The Need for Different Metrics
Peter C. Carstensen
Young-Bascom Professor of Law
University of Wisconsin Law School
975 Bascom Mall
Madison, Wisconsin 53706
Prepared for the Workshop on Merger Enforcement
held by the Antitrust Division and the Federal Trade Commission
It is a honor to be included in these review discussions and a special privilege to be included in a panel with such distinguished scholars. My interest in the problems that buyer power poses for antitrust analysis comes out of my study of the problems farmers and ranchers face in the marketing of their products. Because of the disproportionate allocation of market power in these markets, the producers of agricultural products face a number of problems in achieving fair, open, accessible, efficient and transparent markets. While some of those problems can only be addressed through reform in the structure of the laws that constitute and govern such markets including the Packers and Stockyards Act, the Agricultural Fair Practices Act, the Agricultural Marketing Agreements Act and the Capper-Volstead Act,(1) the structure of buying markets plays a very significant role in both creating incentives for and the opportunity to exploit buyer power. Mergers that increase such power, both in agricultural and non-agricultural markets, present significant threats to both the short run and long run efficiency of those markets in static as well as dynamic terms.
While antitrust law lacks robust and fully developed economic criteria for determining the likely competitive harms that come from increased buyer power,(2) we do have some on the ground observations that are highly suggestive of the kinds of issues and concerns that ought to be considered. Moreover, and of great importance, we have a commitment by the Division's leadership under both the past and present administrations to pursue buyer power cases even when there is no explicit consumer harm. This session of this workshop betokens a further commitment to continue to evaluate the potential of such mergers to cause undesirable competitive effects. It is important to identify relevant types of potential competitive effects and the kinds of market characteristics and structural indicia that suggest when such effects are reasonably likely to occur.
For these comments, I take these commitments to make it unnecessary in this forum to argue for the basic proposition that buyer power deserves attention from antitrust authorities. Hence, I want to focus on the ways in which buyer power issues are different from those presented in most seller side evaluations. My fundamental objective is to demonstrate that buyer power analysis requires metrics that measure both power and effects grounded in the economic realities of the buying side of the marketplace.
The oft repeated declaration that monopsony is the mirror image of monopoly stands in the way of the kind of thoughtful and critical evaluation of such transactions that good public policy requires. Of course, exploitation and exclusion whether of sellers or buyers has a central commonality. Moreover, most of the effects that are prominent on the buying side of the market have analogs on the selling side. The analysis of buyer power requires the same tools and economic sensitivity to the problems of coercion, exploitation and efficiency that guide merger analysis on the seller side. But the metrics by which likely effects should be measured and the more specific typology of likely effects necessarily must reflect the difference between the economic context of buying and selling. Consequently, the specific analysis will be different exactly because buyers have different incentives than sellers, the markets are different, and the barriers to effective competition on the buying side can be qualitatively different. Hence the conventional criteria for judging whether a merger may merit further inquiry will not identify all the transactions that should be investigated. I want to highlight today some of the observations that I have made concerning both the likely adverse effects on competition and the appropriate measures of markets and market power on the buying side of markets. I should also acknowledge as Assistant Attorney General Pate did in October to the Senate Judiciary Committee that more work needs to be done to gain a better understanding of the competitive implications of increased buyer power in various market contexts.
These comments, first, identify key considerations in understanding the implications of the buying side impact of mergers and acquisitions. There follows a discussion of how these points relate to elements of merger analysis.
Although I am not an economic theorist or modeler, I do observe patterns in the reported cases and empirical and theoretical work of economists who have evaluated the issues associated with buyer power. This body of information suggests the following salient points about buyer power and its anticompetitive exploitation. The following paragraphs provide a brief review of the concerns that I feel are central to an understanding and evaluation of mergers that affect buying power.
In evaluating the risk of anticompetitive consequences from a merger that increases market power on either the buying or selling side, it is important to consider the incentives of the merged firm to engage in such conduct, the kinds of conduct that power makes possible as well as the potential market forces that might deter such conduct. In selling markets, there is often a rational basis to be skeptical about the gain to the seller from trying to raise price either through coordination or unilaterally because of the potential that competitors would undercut that price. Hence, it is often said that there needs to be an alignment of interests and incentives for an oligopoly to engage in coordinated market exploitation and that unilateral efforts similarly require concern for both the barriers to new entry and to the potential for existing competitors to respond.
While in general the same considerations are relevant on the buying side, there are important differences with respect to incentives and opportunities. Moreover, the potential deterrent effect of competitors is, in many circumstances, less likely to exist than is the case on the selling side of the market. An important analogy is to the work on auctions that shows that bidder collusion is a substantial risk because of the strong incentives to participate in such conspiracies and exploit sellers.(3) Moreover, auction cartels are rational despite revolving membership exactly because of the mutual advantages of such conduct. Further, potential bidders do not have the same capacity to disrupt such cartels as occurs on the selling side of the market.(4) All of this follows because the object of collusion is the bid price and not the price at which competitors will sell their products.
This observation generalizes to many buying market contexts. A buyer wishing to compete must raise its price to the seller. This will raise the buyer's costs of production and so make its downstream products more costly. Thus, all buyers in any market have a shared interest in keeping prices for inputs down. Only if the demand for the input exceeds supply at a current price will buyers have any incentive to raise prices. "Cheating" by raising prices only provides a larger input supply. It does not immediately increase either sales or profits. When the downstream market is also oligopolistic, there is a further incentive to keep prices of inputs down because this will both increase the margins between input and output prices as well as potentially limiting the supply of inputs available at the prevailing, depressed market price. Reduced input availability in turn provides the oligopolist with increased capacity to raise price in the downstream market with the knowledge that its competitors are unlikely to bid up input prices (lowering their margins) in order to reduce resale prices (still further lowering the total margin). Thus, there is both a general incentive to exploit buyer power and less capacity for self-correction of such market distortions.
This obvious fact about the incentives of buyers has two implications for merger analysis. First, coordinated efforts to keep input prices down can encompass more potential participants without serious risk of opportunistic defection. The Second Circuit remarked on this fact when it upheld a complaint charging a number of major employers in the oil and gas industry of conspiring to suppress wage competition among potential employees.(5) Moreover, such efforts at price limitations would require very little coordination among buyers. The retention of low input prices being in the obvious self-interest of each buyer.
Second, even in the absence of overt or even tacit coordination, a buyer with some monopsony or oligopsony power will have an obvious incentive to drive down input prices and increase the price spread between input and output markets. Its competitors in the downstream markets will have little incentive to disrupt such unilateral conduct in many situations. To be sure, raising rivals costs (i.e., enter the rival's upstream market and raise the price of the input) is a plausible strategy, but the costs of bidding up a competitor's input prices must be balanced against the potential for retaliation and the mutual harm, relative to other producers, in terms of costs for the products being sold. Indeed, as discussed below, a better strategy, arguably unilateral, is to allocate input markets so that each major firm has its own set of suppliers as to which it can exercise buyer power. This strategy reduces input prices and avoids the kinds of competitive confrontations that would otherwise have to occur if a rival wanted to compete for the same sources of inputs. It is better long term strategy for a buyer to have its own set of semi-captive suppliers than to be in a market where it must compete with other buyers. The incentive to pursue such a strategy is particularly strong when the cost of shipping the input, e.g., livestock, is significant, and many areas of the country are amenable to the production of the input.
Where shipping costs are low and the input has multiple uses, the ability to exploit even a large position in some specific segment of the buying market is unlikely. I would expect this to be true in the case of many generic manufactured inputs. On the other hand, if longer term relational contracting is important, there is significant sunk investment in relatively specialized equipment, transportation is a significant cost factor, and buyers are themselves geographically dispersed, then manufacturers of inputs will look a lot like many contemporary agricultural producers.(6)
Thus, in many contexts a firm with buyer power will have the incentive to exploit that power without significant concerns for immediate market reaction. The most basic way such power can be exploited, as implicitly assumed in the preceding discussion, is to lower prices for inputs generally. The recent cases involving a conspiracy among blueberry buyers and tobacco buyers to hold down the prices paid for those crops are contemporary examples.(7) However, the dynamics of the buying side of the market show that it is often the case that other opportunities exist to exploit buyer power that seem to be attractive to firms so endowed. Two additional adverse competitive effects merit particular reference.
First, as a firm becomes a larger buyer in any specific product line, it has increased incentive to engage in manipulation of public prices for the input. The prices for many agricultural commodities, even when sold under contract, rely on prices set in the residual public trading market. For example, most hog contracts throughout the country use the prices from the northern Iowa, southern Minnesota market as the base for setting prices. Cheese and other diary products(8) rely on the prices from the Mercantile Exchange in Chicago for their base. Buyers (or sellers) of large quantities of such commodities have strong incentives to try to manipulate the "public" price in order to gain advantage in the great volume of goods that are traded off exchange. Such manipulative use of public market prices is an old story.(9) The recent cheese litigation is a current example of this problem.(10)
A crucial question is how much of a market share makes it worthwhile for a buyer to seek to influence the public market price? How easy is it for buyers to coordinate their actions given the public character of such markets which may both facilitate transparency among the actors, but may also reveal their actions to greater scrutiny? The risk posed by increased incentives to manipulate market prices is clear as the size of the buyers increases. Certainly, where the input is sold in a public market that handles only a small fraction of the overall sales but which is the source of prices for large numbers of off exchange transactions, it should be incumbent upon those reviewing a merger that will change appreciably the market shares of such a buyer to evaluate careful this risk.
The second risk that a merger among competing input buyers creates is for direct price discrimination among sellers or other non-price exploitation. A monopsonist has a great deal of discretion with whom it will deal and on what terms. This results in two kinds of problems. First, there can be, and is currently in livestock markets, substantial discrimination among producers with respect to the prices and other terms of trade they receive. In beef cattle for example, favored sellers get the advantage of contractual arrangements that assure such producers of prices at or above the cash market. Disfavored sellers are compelled to sell in the cash market at whatever prices the buyers offer. Such cash sellers have not necessarily voluntarily chosen to rely on the cash market. Rather it is the capacity of the buyer to refuse to offer contracts combined with the lack of a viable market in such contracts that creates this effect.(11) Reducing the number of competitors thus increases the potential for such discrimination on both price and access.(12)
In addition, monopoly buyers are often able to dictate terms and conditions that transfer risks to the producer without commensurate compensation.(13) The poultry industry is almost entirely dominated by contracts of adhesion many of which have exploitative terms resulting in uncompensated economic cost and risk falling on the producer. By reducing input costs, the processors are more able to cover their own processing costs when facing a downstream market in which the subsequent buyer also has buyer power. Such exploitation may be counter-productive in the longer run because it effectively consumes the capital invested in the production of the input, but for the intermediate processor to survive in the face of a powerful downstream customer that demands, on an all or nothing basis, lower prices, there may be a kind of economic compulsion. The downstream buyer recognizing that the its supplier has potential buyer power further upstream thus impels the exploitation of the powerless producers that have few, usually no, choices among buyers. These producers are very vulnerable because of the combination of sunk costs and very great switching costs.(14) Hence, there is a kind of rachetting effect in which the downstream pressures cause the processors to transfer more and more risk and cost to the producer while holding price of the input down.(15) Mergers that create increased buying power in the more distant, downstream market can, therefore, exacerbate these problems of upstream buyer power exploitation in both price and non-price terms.(16)
While some might regard such exploitative conduct as basically a matter of wealth transfer having no effect on market competition, it is important to appreciate that the long run incentives to participate in production markets are a function of the expected gains from the activity. In a dynamic analysis of incentives, it should be obvious that when others appropriate most of the wealth produced by an activity, the attractiveness of entry or innovation in that activity will be greatly reduced or eliminated. Thus, the greater appropriation of wealth created by farmers through their use of modern technology and efficient methods, the greater is the disincentive for the next generation to enter farming. Moreover, as the markets for agricultural products fail, the fundamental public interest in retaining a viable agricultural sector is likely to call forth further subsidies and other market distortions intended to prop up producers. The dynamic inter-connection of market practices with these longer run considerations provide another of the indirect effects of increased buyer power and consequent incentives to engage in conduct that has overall adverse implications for the competitive market even if it appears to be rational conduct for the firm.
As switching cost increases, the opportunity to exploit sellers grows. Sellers in agriculture. have substantial sunk cost in product specific markets and so can neither find other buyers for existing lines nor switch to another product in many cases. Professor Schwartz in his 1999 talk about buyer power highlighted this issue in the context of HMO competition for doctors.(17) In reading that paper, one should add farmers to doctors to appreciate that the issues of switching costs are more general than might appear at first.
A recurrent observation is that high volume retailers have significant leverage over their suppliers.(18) Because of the need to have access to a large percentage of outlets in order to obtain a sufficient volume of sales, producers are put at the mercy of each of their large volume customers.(19) In such a context, a firm with a 20% share of the national market in such a class of products is likely to have substantial power over its suppliers because of the threat that the supplier could loose one-fifth or more of its outlets. Reducing the number of outlets stocking a product line to 80% or less of prior outlets will inherently create a serious problem for a supplier. The central insight here is that for mass marketers to operate efficiently they need to have a very large scale presence in retail outlets. Small producers in such products are, in some respects, even worse off. They may require only a regional presence, but they are likely to be more dependent on the retailers that are willing to give them an outlet. In either case, the threat by a major retailer (relative to the total volume produced by the upstream supplier) to drop or curtail its selling effort creates a serious economic problem for the manufacturer. Unless it can switch products or outlets easily, such a producer faces a bargaining situation in which the downstream retailer has the dominant position. The use of slotting allowances, advertising support, and other payments by producers to retailers confirm the relative power.(20)
Despite Robinson-Patman, some of the power of large retailers is used to obtain better price terms on the goods being purchased. Thus, the large retailer gets, directly or indirectly, lower prices than other retailers even though there is no economic justification for this difference. Law enforcers should be skeptical of the social advantage that may accrue from such selective discounts. When not cost justified, they serve primarily to disadvantage classes of buyers who lack power and thus indirectly reinforce the market power of the dominant buyer.
More generally, one observes as the justification for mergers in retailing that the combined firm will achieve lower prices. Whenever the expected lower prices are to arise from lower input prices, there are grounds for concern that what is about to occur is the kind of racheting effect described previously wherein the retailer forces an intermediary to lower its price and that lower price is reflected back up the chain until it comes to rest with those lacking power to pass on the reduced price.
Another important implication of significant retailer buying power is that it creates a capacity to engage in exclusionary conduct. From Interstate Circuit and Klors to Toys R Us we have a number of illustrations in which downstream retailers are able to coerce upstream suppliers into adopting practices that excluded or imposed serious competitive disadvantages on their rivals.
The common theme of these cases is that the upstream producer needs to sell in volume and so needs access to a large number of outlets for its product. This in turn empowers the downstream retailer to make demands both about price and non-price exclusionary conditions. Such demands can be, as in Toys R Us, unilateral with respect to the source of power. The major toy producers required access to the 20% of the market represented by sales through the Toys R Us system. Such retailer demands can also require coordination among producers as in Klors or Interstate Circuit. In my view, such coordination is more likely to occur than outright price fixing or market allocation because each participating upstream firm has more incentive to adhere to the scheme because its own sales are at risk if it cheats.(21)
This phenomenon, which I have located in the retailing end of the distribution chain, may also exist in other contexts where either buyers or sellers require access to the capacity of a substantial fraction of those on the other side of the transaction.(22) What is important from the perspective of merger analysis is that when such a need for use or access exists, as concentration increases beyond a very competitive structure, the opportunity to exercise such power will arise. Again, the paradigm is Toys R Us which had a 20% share of the toy market, but compelled its major suppliers to engage in exclusionary refusals to deal with its more efficient competitors.
Market definition on the buying side requires careful consideration. First, buyers in moderately oligopsonistic markets often have the incentive and capacity to engage in price discrimination among their suppliers. The incentives are obvious and the capacity arises whenever some set of sellers has few options either because modest scale makes it impossible to search a larger market area, other limits on the options available to sellers either inherently or because buyers have imposed some form of allocation of sellers that precludes getting competing bids, or other kinds of significant switching costs.(23) Hence, just as the merger guidelines recognize in selling markets that the capacity to separate out groups of buyers and charge them a higher price reflects a relevant market for purposes of analyzing competitive effects, so too on the buying side, the analysis has to consider whether the same potential to discriminate among classes of sellers might exist. For example, I have been told that volume hog producers and those working at a lower volume have very comparable production costs per hog, but face different transportation costs. The large feeder can ship semi-trailer loads of hogs while the lower volume feeder will use a small trailer attached to a pickup. The differences in the per hog cost of transportation allow the bigger load to go further. In such a situation it is important to examine the options available to both classes of producers to ensure that a merger does not create an undue risk of discrimination against the class with more limited options. It is my sense that the Division may have made an error in its review of the recent Smithfield acquisition of Farmland because it did not fully examine the capacity to discriminate among hog producers based on their ability to transport hogs.
Second, buyers markets have greater potential to develop anticompetitive, linked oligopolies than is likely in seller markets.(24) Because of the shared interest in retaining the lowest input prices possible, buyers will find it attractive to avoid competition for inputs in so far as possible. Moreover, the risk of retaliation across markets when firms compete in the purchase of multiple inputs at multiple locations provides a context in which tacit understandings become more enforceable. Each firm will have an incentive to develop its own sources of supply and not engage in vigorous competition with its downstream rivals. In addition, it is observable in agricultural markets that processors tend to avoid building facilities that compete very directly. Each seeks to operate so that it competes only at the geographic margins of its supply territory. One form this interaction can take is for a buyer to buy and then close plants that are "too competitive" with those of other buyers. By doing so, the new owner reduces competitive pressure on its rival within the rival's sphere of influence and might expect that the rival would reciprocate by focusing its buying interests away from other areas of potential overlap. Again a central motivation for this ad hoc creation of spheres of influence is that each party can expect to gain to the extent that its buyer power is enhanced. Thus, unlike allocating selling markets, the buying side allocations have greater potential for self-reinforcement.
In standard merger analysis concerning seller power, the impact on the immediate downstream market is the primary focus. This is sensible because if the merger will result in higher prices to that market then the downstream effects can be predicted. While in general, the same can be said of buyer power, there is more ambiguity as to the explanation for lower input prices that might result from a merger. On their face, such "cost savings" appear to promise an efficiency gain. But the capacity of the upstream supplier to provide further price discounts to a consolidated buyer may reflect only the ability of that supplier to exert further monopsony power against its upstream input providers. Thus, the adverse effect of a consolidation in grocery retailing, as discussed previously, or the manufacture of food products may primarily impact farmers and ranchers who see the price for the basic ingredients driven down.(25) These impacts will present themselves in the form of lower prices in public markets and contracts for supplies. This price reduction is itself a result of the fact that the downstream buyer can induce its supplier to reduce price because of the creditable threat of moving the business to another supplier. Consolidation among such downstream buyers creates more leverage over their suppliers and so has the capacity to create a new downward pressure on input prices.
Another example of this phenomenon is the manipulation of cheese prices.(26) The direct goal of such manipulation was to lower the price of cheese for the advantage of major cheese buyers. By selling cheese on the old Green Bay cheese exchange, the major buyers of cheese could manipulate the price downward and retard its increase in many situations. This allowed them to buy cheese by contract based on the exchange price at lower rates than would otherwise exist. Milk converts to cheese on a relatively fixed ratio, and so when the price of their product declined, cheese makers reduced the price paid for milk. While the cheese makers may have absorbed a small amount of the price reduction, the great bulk of this buyer power was passed on to the dairy farmers. Obviously, further concentration on the cheese buying side of the market would increase the incentive and capacity to engage in such manipulation further threatening the income of dairy farmers.
The point is that the impact of anticompetitive conduct on the buyer side of the market is often at one or two stages prior to the party bearing the immediate impact of the lower price. A full analysis of the effects of a merger that increases buying power needs to trace out the potential for such indirect impacts if it is to provide a comprehensive evaluation of the potential competitive harms from a transaction.
Such pressures also need to be carefully distinguished from transformations in the technology of production or the methods of transaction that generate both real cost savings and may result in significant changes in demand for inputs. As discussed below, to obtain overall static allocative efficiency gains from mergers creating both some productive efficiency and increased buyer power requires that the productive efficiency gains outweigh the losses. But from a dynamic perspective, it is essential to consider the impact of wealth transfers as well. Hence, in most cases, the combined static and dynamic adverse effects on immediate or more distant parts of the supply chain will not be offset by any modest gain to productive efficiency.
The last point that needs reference is the tension between potential economies resulting from combinations increasing buying side concentration and the risks of adverse effects. It is a commonplace of merger analysis that efficiencies are easily to claim and hard to prove. In my own view this is in substantial part because very few efficiencies are unique to a particular firm specific market organization. For the great majority there are, in the words of Mao, "many roads" to efficiency. This is especially the case when each of the combining firms is already a multi-plant operation enjoying most or all of the economies of scale. In beef and pork, a rough estimate is that 8 to 10 firms each operating two or three plants could exist in this country and all those enterprises would be at or above minimum efficient scale.(27) In such contexts, there is unlikely to be any appreciable real production cost savings resulting from the combination of such firms. It is equally hard to imagine that combining such buyers would significantly reduce transaction costs of buying inputs. Hence, any significant reduction in costs from such a merger is likely to arise only from driving down the price of inputs. But absent evidence of seller cartelization or other market power on the selling side, there is no reason to believe that seller prices are excessive. Therefore, the most likely explanation for the putative "efficiency" gains is that they represent the potential for exploitation of monopsony power. It bears emphasis that a monopsonist has no incentive to share any of those gains with downstream customers unless, as discussed in the case of poultry, downstream buyer power drives the upstream market exploitation.
This is not to deny that there could not be some efficiency gain combined with the increased monopsony power that a merger generates. The questions are whether such a combination will advance static allocative efficiency and the long run need for dynamic efficiency in the market. The famous Williamson tradeoff argument suggests that there is a real possibility of a gain. There are those who are critical of Williamson's model, but more importantly, for our purposes, it focuses on the downstream markets in which goods are sold. Richard Sexton and Mingxia Zhang have examined the trade off between increased monopsony power and increased efficiency in production.(28) Their conclusions are that the balance tips strongly against a net allocative efficiency gain from such combinations if the firm has both buyer and seller power unless the productive efficiency gain is very large. Indeed, both consumers and producers are at risk of exploitation.(29) Hence, the use of combined market power is likely to overwhelm the incentives to lower price and increase output arising from some productive efficiencies. This again suggests the need for a more complete analysis of mergers where there are increases in power in both buying and selling markets. In combination the adverse effects can be much more significant than would appear if the investigation focused on only one side or the other.
If the firm has only buyer power, the Sexton-Zhang model shows that the static trade offs require less productive efficiency gain to offset the deadweight loss resulting from increased monopsony power. But the model also shows that there is a very substantial wealth transfer from sellers to such a buyer. In dynamic terms, such a transfer would create a strong disincentive to enter or innovate in the selling side of such a market. Thus the exploitation of buyer power in this way creates a long run adverse competitive effect
If these conclusions hold up on further consideration, they provide an important background reason for taking a stricter view of mergers on the supply side of the market than comparable mergers affecting only the selling side.
The foregoing factors do not yield a focused set of criteria for evaluating the buying side of mergers. However, they strongly suggest the need for a more naunced and informed evaluation of combinations that raise such concerns. The biggest question is what conditions create the necessary level of concern and what kinds of competitive effects should the considered. The prior discussion makes a case for new metrics to measure the structural thresholds and more flexible inquiry into the likely competitive effects of those transactions that cross the structural threshold.
While it is generally recognized that buying side markets do not correlate with the selling side, it is not entirely clear that antitrust enforcers have fully accepted and internalized that insight. Illustrative of this myopia is a recent acquisition in the turkey business that reduced the number of buyers in a part of the northern Midwest from three to two. The acquisition was not challenged. My speculation is that the downstream market for turkeys, most sold frozen and easily shipped, appeared reasonably competitive post merger and therefore no critical analysis of the upstream buying situation occurred. Yet the farmers in the turkey business in that region with one fewer bidder for their business experienced a substantial reduction in the price received for birds in comparison with the prices paid in other regions where sellers had more options.
The central point is that for many commodities especially basic agricultural commodities processing must occur relatively close to the place of production. This is both an issue of the costs of transportation (livestock need special handling, crops spoil) and of timing because many commodities once ready for market need prompt processing. Even in the case of grain, local markets are vital because of the necessity of combining very large quantities in order to obtain the necessary volumes for long distance transportation. When this fact is combined with the potential that lower volume producers may have fewer options among buyers, it should be apparent that the many buying markets ought to have narrow geographic definitions where directly competing buyers combine.(30)
On the other hand, where firms buy in adjacent markets, the market definition problem is more complex because the most likely impacts of such a combination is the loss of potential competition and the entrenchment of dominance in the combined region, but these theories are now disfavored in antitrust analysis. Hence, if indeed some sellers cross over the boundary, it is important to access the impact of the merger in an appropriately larger context of interlinked markets to see how seriously it affects sellers. In this, the analysis is similar to that done in selling markets where branded products that are imperfect substitutes are sold.(31)
A different metric also should inform the buying side analysis when the focus is on retailer buying power as in Toys R Us. The buying power that can inhere in a national retailer creates a real potential for imposing anticompetitive conditions on producers of retail goods. Hence, it is plausible to look at broad categories of goods, e.g., toys, bought in the national market as well as more regional market contexts. The central issue is producer access to a sufficient number of outlets to ensure that it can market its goods successfully. The appropriate scope of such access has to be measured in light of the classes of products and the options for their distribution to the final consumer. Toy manufacturers were the target of Toys R Us; all cheese makers and behind them all dairy farmers were the victims of the manipulation of the cheese market. In these cases, the significance of the buyer in the market, as discussed further below, is an important factor. Where, as in the toy market, a relatively major retailer has the capacity to affect very much the access of a producer to the ultimate consumer, substantial power inheres in such a buyer. Combinations in such a market creating an additional such buyer pose particular competitive concerns for both existing producers and those seeking to enter or expand in the product market.
Similarly, the risk of upstream price manipulation through a public market having limited volume increases with the overall volume of inputs whose price is set by that market that such a firm buys. The incentive to manipulate prices that affect an economically important category of the buyer's inputs will arise directly from being a large buyer of such products whether measured in market share or dollar value.
In sum, market definition has to go forward with an eye to the relevant aspects of the merger being evaluated for its impact on buying. Once, it is accepted that there are different buying and selling markets, it should not cause concern to recognize that the markets most relevant to the evaluation of specific competitive concerns will themselves vary.
An important change in merger analysis from the earlier eras is the inference to be drawn from market share analysis. When I did merger work many years ago, the structural change was the conclusive fact on which our analysis turned. Modern enforcement policy regards the structural change as an initial threshold which permits easy dismissal of combinations that fall below that threshold and requires more focused analysis of potential competitive dangers in those mergers that exceed the threshold. My central point is that in markets where buyer power should be a concern, the initial thresholds that trigger concerns should rest on experience with buyer power and its impacts rather than by analogy to the selling side of the market.
As a generalization, when buying side issues are being evaluated, antitrust enforcers should employ lower thresholds for both increases in concentration resulting from the merger and in post-merger concentration levels. This is a central lesson from auction theory.(32) Buyers have more incentive to collude because it is easier to share gains among themselves and, as discussed earlier, the nature of buying discourages defection from the collusive understanding. Moreover, tacit understandings and unilateral exploitation of buyer power are less likely to induce disruptive responses from downstream rivals. Furthermore, individual buyers who buy significant shares of a product either nationally (e.g. Toys R Us) or regionally (e.g., Klors, Interstate Circuit) have and can exercise significant unilateral buyer power over their suppliers. The exploitation of this power can lead either to price discrimination among buyers or exclusionary practices with respect to the competitors of such dominant buyers.
On the selling side, it appears that most merger cases have focused on challenging combinations that reduced major competitors from three to two or four to three. The incentives to engage in anticompetitive uses of buyer power, the improbability of efficiency gains that will offset the allocative detriments, and the incremental effects of the loss of a bidder in most selling markets, all argue for using a lower threshold. Combinations that reduce the number of substantial buyers to six or fewer in any market are likely to affect prices and consequently the dynamics of the supply market. It is for this reason, as well, that I question the decision of the Division not to conduct a focused investigation of Smithfield's acquisition of Farmland, given that these were two of six major competitors in the pivotal price making market area for hogs.
In retail markets, relatively modest shares of national buying are associated with significant buyer power. It is my suggestion that when a retailer buys 20% or more of the national market in some class of goods, that retailer has the potential to exercise substantial buying power.(33) It would follow that combinations creating firms with shares above that threshold poss a serious threat to the input market.(34) The same would follow for combinations among producers of goods that buy on a national basis similar shares. They have incentives both to manipulate public prices and to employ tactics that force down the prices of their own input suppliers.
In defining buying markets for raw materials or other basic inputs, it is important to focus on the potential for discrimination among producers. The ability to refuse to deal as well as transportation costs can differentiate some classes of market participants and so warrant narrowly drawn markets when input buyers propose to consolidate regardless of the state of the downstream product markets. The experience in pork and turkeys seem to me to confirm the fact that a more thoughtful and nuanced analysis of such markets is long over due.
Once defined, it is likely that the buying side concentration in many such markets will be very high under any standard and that a merger involving competing buyers is likely to produce a reduction from three or four to two or three buyers. Such markets face very serious monopsonistic risks if such combinations are allowed. It is important, however, to recognize that switching costs will be substantial even in markets with five or six buyers. Many sellers are likely to have long term relationships with only a few of those buyers taking a substantial proportion of their output.(35) Consolidation in such markets because of the switching costs are likely to create more buyer power than might otherwise be evident based on a raw count of potential buyers.
An even greater departure from conventional seller side orientation is necessary for appropriate competitive effects analysis on the buying side. Again, the shifts are not in the fundamental conceptions of harm but in where and how those risks will manifest themselves.
As discussed earlier, buyer power is often reflected back up the stream of inputs until it reaches a market in which the sellers are powerless to affect their input suppliers. Examples are the animal meal business where the consolidated buyer can drive down its prices to its suppliers who in turn will pass the reduced price on up the line until it results in lower prices for cattle.(36) Similar consequences occurred in the cheese business when cheese buyers drove down the price of cheese.(37) The result was a pass through of most of the reduced price to dairy farmers. Unless antitrust law enforcers pay attention to the ways in which initial buying power impacts are reflected through initial sellers to more remote suppliers, the full implications of a merger for competition may go unappreciated.
The prior discussion has highlighted the range of adverse competitive effects that can impact both input suppliers and competitors of the firm with buying power. Such a firm will have an incentive to manipulate any public market in which inputs are priced. This effect will occur whenever there is a public market or other price disclosure system that involves a limited volume, but which serves as the benchmark for a large number of contracts made outside that market. In this context, the likelihood of an adverse effect is a pretty direct function of the market share of the buyer.
A second effect from enlarged buying power is an incentive to engage in discrimination among its input suppliers when those inputs are relatively generic, as in livestock. The buyer uses its power to drive down the total cost of its inputs. This effect is contingent, I believe, on the suppliers having high switching costs resulting from product specific investments and lack of alternative markets into which they could sell.
Third, major buyers in national markets can both obtain discriminatorily favorable prices and impose other disadvantages on their competitors. Here the pivotal market characteristic appears to be the need for producers to have access to a large number of outlets through which their products will reach consumers. In such a context, a firm with a 20% share of the national market in such a class of products is likely to have substantial power over its suppliers because of the threat that the supplier could loose one-fifth or more of its outlets. Obviously, smaller suppliers have more focused sets of outlets, either regionally or nationally, will face even more coercive impact if two of its customers consolidate. This again assumes switching costs are substantial for such a producer.
The central lesson from auction theory that carries over into the more generalized arena of buying is that as buyers, firms have more incentive to collude and less incentive to betray the conspiracy. This makes overt collusion a more serious risk, but more importantly shows that tacit coordination, in particular market allocation, is a very likely outcome of increased buyer power.
It is because of the incentives to coordinate that lower market share thresholds are appropriate to the evaluation of mergers where a relatively small, six to eight, firm group are the primary buyers of some input. The incentive to reduce prices by collective action will necessarily exist. The important factual issue for investigation is whether other market factors make it unlikely that the parties will in fact act in a coordinated manner. Entry conditions and the capacity of suppliers to switch to producing other products or selling to other buyers at relatively low cost are likely to limit the capacity of even a limited number of firms to exploit their potential for coordination.
Depending on the characteristics of the input market, buyers may find it attractive to create geographic spheres of influence in which only one or two firms do the bulk of buying. This is a common feature of agricultural product markets. Moreover, when there are limited number of buyers, each buyer can establish a course of dealing with a few suppliers which again further allocates the supply market creating more power in the buyer to reduce price or impose other conditions on the semi-captive supplier. In other markets, the process of collective action may require the use of various strategies to disclose prices and avoid bidding them up. Investigations need to be attentive to the methods of price making and communication on the buying side of the market. While it is true buyers need some information, the scope and nature of the information can be a signal of tacit coordination. In any such situation combinations among such firms are very likely to exacerbate the problems.
The standard justification for mergers is that they will increase the efficiency of the resulting firm. It is by now a common place that such claims are often questionable. This is particularly the case in mature markets where mergers result in significant increases in buying power. It is prima facie unlikely that such markets are inefficient in the transactional systems currently in place or that the prices of inputs, given a workably competitive input market, are significantly out of line in relation to costs. Hence, the claim of substantial efficiencies on the input side most be regarded with great skepticism.
Indeed, when merging parties assert that they will be able to reduce input costs substantially, absent some major technological change that will result and is merger specific, the most likely explanation is that the merged entity expects to acquire buyer power and so become able to drive down input prices through the exercise of monopsony or oligopsony power. Indeed, we see a ratchet like process in which the merger of grocery chains that produced buyer power is met by mergers in the food processing business intended both to create off-setting seller power and increase upstream buyer power for the merged processor so that it can pass back the lower prices for its products to its upstream suppliers. Hence enforcers should be highly skeptical of claims of cost savings when linked to the buying side of the merged enterprise.
In addition, antitrust enforcers should be particularly concerned with Professor Sexton's observations that mergers increasing combined buyer and seller power, even if they make modest contributions to productive efficiency, will on balance result in substantial loss on a static allocative efficiency basis.(38) This insight means that efficiency claims are less meaningful to merger analysis where such combined power is involved except when the gains are clear, substantial and merger specific. Again as noted earlier, the implication of this analysis is that where mergers create both types of power, the analysis must consider the combined impact. Further, the dynamic implications of wealth transfers that will result from enhanced buyer power along should be a further cause of concern. Such transfers have, as discussed previously, clearly negative implications for the long run competitive efficiency of the selling side of such markets
Some times merging parties contend that the barriers to entry are such they can not exercise significant power in the market despite a large market share. While it would certainly be the case that if entry into the buying side of the market were easy, that would be likely to modify the capacity of firms to exploit apparent buyer power, it is essential to look closely at the realities of such entry claims. It is not sufficient to show that a plant can be built cheaply. Indeed, such a claim raises an important question as to why the merger is occurring at all. In most of the contexts I have described, the entrant will need a well developed channel to bring its goods to the ultimate consumer as well as the capacity to obtain inputs easily. In markets where established firms operate multiple plants and buy from suppliers who can serve efficiently only some of those locations, any entrant will face the potential that its established rivals will bid up the cost of inputs in its immediate area. This will raise the rival's cost of doing business and disable its competitive potential. Realizing this risk, investors, potential suppliers and customers are likely to be skeptical. The suggestion then is that the character of the supply market, especially its potential to be allocated among existing buyers, creates an additional potential barrier to entry beyond the basic costs of building a plant and developing a channel for distribution.
Antitrust law has long recognized that buyer power creates competitive concerns just as seller power does. However, for too long, the primary focus of antitrust analysis, especially in mergers, has been on the selling side. It is time to redress the balance. This session is a step in that direction.
Enforcers need to develop a deeper understanding of the unique characteristics of the buying side of the market place. This calls for appropriate metrics. A mindless transposition of seller side criteria for market shares or competitive effects can result in a deeply flawed analysis of the buyer power implications of mergers. In some cases, mergers that do not create serious buying power problems might be challenged but, far worse, mergers which create serious competitive risks in one or more buying markets may be ignored because of the failure to employ appropriate market definition and competitive effect analysis. It is my earnest hope that both the FTC and the Antitrust Division will make sustained efforts to understand the different aspects of buyer markets and buyer market power. Only with that kind of effort can merger enforcement continue to fulfill its assigned responsibility as a guardian of our competitive market system.
1. I have elsewhere begun to address this broader problem including the role of antitrust law as a component in the overall constituting of fair and efficient markets for agricultural products. See, Peter Carstensen, The Role of Law in Markets for Agricultural Products (paper presented at the annual meeting of the Law and Society Association, Pittsburgh, June 5-8, 2003).
In line with the long standing commitment of both the Antitrust Division and the Federal Trade Commission to inter-agency advocacy for efficient, competitive and open markets, it is my suggestion that both agencies should devote time and attention to encouraging the Department of Agriculture to implement more effective market facilitation within the scope of the existing legal regime and to urging Congress to adopt more inclusive statutory standards to achieve those goals for agriculture more generally.
2. I have always thought it significant that the economic basis for the Philadelphia Bank presumption that increased concentration in a market created a rebuttable presumption of adverse competitive effects rested on a modest amount of scholarship. Only after that decision and, I believe, in response to it, did economists begin to undertake the large scale study and testing of the relationship between concentration and competition. As one of the old school, I find it quite heartening that some of the most recent scholarship building on the work of my late colleague Leonard Weiss confirms that concentration and higher price are frequently correlated. See Leonard Weiss, Concentration and Price (1989); see also, Richard Schalenmsee, Inter-Industry Studies of Structure and Performance in 2 Handbook of Industrial Organization at 988 (R. Schmalensee, R Willig, eds. 1989); Jonathan Baker, Econometric Analysis in FTC v. Staples, 18 J. Pub. Pol'y & Marketing 11 (1999). My expectation is that a more active pursuit of monopsony and oligopsony will result in a similar flowering of economic analysis.
3. Robert Marshall, Michael Meuer, Bidder Collusion and Antitrust Law: Refining the Analysis of Price Fxising to Account for the Special Features of Auction Markets, 2-3 (2004, paper available from the authors).
4. Id. at 2 ("Bidder collusion is common because cheating is relatively difficult.").
5. Todd v. Exxon, 275 F3d 191 (2nd Cir. 2001).
6. Richard Rogers, Richard Sexton, Assessing the Importance of Oligopsony Power in Agricultural Markets, 76 Amer. J. Ag. Econ 1143 (1994)
7. See, Deloach v. Philip Morris, 2003 WL 23094907 (MDNC 2003) (buyer price fixing in tobacco settled with $200 million payment to tobacco farmers); reports on the blueberry case which is in state court in Maine can be found in the Bangor News on various dates including December 9, 2003.
8. The price of such dairy products directly determine the price of milk used for such manufacturing purposes as well as affecting the price for fluid milk.
9. See, Peter Carstensen, The Content of the Hollow Core of Antitrust: The Chicago Board of Trade Case and the Meaning of the "Rule of Reason" in Restraint of Trade Analysis, 15 Res. in Law and Econ. 1, 42-43.
10. The basic analysis of this manipulation is found in Willard Mueller, Bruce Marion, et at, Cheese Pricing: A Study of the National Cheese Exchange (report to the Wisconsin Department of Agriculture, Trade and Consumer Protection, 1996). The reported cases are Knevelbaard Dairies v. Kraft, 232 F3d 979 (9th Cir. 2000)(upholding law suit by dairy farmers under California indirect purchaser law challenging market manipulation as antitrust violation), and Servais v. Kraft, 246 Wis.2nd 920 (2001), aff'd by an equally divided court, 252 Wis.2nd 145 (2002), cert. denied 537 US 1047 (rejecting a state antitrust challenge to this same conduct based on state law filed rate doctrine).
11. One might contrast the unregulated nature of the market for livestock contracts with the carefully worked out procedures of the securities markets governing both the initial offering and subsequent resale of publicly traded securities. Public capital markets suffered from seller power to engage in arbitrary, manipulative and discriminatory conduct toward buyers. When that conduct was regulated by the 1933 Securities Act and the 1934 Securities Exchange Act, public capital markets have prospered and grown dramatically. Livestock markets present the reverse situation, buyer rather than seller power. In each case, part of the public policy response should be to regulate the uses of such power directly (see note 1, supra); but in addition maintaining as competitive a market structure on the side with power reduces the capacity to exploit that power as effectively because the powerless side of the market will have more options.
12. See, Rogers, Sexton, supra note 6; cf., Steve McCorriston, Why Should Imperfect Competition Matter to Agricultural Economists?, 29 Rev. of Ag. Econ 349 (2002).
13. See generally, Joseph Miller, Contracting in Agriculture: Potential Problems, 8 Drake J. of Ag. Law 57 (2003).
14. Rogers, Sexton, supra note 6, at 1143.
15. Solutions to important parts of the contract problem must come from other elements of agricultural market law. Antitrust is not well designed nor does it have the precedents to provide regulation over such conduct; but it can seek to reduce the risk of such conduct by its attention to the structure of markets.
16. Vertical integration by a monopsonist can also facilitate exploitation of latent buying power producing, under some conditions, lower input prices and higher consumer prices. Catherine de Fontenay, Joshua Gans, Can Vertical Integration by a Monopsonist Harm Consumer Welfare? (2003, available at www.mbs.edu/jgans).
17. Marius Schwartz, Speech, 5th Annual Health Care Antitrust Forum, Northwestern University School of Law, Chicago, Illinois, October 20, 1999.
18. See, Toys R Us v. FTC, 221 F3d 928 (7th Cir. 2000). Additional support comes from Europe. Roger Clarke, Stephen Davies, Paul Dobson, Michael Waterson, Buyer Power and Competition in European Food Retailing (2002).
19. One potential reason why the automobile industry has so long insisted on using many independent dealers rather than chains of retailers is the potential power chains would acquire to dictate to the manufacturer. Palamountain suggests a similar explanation for the historical experience of the gasoline industry which initially tried ownership integration into retailing but then moved to individually owned gas stations over which the producers still sought to exercise substantial control. Joseph Palamountain, The Politics of Distribution (1955).
20. I have read that brand name products are often not much more profitable to their producers than are generic products once all costs including advertising are factored in. But a brand name producer can, by successful promotion of its product lines, insulate itself, to some degree, from a retailer's refusal to deal. If consumers expect certain brands to be present, the retailer will ensure that they are in fact on the shelf. Thus, manufacturer brand advertising can serve as a method of creating offsetting power in producers.
21. For a more general discussion of this kind of restraint, see Robert Lande and Howard Marvel, The Three Types of Collusion: Fixing Prices, Rivals, and Rules, 2000 Wisc. L. Rev.941.
22. In electric power, the peak load phenomenon creates a similar potential under some pricing schemes. The owner of base capacity has an incentive to acquire and withhold mid-level power production so that peak prices go higher by requiring the market to call on even more expensive generation. This results in an increased profit for the base component of the company's system. What makes this strategic conduct rational is the fact that the market is taking the entire industry output as a unit and pricing all units at a single price.
23. See, Lynn Hunnicut, DeeVon Baily, Michelle Crook, Rigidity in Packer-Feedlot Relationships (paper provided to Senate Judiciary Committee hearing on Monopsony Issues in Agriculture, October 30, 2003).
24. Elinor Solomon, Bank Merger Policies and Problems: A Linkage Theory of Oligopoly, 2 J. Money Credit & Banking 323 (1970)
25. See, Einer Elhauge, Analysis of Proposed Nestle-Ralston Purina Merger (on behalf of the National Grange)(copy on file with Peter Carstensen).
26. Willard F. Mueller, Bruce Marion, et al, supra note 10.
27. Peter Carstensen, Concentration and the Destruction of Competition in Agricultural Markets: The Case for Change in Public Policy, 2000 Wis. L. Rev. 531, 537.
28. Richard Sexton, Mingxia Zhang, An Assessment of the Impact of Food Industry Market Power on U.S. Consumers, 17 Agribusiness 59 (2001); see also Rogers, Sexton, supra note 6.
29. Sexton, Zhang, supra note 28; see also, Fontenay, Gans, supra note 16.
30. Rogers, Sexton, supra note 6.
31. See, Roy Epstein, Daniel Rubinfeld, Merger Simulation: A Simplified Approach with New Applications, 69 Antitrust L. J. 883 (2002).
32. Marshall, Meuer, supra note 3.
33. There is no available data, that I am aware of, concerning the market position of, e.g., Broadway-Hale in its markets. Where national representation is important because of advertising or other economic imperatives, it is very likely that a local or regional buyer with a substantial market share, i.e., 20% to 25% of total demand in the market may well have comparable capacity to affect the seller's treatment of other buyers, especially those with lower volume. This is an example of an issue that demands careful investigation in the context of cases that present such risks. Only with careful, case specific investigation can enforcers learn whether and under what additional market facts the potential risks are significant.
34. This implies when the post merger HHI is at or above 1600, even modest increases are likely to result in significant buyer power.
35. Hunnicut, et al, supra note 23.
36. Elhauge, supra note 25.
37. Mueller, Marion, et al, supra note 10.
38. Sexton, Zhang, supra note 28.