm, manufacturers will as indicated above, face a larger aggregate demand for their product.
Hence, they will find it more profitable (at given prices for the components) to produce an aircraft which includes the component of the integrated firm. 46 Prices for the components however, need not be constant; one would indeed expect that the competitors will charge a lower price. Assume, to fix ideas, that the integrated firm does not change its price (an assumption which is favorable to the competitors). In order to convince manufacturers to choose their components, they would have to charge a price which compensates manufacturers for the increased sales which are triggered by the choice of the component produced by the integrated firm.
Given the value of the components relative to that of the aircraft, they may not be in a position to do so. As told, the theory of “pivotal” leveraging relies on extreme assumptions and in particular the assumption that components are not differentiated. In those circumstances, even a small presence of the integrated firm as a downstream buyer will change the behavior of the manufacturer. This assumption, at least in such extreme form, is probably not consistent with the facts; it implies, in particular that the prices of the 45
This is, in essence, a Cournot effect again, as the integrated firm considers an external effect across complements (components and financing). 46 And may also increase the price of the aircraft. 19 components should be driven to marginal cost. Given the importance of fixed costs in the development of components, this would in turn imply that firms make losses in the component market, which cannot be an equilibrium situation. The question then arises to what extent “pivotal” leveraging still operates when there is some product differentiation in the component market.
Product differentiation will clearly reduce the extent to which the choice of component will be affected by integration (moving away from this would otherwise be an optimal choice and will be costly for the manufacturers) but should not annul it altogether, and it makes integration less attractive but not necessarily unprofitable. Pivotal leveraging also works in the example above even if the integrated firm has a small presence in the downstream market. At the extreme, any shift in aggregate demand can tilt the choice of the manufacturer.
When there is some product differentiation among components, the magnitude of the shift, and hence the market presence of the integrated firm in the downstream market, will matter. Hence, whether pivotal leveraging can work in more realistic circumstances, is inherently an empirical question. At the same time, the specificity of this construct is such that it can hardly be presumed. In order to give some credence to pivotal leveraging, the Commission considered the effect that GECAS had with respect to the choice of engines.
Engines are typically not supplier furnished (embedded) and hence do not fit with the assumptions of the model underlying pivotal leveraging, but the Commission found it informative. Indeed, if GECAS was able to substantially shift purchases of buyer furnished equipment like engines, one would expect that it would a fortiori be in a position to do so with a SFE. In the context of BFE, suppliers will not be exclusivity upstream. The integrated firm can nonetheless affect the relative position of these suppliers through its own purchasing policy, like a GE only policy for engines fitted on aircrafts bought by GECAS.
Archimedean Leveraging:The Practice: Some debate first arose with respect to the market share of GECAS. As Gotz Drauz, (then head of the merger task force) (2002, pg 194), states in his subsequent defense of the decision, “GECAS is the largest purchaser of new aircraft, ahead of any individual airline and or other leasing company. GECAS is also reported to have the largest single fleet of aircraft in service, as well as the largest share of aircraft on order and options”. However, he neglects to mention GECAS low market share in a fragmented and competitive market. The EC states GECAS market share as approximately 10% 47.
However, Nalebuff (2003) argues that this overstates the figure. He gives evidence that it is closer to 7% over the last 5 years as GECAS aircraft purchases are irregular, and it often has to clear a backlog. In dollar ter