AAI attacks Alcoa's proposed acquisition of Reynolds, in letter to Klein and monograph

Feb 23 2000
Testimony and Interventions

The Hon. Joel KleinAssistant Attorney General for AntitrustU.S. Department of JusticeWashington, DC 20530

Dear Mr. Klein:

On February 23, the American Antitrust Institute sent you an open letter detailing why the proposed merger between Alcoa Inc. and Reynolds Metals Company poses a severe competitive threat in the market for alumina, the main feedstock for the smelting of aluminum. The AAI urged that the U.S. Department of Justice either reject the merger outright or condition approval on divestitures that would fully address the competitive concerns. We were dubious that our concerns could be satisfied by divestitures.1

On March 14, events occurred in Europe that are relevant to the positions taken by the AAI in its report and letter. Antitrust authorities in the European Community announced that they would allow a merger between Canada's Alcan Aluminum Ltd. and Switzerland's algroup, provided the firms make certain divestitures. At the same time, the EC announced that Mario Monti, the Commissioner in charge of competition policy, had proposed that his colleagues block the $7-billion merger between Alcan and France's Pechiney SA. However, before the Commission acted on this latter merger, Alcan and Pechiney withdrew their proposal, saying they would retool it to better address antitrust concerns and then resubmit it to the EC.

All of these deals were announced last August. First, Alcan announced its separate offers to buy algroup and Pechiney, and its hope to merge the three companies together into a $21.6-billion-a-year company called "APA." Then, about a week later, Alcoa and Reynolds agreed to merge as well, creating a company with about $20.5 billion in sales. These proposed mergers would make Alcoa/Reynolds and APA by far the largest aluminum companies in the world.

When the proposed mergers were announced, some analysts felt that they might be mutually re-enforcing. EU officials, it was thought, might approve the APA merger in part to boost European power in the aluminum industry, since European players might now have to face a merged Alcoa/Reynolds across the Atlantic. Meanwhile, enforcement officials at the Justice Department might be swayed by the fact that Alcoa/Reynolds would have to compete with a much more powerful Alcan across the Canadian border, assuming the APA merger was approved.

The three APA companies still hope to carry out their merger. However, the EC has already said that it would not accept various proposals by Alcan and Pechiney to address competitive concerns.2 So it is now possible-perhaps likely-that the Alcan/algroup merger will go forward without the companion merger of Alcan and Pechiney.

Now that the EC has shown a willingness to block the larger of the two APA deals, the AAI urges the Justice Department to do its part to maintain competition in the global aluminum business.

If only the Alcan/algroup portion of the merger goes through, and Pechiney remains a separate company (as appears possible, if not likely), the AAI's conclusion about the anticompetitive effects of the Alcoa/Reynolds merger is affected hardly at all.

The market singled out by the AAI in its February 24 report was the market for third-party Western metallurgical alumina.3

As the report explains, if one assumes that the three-way APA merger is approved prior to a decision on Alcoa/Reynolds, the Hirschmann-Herfendahl Index (HHI) would increase from 2396 to 3119 due to the Alcoa/Reynolds merger, and the 4-firm ratio would climb from 73% to 80%. (The AAI used projected year-2001 market shares, and a market definition as given above.)

On the other hand if, as now appears likely, only the Alcan/algroup portion of the APA merger precedes the Alcoa/Reynolds decision, the numbers change slightly. In 2001, algroup will be a net producer of metallurgical alumina, in the amount of around 775 kt.4 Alcan by itself would have a deficit of roughly 200 kt. A merged Alcan/algroup would thus have a surplus of 775 - 200 = 575 kt, whereas the merged APA (including Pechiney) would have been roughly "balanced," producing as much alumina as it consumed. Thus, the two-way Alcan/algroup merger would leave a small seller in the market, which would not exist if the full APA merger were to occur.

Thus, if only the two-way merger is allowed, the Alcoa/Reynolds deal increases the HHI from 2236 to 2907, and the 4-firm ratio increases from 71% to 77%. The HHI increase is not quite as great as with the full APA merger (+671 instead of +723). And the absolute numbers are lower by about 200 points. But in either case, the Alcoa/Reynolds deal still falls far outside the safety zone specified in the Justice Department's merger guidelines.5

On March 17 Bloomberg reported that Alcoa was offering to sell Reynolds's alumina refinery in Corpus Christi, Texas, and 50% stake in a refinery in State, Germany, to satisfy EC competition officials. According to the Bloomberg story, Alcoa had not offered to divest Reynolds's 56% stake in the Worsley refinery in Western Australia.6

Alcoa and Reynolds were expected to offer the Texas and German plants to appease competition authorities. As described on pp. 50-51 of the AAI report, North American and European refineries are among the highest-cost facilities in the world, running variable costs of $170-$190/ton.

Refineries in Western Australia have very low costs ($100-$130/ton), and are known to have excellent expansion opportunities. (The Worsley plant "is the world's lowest-cost facility in the industry and has the capacity for massive expansion," the Bloomberg report said, citing the opinions of analysts.) AAI has noted that the four refineries of Australia's Darling Range-three controlled by Alcoa, the fourth, Worsley, majority-owned by Reynolds-represent a huge fraction of the world's low-cost expansion opportunities, that is, opportunities for adding significant capacity to existing low-cost refineries. All four refineries would be united under an Alcoa/Reynolds merger.

If the world were to sustain a period of moderate alumina prices, it may very well be that virtually all of the clearly profitable expansion opportunities would come from the Darling Range, because of the region's low cost. With two competitors in the region, as we have now, we might see that expansion occur. Indeed, a 1.2-Mt expansion is underway at the Worsley plant at this very moment. An independent Reynolds may choose to expand more in the future, because the benefits to Reynolds of the additional sales might outweigh the harm from lower alumina prices worldwide-since Reynolds's existing alumina sales in 2001 will amount to only about 1,200 kt. On the other hand, a merged Alcoa/Reynolds, with projected global 2001 sales of 8,000 kt, stands to lose nearly seven times more from lower alumina prices,7 and may therefore use its influence to prevent expansions in the Darling Range. Thus, the world would have to wait for a period of high prices before any expansions would occur.

We believe that this is a likely scenario by which a merged Alcoa/Reynolds would use its Western Australian assets to retard the expansion of global alumina supply. Alcoa's proposed divestitures of the Corpus Christi and State plants do nothing to address this problem. The AAI therefore repeats its advice to antitrust authorities to consider the likely competitive impact of a single company having substantial influence over production decisions at all four of the alumina refineries in Australia's Darling Range.


Albert A. FoerPresident

Matthew SiegelResearch Fellow

1See the following link for the AAI letter and a link to the full report.

2 See the following link for a detailed description of these proposals and of the EC's reasons for rejecting them.

3 The report defends this market definition in detail. "Third-party alumina" is alumina currently being sold to third parties rather than consumed in the producer's own aluminum smelters. "Metallurgical" alumina is alumina that is physically suitable for the smelting of aluminum. And "Western alumina" excludes all alumina refined in China and the formerly communist countries of Eurasia (the Eastern bloc).

4 This estimate comes from Alcan. The company currently has a surplus of about 600 kt; it's new smelter, which will probably start operating toward the end of 2000, will require about 600 kt in 2001; likely re-starts of idle smelting capacity will require about 250 kt, and alumina refining may increase by 50 kt or so due to more intensive use of existing assets. That leaves a net production of 600 kt - 600 kt - 250 kt + 50 kt = - 200 kt.

5 Any merger that leaves the HHI at more than 1800, and that represents an increase of more than 100 points, is presumed to be anticompetitive. See this link.

6 See the following link for the Bloomberg story, written by Seena Simon.

1 That is, 8000/1200 = 6.67.


The 53-page monograph "Anticompetitive Effects of the Proposed Acquisition of Reynolds Metals by ALCOA" can be downloaded here in Adobe Acrobat.Acrobat Reader can be downloaded for free from Adobe.


The Hon. Joel KleinAssistant Attorney General for AntitrustUS Department of JusticeWashington, DC 20530

Dear Mr. Klein:

The American Antitrust Institute is an independent non-profit organization that advocates a strong role for antitrust in the national economy. We have been studying the public record and consulting with experts in the aluminum industry in order to understand the competitive impact of the proposed acquisition of the Reynolds Metals Company by Alcoa, Inc. A detailed briefing paper by AAI Research Fellow Matthew Siegel is attached and available on our website, www.antitrustinstitute.org. We are also making this available to the media to enhance their ability to report on this important merger.

Our conclusion is that the proposed merger between Alcoa and Reynolds poses serious competitive problems in the market for metallurgical alumina, the feedstock needed for aluminum smelting. (We do not express an opinion on the markets for other products.) The Antitrust Division, in our opinion, should probably disallow the deal entirely, since it will be very difficult to find appropriate buyers for divested assets who are likely to assure the current level of competition. Moreover, the Division should analyze this merger with the Clayton Act's incipiency nature clearly in mind: this merger, coupled with the large pending merger of Alcan Aluminum Limited, Pechiney S.A. and Alusuisse Lonza Group AG (algroup), is likely to trigger additional mergers that will further concentrate an already concentrated industry. Serious antitrust issues in this particular merger ought to be dealt with in a way that discourages any likely lemming effect.

Alumina represents a significant part of the cost of smelting aluminum-usually amounting to roughly 25% of the selling price of an aluminum ingot. Of the roughly 3.8 million metric tons (3.8 Mt) of non-recycled aluminum produced in the U.S. each year, some 35% is made by independent smelters, which must buy their alumina on the open market.1 The rest of the non-recycled aluminum is produced by integrated producers, such as Alcoa, Reynolds and Alcan, which refine their own alumina and therefore do not need to buy alumina on the open market. We focus on the buyer most at hazard from rising alumina prices: the independent U.S. smelter.

The relevant geographic market includes most but not all of the world. The independent U.S. smelter can only buy alumina from the "West"-not from China or formerly communist Eurasia, because this Eastern alumina is unreliable, low in quality, and in many cases produced at very high cost.2

The relevant product market reflects the fact that an independent smelter cannot buy alumina that an integrated producer makes for its own smelters only, since that is by definition not available in the market. (Our research has revealed that, for various reasons, the alumina that a company makes for its own smelters is essentially never made available to outside buyers, no matter how high alumina prices may go.) Also, we are only concerned with the 91% of the world's alumina that is "metallurgical grade," that is, fit for smelting. So, in our view, the appropriate market definition for antitrust purposes is the market for Western-produced metallurgical alumina for sale to third parties-so-called "third-party alumina."

We estimate the shares for this market as follows:

Forecast Net Production of Western Third-Party Metallurgical Alumina for 2001Company Forecast
Net Production [kt]3
Market ShareAlcoa
4.9%Jamaican Govt.

If the proposed 3-way merger among Alcan, Pechiney and algroup occurs, it will have the effect of removing algroup from third-party sales, because algroup's current third-party alumina will be consumed within the enlarged company. In that case, the Hirschmann-Herfendahl Index (HHI) before the Alcoa/Reynolds merger is 2396, and after the merger is 3119. Even if algroup were to remain in the market, the pre-merger HHI would be 2191, and the post-merger figure would be 2844. In both cases, the HHI ends up over 1800 and increases by well over 100 points. That puts the Alcoa/Reynolds merger in the Merger Guidelines' "danger zone," where it is presumed likely to lead to higher prices, unless that presumption can be overcome by other factors.

Let's examine additional potentially relevant factors. It could be that certain firms that do not now sell third-party alumina should be considered part of the market, since they could enter opportunistically, very quickly, and at low cost-so-called "uncommitted entrants." But all candidates for easy entry prove to be unlikely uncommitted entrants:

  • Integrated producers virtually never shut down aluminum smelting capacity to free up extra alumina for third-party sales. No one has done this even since the Gramercy explosion, which ushered in a period of extremely high alumina prices.
  • In the West there is usually very little idle alumina-refining capacity that could be pressed into immediate service. (In our report, we suggest that as of last July the West probably had about 385 kt/year of true idle capacity, of which only 50 kt/year is in the hands of companies that do not already sell to third parties, and thus could "enter" the market.)
  • No one is quite sure how much true idle capacity exists in the East. But some who are knowledgeable about the industry believe it is very little or none at all. In any case, for various reasons, one cannot count on Eastern entry or supply expansion in response to higher Western prices, even if these strategies would make economic sense.
  • Even if the East and West together were able to supply 2.0 Mt's worth of uncommitted entry-a number that is much higher than any reasonable estimate of idle capacity-the Alcoa/Reynolds merger would still cause an impermissibly high increase in the HHI.

What about long-term entrants?

  • To plan and build a refinery takes 4-5 years: too long to avert harm.
  • Entry is rare in this industry: During all of the 1990s we know of only two cases of significant entry. One was fleeting, a by-product of Alcan's merger and acquisition activity; the other put a mere 150 kt of new alumina into the third-party market.
  • An existing refiner could expand an existing plant in 12-18 months, which would not be too long for meaningful entry. However, only five Western companies are likely candidates for this sort of entry, and two of them would be combined in the APA merger.4
  • All large potential Western entrants produce aluminum as well as alumina. They thus have an incentive to keep alumina prices high, so that prices will stay high in the downstream aluminum markets into which they sell. This is also a strategy which has the effect of raising rivals' costs, thereby giving the integrated companies a competitive edge.

In some cases, merging companies can realize efficiencies that offset any likely harm to buyers. Alcoa/Reynolds is not such a case. Almost superhuman efficiencies would be required to counteract the enormous increases in market concentration. Alcoa is claiming only $200 million in efficiencies (equivalent to a 1% across-the-board price hike by the combined company5), and much of that savings would likely come from fixed costs, such as layoffs among Reynolds's headquarters staff. Fixed-cost savings do not typically affect pricing dynamics, and therefore are unlikely to benefit buyers. The only other plausible efficiencies are due to Alcoa's allegedly superior management. This sort of "management efficiency" (the possibility of which we do not deny) tends to be less merger-specific and less verifiable than other types of efficiency, according to the Merger Guidelines, and therefore should be weighed lightly in antitrust analysis.

Thus, the merger appears to be illegal because of its tendency to substantially lessen competition in the market for alumina. Can it be "fixed" by negotiated conditions?

Making this merger work through divestitures may prove particularly tough. First, to maintain the same level of competition in alumina as existed prior to the merger, it would be necessary to divest the equivalent of Reynolds's capacity for producing third-party alumina. This entails finding a buyer that will in fact both continue to produce the alumina and sell it into the third-party market. A buyer that would use it to feed internal growth or to reduce its own purchases from the market should not be satisfactory.

Further, the most likely potential buyers are integrated producers of aluminum, and (as we have noted) they will have a natural incentive to keep alumina prices high, since doing so raises the costs of their competitors in the downstream aluminum market.

Finally, we note that a huge fraction of the world's low-cost expansion opportunities lie in the Darling Range, a region of Western Australia. At the moment there are four refineries in the Darling Range, three controlled by Alcoa, the fourth 56%-owned by Reynolds. Under the substantial control of a single company, the Darling Range could be used as a lever point to control global prices. Alcoa/Reynolds could drag its feet on future expansions in the region, and could use the potential for low-cost expansion to inhibit any construction of new refineries around the world.6 This suggests that at least some divested assets would have to come from the Darling Range area.

We hope that this analysis is useful to the Antitrust Division and to those who will be covering this transaction in the media. Our detailed briefing paper is enclosed.


Albert A. FoerPresident

Matthew SiegelResearch Fellow

1 These independent smelters include Glencore and Columbia Falls Aluminum Company (to merge); Goldendale Aluminum Company; Noranda; Northwest Aluminum Company; NSA, a division of The Southwire Company; and Venalco. Total alumina costs to the independent smelters amount to around $540 million annually, when alumina prices are at typical levels. (At the moment, those prices are about double their normal level, due to the explosion last summer at a Kaiser-owned alumina plant in Gramercy, Louisiana.)

2 Transport costs are relatively small, as long as no major overland transportation is required.

3 "Forecast Net production" is a company's forecast production of alumina, minus its smelters' forecast consumption of alumina. Except for (small) changes in inventory, this is exactly equal to "net sales": the quantity sold in the open market minus the quantity purchased in the open market. We believe this provides an accurate measure of whether a company should prefer high market prices or low market prices, and is therefore a valid index of market share for antitrust purposes.

4 We are referring to Alcan, Pechiney, Hydro Aluminium, Billiton, and Comalco.

5 Annual sales of the combined company would be about $20.5 billion.

6 Alcoa/Reynolds could render any such construction uneconomic by simply expanding its Australian production after construction has already begun on the green-field plant. Expansion is both quicker than new-plant construction and only half as costly per ton of capacity.

The 53-page monograph "Anticompetitive Effects of the Proposed Acquisition of Reynolds Metals by ALCOA" can be downloaded here in Adobe Acrobat.Acrobat Reader can be downloaded for free from Adobe