This case comes before us on petition for review of a decision of the Federal Trade Commission holding Boise Cascade Corporation in violation of section 2(f) of the Robinson-Patman Act, 15 U.S.C. Sec. 13(f) (1982). The Commission determined that Boise's receipt of price discriminations, in the form of discounts on office products it purchased for resale to consumers, tended to cause competitive injury to dealers in the office products industry with whom Boise competes. In so concluding, the Commission relied upon the "inference" of competitive injury, articulated by the Supreme Court in FTC v. Morton Salt, 334 U.S. 37, 68 S. Ct. 822, 92 L. Ed. 1196 (1948), that arises when a substantial price discrimination exists over time. For the reasons to be set forth, we grant the petition for review.

* A

Before examining the facts in this case, we first pause briefly to describe the pertinent statutory framework. Section 2(f) of the Robinson-Patman Act makes it unlawful for any person "knowingly to induce or receive a discrimination in price which is prohibited [under the Robinson-Patman Act]."1 15 U.S.C. Sec. 13(f). Since the Act directly proscribes only a seller's activities, the liability of a buyer under section 2(f) depends on whether the seller discriminated in the buyer's favor in violation of section 2(a) of the Act. See, e.g., Great Atlantic & Pacific Tea Co. v. FTC, 440 U.S. 69, 76-77, 99 S. Ct. 925, 931, 59 L. Ed. 2d 153 (1979); Automatic Canteen Co. v. FTC, 346 U.S. 61, 70-71, 73 S. Ct. 1017, 1022-23, 97 L. Ed. 1454 (1953). Section 2(a) prohibits price discrimination between different purchasers "where the effect of such discrimination may be substantially to lessen competition or tend to create a monopoly in any line of commerce, or to injure, destroy, or prevent competition with any person who either grants or knowingly receives the benefit of such discrimination, or with customers of either of them."2 Id. Sec. 13(a).

To establish a prima facie case under section 2(f), the Commission must establish two things: first, that the buyer received a lower price than its competitors, and second, that the price discrimination caused, or reasonably might cause, competitive injury.

The statute also specifies two defenses to a seller's (and therefore to a buyer's derivative) liability. No violation exists where the discount reflects the lower cost to the seller of selling to the favored customer (the "cost-justification" defense), nor where the seller granted the discount in good faith to meet a competing seller's price (the "meeting competition" defense).3 Id. Sec. 13(a)-(b); see also Automatic Canteen, 346 U.S. 61, 73 S. Ct. 1017 (interpreting "cost-justification" defense); FTC v. Sun Oil Co., 371 U.S. 505, 83 S. Ct. 358, 9 L. Ed. 2d 466 (1963) (interpreting "meeting competition" defense). In addition to these statutory defenses, the Commission recognizes a defense, or more precisely, finds an absence of competitive injury, where the discounts are generally and practically available to competitors of the favored customer (the "practical availability" defense). See, e.g., FLM Collision Parts, Inc. v. Ford Motor Co., 543 F.2d 1019, 1025-26 (2d Cir.1976), cert. denied, 429 U.S. 1097, 97 S. Ct. 1116, 51 L. Ed. 2d 545 (1977).

On April 23, 1980, the Commission issued a complaint, over vigorous dissent, charging that Boise, as a "dual distributor" (both wholesaler and retailer) of office products, had violated section 2(f) of the Robinson-Patman Act. The complaint alleged that Boise's receipt of wholesale discounts from manufacturers on products that it then resold in a retail capacity to consumers constituted price discrimination for purposes of section 2(a). (No charge as to wholesale discounts for products that Boise thereafter sold in a wholesale capacity was pursued). The complaint alleged that Boise received these discounts knowingly and that the effect of Boise's receipt of such favorable pricing (and the resulting differential in prices) "has been or may be substantially to lessen, injure, destroy, or prevent competition" with dealers that were ineligible for the wholesale discounts. See Complaint, Boise Cascade Corp., FTC No. 9133 (Apr. 23, 1980), Joint Appendix ("J.A.") at 3-4.

In issuing the complaint, the Commission directed the Administrative Law Judge to receive evidence and to make findings of fact sufficient for disposition of the case under the competing standards of two FTC cases, namely Mueller Co., 60 F.T.C. 120 (1962), aff'd, 323 F.2d 44 (7th Cir.1963), cert. denied, 377 U.S. 923, 84 S. Ct. 1219 (1964), and an earlier FTC case that Mueller had overruled, Doubleday & Co., 52 F.T.C. 169 (1955). Doubleday, in brief, had articulated an additional defense under section 2(f) when the amount of the price disparity was reasonably related to the buyer's cost (as distinguished from the seller's cost savings) of providing additional marketing services. As we shall presently see in fuller detail, Mueller rejected entirely Doubleda 's rationale, but both cases nonetheless seemed to live on within FTC chambers as competing strains of Robinson-Patman thought.

At the outset of our examination of the office products industry, it is appropriate to sound a note of caution as the applicable law meets the rich web of facts. The voluminous record of this case eloquently attests to the fact that the manufacture and sale of office products have given rise to a complex, dynamic industry with a variety of players performing discrete but interrelated roles. Resolving the controversy at hand requires applying to this complex business environment a set of legal ground rules which are in a state of considerable uncertainty. We are therefore well advised to follow the lead of courts which, confronted with cases arising under Robinson-Patman, have found it necessary to map out in some detail the competitive terrain upon which the players do battle. See, e.g., FLM Collision Parts, 543 F.2d 1019; see also Edward J. Sweeney & Sons, Inc. v. Texaco, 637 F.2d 105 (3d Cir.1980), cert. denied 451 U.S. 911, 101 S. Ct. 1981, 68 L. Ed. 2d 300 (1981). We hasten to observe that this fact-intensive approach is dictated by the statute itself, which as we indicated at the outset calls for an inquiry into whether the effect of a price discrimination has been or "may be substantially to lessen ... injure, destroy or prevent competition." 15 U.S.C. Sec. 13(a).

* The office products industry involves the manufacture, distribution, and sale of stationery, office supplies and office furniture.4 ALJ Finding 6, J.A. at 46. A wide range of items falls within the broad compass of "office products," including, to name just a few, business forms, appointment books, portfolios, address books, file cabinets, desk stands, ring binder notebooks, work sheet pads, and paper punches. See generally id. 103-06, 157-59, 202-20, 252, 287-90, 330-40; J.A. at 66, 76, 83-85, 89-90, 95, 102-04.

The sale of office products has traditionally occurred at three levels: manufacture, wholesale, and retail.

Manufacturers. Manufacturers in this industry sell their goods to wholesalers, dealers, and, on occasion, directly to end-users. Id. 7, J.A. at 46. Although the record provides no precise numbers, compare id., J.A. at 46 with id. 65, J.A. at 59, office products are apparently produced by at least 1,000 manufacturers in the United States. Id. 7, J.A. at 46. These manufacturers produce and market a bewildering array of products. For instance, Shaeffer Eaton, a division of Textron Inc., one of the six manufacturers that the Commission selected in its effort to demonstrate that Boise had violated section 2(f), produces writing instruments, social stationery, record books, typewriter paper, and other sundry items. Transcript at 1059. These products are, of course, manufactured in a wide variety of colors and sizes, creating hundreds of separate Shaeffer Eaton items. This proliferation of products and product lines is apparently not at all unusual in the industry.

Wholesalers. Wholesalers are defined as firms or individuals that buy goods for resale to dealers. Id. 23, J.A. at 50. There are approximately 100 wholesalers in the industry. Most are small, privately-owned businesses. Id. 27, J.A. at 51; Transcript at 6587-88. Sales at the wholesale level are dominated by five nationwide companies (the largest of which is Boise). These "Big Five" firms account for about 60% of the total volume of wholesale sales. ALJ Findings 26, 28, J.A. at 50-51. Boise, as we shall presently see, fits within the category of "dual distributors," that is to say an integrated wholesaler that carries on retail as well as wholesale functions.

Dealers. Dealers, by definition, purchase office products from manufacturers or wholesalers (or both) for resale to consumers. Id. 43, J.A. at 54. Within the broad genus of dealers (also referred to as "retailers"), one finds a number of species. Take, for example, the "contract stationer," a title generally used to denote large dealers (but sometimes employed to refer to any dealer that has on-going contracts with industrial customers). Id. 44, J.A. at 54. There is also (1) the "rack jobber," a specialized breed that maintains the stock of one line of products on a lease-type basis with retail stores; (2) the more familiar "mass marketer," which is a food or variety chain store, such as K-Mart or Walmart; (3) the "mail order house," which advertises and sells to consumers through the mail; and (4) the "converter," which buys component parts, assembles them, and markets them under the manufacturer's or its own label. See, e.g., Transcript at 1209, 1292-95. In addition, formal and informal dealer "buying groups"--coalitions of dealers formed in order to garner advantageous discounts--exist in most areas of the country. ALJ Finding 455, J.A. at 129. There are approximately 8,000 dealers in the industry. Id. 47, J.A. at 55.

Dual Distributors. In addition to these three major categories of participants, there is another player on the office products field, as we alluded to above. The "dual distributor" purchases products for resale to both dealers and end-users. Id. 5, J.A. at 46. Although the record does not specify how many wholesalers engage in dual distribution, Boise contends that most do. Boise's Brief at 5; Transcript at 5090. Of the "Big Five" wholesalers, however, only Boise (and possibly one other) is a dual distributor. Commission's Brief at 5. Due to the understandable reluctance of dealers to buy from their competitors, dealers rarely move into the class of dual distributors. See Transcript at 3597; see also ALJ Findings 63, JA. at 58.B

In the office products industry, price is typically expressed in the form of a "manufacturer's suggested list price," less a discount, if any. To illustrate by way of a simple example, suppose the manufacturer's suggested list price for a product is $1.00. If the manufacturer offers a buyer a 50% discount, the list price is reduced by 50% and the product is thus sold to that buyer for 50 cents. If the manufacturer offers a 50-20% discount, the list price is reduced first by 50%. A 20% discount is then applied to that amount (50 cents), thus leaving 40 cents as the net price to the buyer. Id. 8, J.A. at 46.

Manufacturers' discounts take many forms.5 Perhaps best-known are quantity and volume discounts. To state the obvious, quantity discounts are lower per unit prices charged for large orders of a product or group of products, as opposed to smaller orders. Id. 9-10, J.A. at 47. Volume discounts, on the other hand, are tied to the aggregate annual amount purchased from the manufacturer by a particular buyer. Id. 10, J.A. at 47. In addition, promotional discounts, which range from additional year-end discounts to extension of payment deadlines, are prevalent in the industry. Id. 20-22, J.A. at 49-50.

To a more limited extent, manufacturers also offer bid discounts. These may arise when large end-users--typically government entities--that need substantial quantities of office products solicit bids from marketing intermediaries (wholesalers and retailers) and seek prices different from those normally offered to end-users. Id. 15-19, J.A. at 49. Manufacturers offer the same "bid" discounts to all distributors submitting bids to these end-users, regardless of their identity as wholesalers or dealers.

Most significant for purposes of this case, however, are functional and trade discounts. A trade discount is one given to purchasers on the basis of the level of trade at which they operate. The discount depends solely on to whom the purchaser resells; it is entirely independent of marketing functions performed by the purchaser. Id. 517, J.A. at 141. A functional discount, in contrast, is offered by a manufacturer to a purchaser for assuming and performing a function that would otherwise be performed by the manufacturer. Id. 518, J.A. at 141. The pure functional discount operates independently of the purchaser's level of trade. In other words, any purchaser that performs the required functions would be eligible for the discount regardless of whether it is nominally a wholesaler or retailer.

At issue in this case is the validity of a hybrid discount which partakes of both a trade and functional discount. This discount, which for purposes of clarity and consistency we will call a "wholesale discount," has existed, without change, for many years in the office products industry. Indeed, the wholesale discount has been extended by most manufacturers of office supplies since well before Boise entered the industry.6 Id. 489, J.A. at 136. Its purpose is to induce wholesalers to undertake distribution and marketing functions on behalf of manufacturers. Id. 447, 460, J.A. at 127, 130. Manufacturers therefore extend the discount to any company, regardless of size, that meets the objective criteria contained in their respective definitions of "wholesaler."

In defining "wholesaler," all manufacturers in this industry require that the candidate resell products to dealers. Significantly, however, manufacturers do not require that the candidate resell exclusively to dealers (or other non-end-users). Nor do manufacturers limit the availability of the discount to those goods that are resold to dealers. Instead, manufacturers typically consider membership in the Wholesale Stationers Association (which, as its name suggests, is a trade association for wholesalers of office products) determinative of eligibility to receive the wholesale discount. Transcript at 1219, 1280-82, 1691-92, 2334. Under the Association's guidelines, any company that resells at least 20% of its office products to dealers qualifies as a wholesaler. Commission's Exhibit 2002D. As a result of this generous definition, most "dual distributors" clear the first definitional hurdle for wholesale discount eligibility, whereas those dealers which sell exclusively to end-users naturally do not. ALJ Findings 449, 454, J.A. at 128, 129.

Beyond the universal requirement that a buyer resell (at least 20%) to dealers, manufacturers differ in their respective criteria for granting wholesale discounts. To illustrate with the Commission's six selected manufacturers in this case, both Rediform Office Products, a division of Moore Business Forms, Inc., and Sheaffer Eaton define a wholesaler as a purchaser which not only sells to dealers but also warehouses and promotes their products, publishes a catalog, and employs and trains salespeople who call on dealers. Id. at 447, 450, J.A. at 127, 128. Kardex Systems, Inc. defines a wholesaler as a purchaser which maintains inventory on a variety of product lines for resale to dealers, publishes a price list, may (but apparently does not have to) publish a catalog, and employs outside salespeople. Id. 449, J.A. at 128. In contrast, the sole criterion imposed by both Boorum & Pease Company and Master Products Manufacturing Co. is that the purchaser resell to dealers and contract stationers. Id. at 451, 453, J.A. at 128, 129. Finally, Bates Manufacturing Company requires that the purchaser resell to dealers, carry an inventory of Bates' products, and publish a catalog. Id. 452, J.A. at 128-29.

Thus, wholesale discounts are trade discounts insofar as they are awarded on the basis of the identity of those to whom the marketing intermediary resells. To the extent that these manufacturers require the recipient to perform functions (such as publication of a catalog and warehousing) in order to qualify for the discount, the wholesale discount partakes of a functional discount. See generally Transcript at 6838-71.

With this overview of the office products industry and its pricing structure in mind, we turn to the specifics of the case before us.

* Boise is an integrated forest products company. It entered the office products industry in 1964 as a dual distributor through the acquisition of Associated Stationers Company. ALJ Finding 3, J.A. at 45. From this beginning, Boise's Office Products Division quickly grew, to the point where today its combined wholesale and retail sales make it the largest distributor of office products in the United States. Id. 5, 52, J.A. at 46, 56. From 1976 to 1980, Boise's sales in its wholesale capacity amounted to slightly over half its total sales. Id. 54-58, J.A. at 56-57. Excluding the Division's retail sales, Boise is one of the two largest wholesalers in the United States. Id 53, J.A. at 56. Notwithstanding its considerable size, Boise's purchases typically account for less than 5% of any single manufacturer's total sales. See Transcript at 6087.

Headquartered in Itasca, Illinois, Boise's Office Products Division operates through distribution centers located in 27 cities. ALJ Finding 2, J.A. at 45. Each distribution center is responsible for marketing, inventory management, purchasing, warehousing, sales, and customer service. Id. 4, J.A. at 46. Accordingly, the purchasing department at each of the 27 locations places its orders directly with manufacturers, which in turn ship the goods directly to and bill the individual distribution center. Id. 67, 463, J.A. at 59, 131. Likewise, each distribution center employs its own sales force to call on Boise's dealer and commercial (end-user) accounts.7 Id. 76, J.A. at 60a.

Although in its capacity as a dealer Boise performs essentially the same marketing functions for manufacturers as do other dealers, Boise is, by virtue of being a dual distributor, able to obtain wholesale discounts unavailable to the 23 dealers selected by the Commission for review in this case. Id. 95, J.A. at 64. These discounts are received on goods of like grade and quality that Boise sells in competition with those dealers. Id. 95-100, J.A. at 64-66. As we previously indicated, see supra text at 8-9, all six manufacturers studied in this case extend to companies (including Boise) that qualify as wholesalers, wholesale discounts that represent reductions in list prices beyond those offered to dealers.8 As a result of the difference between dealer and wholesaler discounts, the 23 selected dealers pay between 5% and 33% more for their purchases from manufacturers than does Boise.9

At the same time, Boise played no role in the adoption or formulation of the manufacturers' standards for offering wholesale discounts. Transcript at 1660-61, 2004, 2345. And, although the Commission points to evidence of Boise's successful negotiation for the discounts, Commission's Brief at 20 n. 19, the Commission candidly admits that the ALJ did not find that Boise had in any way coerced or pressured the manufacturers into extending it the wholesale discounts. Id. at 16-17.

The office products industry is characterized by intense competition. Id. 408, J.A. at 117. The industry, moreover, has flourished during the period in question. From 1976 to 1982, there was a steady rise in the number of dealers in the industry, both nationwide and in the individual States where the FTC's selected dealers are located. Id. 47, J.A. at 55. Boise's economic expert, Dr. Kenneth Elzinga, see supra note 6, interpreted this upward trend as indicating an absence of anti-competitive forces operating in the market. Transcript at 6103-14.

More relevantly for Robinson-Patman purposes, an analysis of the financial condition of 18 of the selected dealers revealed that all 18 had enjoyed an increase in sales during the period in question. Phoenix-based Wist Supply & Equipment Co., for instance, increased total sales from $1.6 million in 1977 to $4 million in 1980. Boise's Exhibits 300, 804. Yorkship Business Supply of Cherry Hill, New Jersey similarly increased sales from $1.3 million in 1977 to $3.5 million in 1980. Id. Pomerantz & Company in Philadelphia increased sales from $14.6 million in 1977 to $36.3 million in 1980. Id. Overall, the average annual growth and increase in gross profits of the dealers from 1977 to 1980 was in excess of 22%. Id. 431, J.A. at 121-22. What is more, this substantial growth occurred during a period of recession in the economy. ALJ Finding 431-32, J.A. at 121-23.

Along with their growth, the selected dealers' credit ratings with Dun & Bradstreet were, on the whole, quite favorable. Id. 432, J.A. at 122-23. And those companies that experienced a change in credit rating from 1977 to 1980 generally improved, again in the teeth of a recessionary economy. Id., J.A. at 122-23.

Data compiled by the National Office Products Association, the major industry trade association, revealed that in each year from 1967 to 1980, dealers' median net profit before taxes as a percent of sales ranged from 3.0% to 3.9% (except in 1974, when the median net profit was a heftier 4.3%). Id. 423-29, J.A. at 120-21. According to Mr. Ronald Rowe, the Commission's accounting expert, the median net profit for the selected dealers ranged from 2.3% to 3.5% for the years 1976 to 1981. Id. 430, J.A. at 121. From 1976 to 1979, Boise's combined wholesale and retail median net profit before taxes as a percent of sales ranged from 3.1% to 5.6%. Id. 87, J.A. at 62. This return on sales, as Boise's 1979-83 Business Plan explained, was comparable to the leading industrial stationers in the industry. Id., J.A. at 62.

In addition to this undisputed evidence of competitive health of the selected dealers and in the industry as a whole, there is another aspect of this case that is relevant to resolution of this controversy and unusual in the context of a Robinson-Patman proceeding--a virtually complete absence of sales lost to Boise by the selected dealers traceable to the price differential caused by wholesale discounts to Boise. Although the Commission aggregated 23 dealers, each of which carries hundreds, and in some cases thousands, of accounts, the record shows only 162 specific accounts identified by the dealers as lost to Boise. Commission's Brief at 33. This stability of dealer accounts is of especial note in an industry where switching of accounts is common. Id. 422, J.A. at 120. Significantly, none of the selected dealers that lost accounts in whole or in part to Boise was able to conclude that the losses were due to the fact that Boise received a greater discount from the six manufacturers. Id. 407, J.A. at 117.

To be more specific, the Commission's selected dealers testified that they lost accounts to Boise, in whole or in part, due to Boise's lower prices, better service, or a combination of these factors. Id. 384, J.A. at 112-13. For example, Victoria Station, a restaurant chain headquartered near San Francisco, switched from one dealer, Gilbert-Clarke, to Boise because of the latter's pricing and services. Id. 390, J.A. at 114. The City of Seattle switched the majority of its business from the incumbent dealer to Boise because of Boise's sophisticated invoicing system, depth of inventory, high fill rates, and usage reports. Id. 393, J.A. at 114. G.E. Stimpson, another dealer, lost the account of one valued customer to Boise's lower prices. Id. 396, J.A. at 115. New England Telephone switched from Union Office Supply to Boise because the latter had a computerized backup system that Union did not possess at the time. Id. 398, J.A. at 115. See also id. 385-406, J.A. at 113-17.

In addition to the small number of lost accounts, the record shows many instances where accounts shifted not because of price or service, but because sales representatives shifted. See, e.g., Transcript at 3855-56, 3276-78, 2498-99, 2527, 2544-45, 2953, 2041, 2071, 3467, 2444-45, 3170-74, 3188, 3402-03. The ALJ quoted Dr. Elzinga's testimony in this respect:

If you were trying to develop a theory of account shifting in the office products industry, probably the most robust theory you could develop is that accounts shift when salespeople shift. Over and again, as I am reading testimony about account shifting, I find that the real reason the account shifted is because a salesperson left. That is, Boise hires a salesperson away from Yorkship, and they get some accounts from Yorkship. Although in that case, if you read on, you find later Yorkship gained some of those back.

.... It is so common in this industry I think they have a term for it. It is called following; the salespeople have a following. The thing that struck me is the amount of evidence that seemed unambiguous of an account shifting that was explicable not by lower prices, not by better service, but by the simple pristine fact that the salesperson shifted.

ALJ Finding 434, J.A. at 123a-124 (quoting Transcript at 6118-19) (footnote omitted).

Further diluting the Commission's attempted showing of diverted sales, Boise introduced considerable evidence of sales that it had lost to the selected dealers. The ALJ, moreover, made specific findings to this effect. For instance, Boise's Philadelphia distribution center lost four large accounts in 1982 to dealers that offered lower prices. Id. 418, J.A. at 119. Likewise, Boise's Boston distribution center lost several accounts to Monroe, one of the selected dealers, because Monroe offered significantly lower prices. Id. 420, J.A. at 120. Seven accounts switched from Boise's Phoenix distribution center to Wist due to Wist's lower prices. Id. 421, J.A. at 120.

In addition to evidence of accounts actually lost, Boise identified many instances where lower prices were offered by the selected dealers and their competitors. Id. 409, J.A. at 117. One of Boise's Washington, D.C. salespersons, for example, testified that a dealer's price list obtained from one of Boise's customers revealed that the dealer was offering prices on average 30% lower than those offered by Boise. Id. 412, J.A. at 118. In the same vein, a sales representative from Boise's Salt Lake City distribution center testified that although he had submitted pricing to two school districts which was only 4% to 5% over Boise's costs, Boise received only 10% to 20% of the districts' business. Id. 414, J.A. at 119. See also id. 410-21, J.A. at 118-20.

The ALJ quoted Boise's expert's analysis and conclusions in this regard:

The evidence that I have studied and reflected on with regard to lost accounts persuades me that you simply cannot look at that evidence and conclude that injury to competition has occurred or is occurring here.

In fact the thing that really strikes me--and here again I speak as an economist--is you look at these [selected] dealers who ... have literally hundreds and hundreds and in some cases thousands of accounts, and when they are asked to give illustrations and to document the accounts they have lost to Boise, they are able to come up with a handful at best. I was really struck at the tiny number, in fact.

In fact, it was in a way even troubling to me as an economist that there were not more accounts being diverted around just through normal market processes. I would have expected much more just through almost random competitive shocks.

... [T]he striking thing to me is how few accounts were lost, and then when you start to get into the record as to the precise facts about those purportedly lost accounts, I find over and again that it is not clear to me they were lost to Boise, much less were they lost to Boise because Boise buys products at lower prices from manufacturers.

Id. 433, J.A. at 123-23a (quoting Transcript at 6119-22).

Notwithstanding his findings (1) that the selected dealers were competitively healthy and (2) that "accounts lost to Boise were counter-balanced by accounts which Boise lost to the dealers," the ALJ concluded that that evidence went only to the price discrimination's lack of effect on market structure. J.A. at 153-54. By virtue of that characterization, the ALJ determined that market-structure evidence was by its nature inadequate to rebut the inference of competitive injury raised by the sustained and substantial price discrimination that Boise received. J.A. at 154-55; see FTC v. Morton Salt Co., 334 U.S. at 46-47, 68 S.Ct. at 828-29. The ALJ stated his position in the following way: "[I]t is inconceivable that the substantial and sustained price differences documented in this record can have had no substantial effect on the ability of the dealers to compete with Boise." J.A. at 155.

Having determined that a prima facie violation of section 2(f) existed, the ALJ went on to conclude that the discounts were not practically available to the selected dealers, J.A. at 155-57, and that Boise knew, or should have known, that the discriminations in price were neither cost-justified nor given in good faith to meet competition, J.A. at 161-70. In addition, the ALJ recommended that the Commission adhere to its decision in Mueller. J.A. at 159-61. On the possibility that the Commission would reembrace the Doubleday rule, the ALJ determined that, because Boise did not perform services not performed by the selected dealers, Boise did not qualify for that "defense." J.A. at 161.

The Commission heard oral argument on the case in June 1984, and in February 1986 issued its order and opinion adopting and affirming the ALJ's findings of fact and conclusions of law. Boise Cascade Corp., FTC No. 9133 (Feb. 11, 1986), J.A. at 174 [hereinafter Commission Decision ]. In its opinion, the Commission expressly relied on Morton Salt's inference that substantial price discrimination between competing purchasers over time causes, or tends to cause, competitive injury. Id. at 187. In doing so, the Commission expressly recognized that the inference " 'may be overcome by evidence breaking the causal connection between a price differential and lost sales or profits.' " Id. (quoting Falls City Industries, Inc. v. Vanco Beverage, Inc., 460 U.S. 428, 435, 103 S. Ct. 1282, 1289, 75 L. Ed. 2d 174 (1983)). Specifically, the Commission suggested that the Morton Salt inference could be overcome by a showing that market conditions unrelated to the price discrimination explained the lost accounts or shifted sales or the effects on the disfavored competitors (presumably lost profits or market share). Commission Decision, J.A. at 191.

The Commission nonetheless determined that Boise's showing of dealer-specific and industry-wide competitive health, in conjunction with the relative absence of lost sales, "fails to rebut the 'self-evident' inference of causation." Id. at 191-92. Indeed, with virtually no elaboration, the Commission concluded that Boise's elaborate evidentiary showing did "not address the causal connection at all." Id. at 191. The FTC emphasized that under Robinson-Patman actual injury to competition need not be shown, but only " 'a reasonable possibility that a price difference may harm competition.' " Id. at 192 (quoting Falls City, 460 U.S. at 434-35, 103 S.Ct. at 1288).

Having found the two essential elements to a prima facie case (price discrimination and resulting competitive injury), the Commission concluded that Boise did not satisfy any of the defenses, statutory or Commission-fashioned, to Robinson-Patman liability. Id. at 204-18. In addition, for reasons that we will detail shortly, the Commission decided that Doubleday should not be revitalized and, accordingly, reaffirmed the anti-Doubleday holding of Mueller. Id. at 193-204.10

The Commission's decision confronts us with a difficult issue arising under a statute that does not represent the highwater mark of skillful draftsmanship. As Justice Frankfurter well put it in speaking for the Court a generation ago in Automatic Canteen Co. v. FTC, 346 U.S. at 65, 73 S.Ct. at 1020, "precision of expression is not an outstanding characteristic of the Robinson-Patman Act." More recently, Justice Powell, joined by Justices Brennan, Marshall and Blackmun, decried judicial interpretations and applications that "increase[ ] the uncertainty inherent in the generalities of the Robinson-Patman Act." J. Truett Payne Co. v. Chrysler Motors Corp., 451 U.S. 557, 570, 101 S. Ct. 1923, 1931, 68 L. Ed. 2d 442 (1981) (dissenting opinion).

The need for clarity, obviously a desideratum for any body of law, has been evidenced by judicial recognition that Robinson-Patman is not to be viewed as an act of Congressional schizophrenia, an anti-competitive island situated in an otherwise turbulent sea of pro-competitive efficiency and maximization of consumer welfare, the hallmarks of the Nation's antitrust laws. The Supreme Court warned early on of the dangers of doctrinaire interpretations of Robinson-Patman that could lead to "conflict with the purposes of other antitrust legislation." Automatic Canteen, 346 U.S. at 63, 73 S.Ct. at 1019. That warning has been repeated in more recent times. See also United States v. United States Gypsum Co., 438 U.S. 422, 450-51, 98 S. Ct. 2864, 2880-81, 57 L. Ed. 2d 854 (1978); Great A & P Tea Co., 440 U.S. at 80, 99 S.Ct. at 933.

The imprecision infecting the statutory language has frequently led courts construing the measure to repair to the backdrop against which the Robinson-Patman amendments were crafted in 1937. For it is in its genesis that the purposes animating Congress in passing this ambiguous statute can best be discerned and then borne carefully in mind in contemporary judicial applications.

The Supreme Court has summarized the statute's background in the following way:

The Robinson-Patman Act was passed in response to the problem perceived in the increased market power and coercive practices of chainstores and other big buyers that threatened the existence of small independent retailers.

Great A & P Tea Co., 440 U.S. at 75-76, 99 S.Ct. at 930-31. Robinson-Patman was thus aimed at protecting small retail businesses against favoritism toward their larger competitors, a threat seen as eventuating from an exception carved out by the original Clayton Act, passed in 1914, permitting discriminations in price based on "differences in the grade, quality or quantity of the commodity sold." Clayton Act, ch. 323, Sec. 2, 38 Stat. 730 (1914), 15 U.S.C. Sec. 13 (1982) (emphasis added). Robinson-Patman, in short, sought to remove the competitive advantage conferred solely by virtue of the size of the buyer's appetite. "The legislative history of the Robinson-Patman Act makes it abundantly clear that Congress considered it to be an evil that a large buyer could secure a competitive advantage over a small buyer solely because of the large buyer's quantity purchasing ability." Morton Salt, 334 U.S. at 43, 68 S.Ct. at 826. See also Falls City, 460 U.S. at 436, 103 S.Ct. at 1289; F. Rowe, Price Discrimination Under the Robinson-Patman Act 3-23 (1962) [hereinafter "Rowe"].11

The Supreme Court addressed early on the operation of Robinson-Patman in the paradigm setting envisioned by Congress--a quantity discount to large buyers. In Morton Salt, the Court was confronted with a salt manufacturer's discount pricing system under which only the five largest retail grocery chains in the country were able to buy enough salt to qualify for the manufacturer's most lucrative discount. The Court found that competitive injury, a prerequisite to liability under section 2(a) of Robinson-Patman,12 was "self-evident" simply by virtue of the longstanding and substantial price disparity:

Here the Commission found what would appear to be obvious, that the competitive opportunities of certain merchants were injured when they had to pay [Morton Salt] substantially more for their goods than their competitors had to pay.

Morton Salt, 334 U.S. at 46-47, 68 S.Ct. at 828.13

In recent years, the Supreme Court has reaffirmed the viability of the Morton Salt inference of competitive injury arising from a substantial price difference existing over time. It has, moreover, declined to cabin the Morton Salt inference to "cases involving 'large buyer preference or seller predation.' " Falls City, 460 U.S. at 436, 103 S.Ct. at 1289. See also J. Truett Payne Co., 451 U.S. at 561, 101 S.Ct. at 1926. To the contrary, the Falls City Court upheld application of Morton Salt's rule to a discriminatory pricing system based upon the geographical location of the buyer. (In that case, Indiana distributors of Falls City beer were charged higher prices than Kentucky beer distributors.)

The Morton Salt inference is thus alive and well in the law. The case before us, however, presents the Commission's modern-day invocation of that inference in the setting of a pricing system which is entirely different from quantity and regional discounts.14 For as we have seen, the Morton Salt inference has been applied in the case at hand to a system of wholesale discounts. (The Commission uses the term "functional discount" generically to describe the wholesale discounts at issue here. We will therefore proceed to use these terms interchangeably. Cf. supra text at 1132.) Acting Chairman Calvani, in writing for the Commission, succinctly described the legislative landscape as to wholesale functional discounts: "The Act does not expressly address functional discounts, and the legislative history is inconclusive." J.A. at 194. Thus, the imprecise statutory language of which the Supreme Court has complained, see supra text at 1138, beclouds the area of functional discounts as well.

The rather chaotic state of the law as to functional discounts was aptly summarized by then-Professor Calvani who, as fate would have it, was destined in another capacity to author the decision now before us:

[T]he uncertain status of functional discounting is primarily due to the failure of Congress, the Federal Trade Commission, and the courts to give explicit and independent recognition to the practice and to define with any modicum of specificity its permissible contours. The result of this failure of recognition has been a lack of focus upon the validity of the functional discount which, in turn, has left the law in a state of confusion, causing often legitimate practices to be condemned.

Calvani, Functional Discounts Under the Robinson-Patman Act, 17 Boston C. Indus. & Com.L.Rev. 543, 543-44 (1976).

Eloquent attestation as to the unsettled nature of the law of functional discounts is found in events surrounding the complaint that brought this case to life. In considering whether to issue a complaint, the Commission divided three to two in favor of sending the staff forward, with Commissioner Clanton and now-Dean Pitofsky in dissent. Dean Pitofsky set forth his views in a thorough opinion, which opened with the following warning:

The Commission today has issued an extremely unwise Robinson-Patman complaint. I would not ordinarily dissent from the issuance of a complaint (and certainly not at such length), but this one has such a profound anticompetitive potential that it ought not to go by without comment.

Complaint, Boise Cascade Corp. (dissenting statement of Comm'r Pitofsky), J.A. at 12.

Further evidencing the divided state of the Trade Commission's house was its order directing the administrative law judge to receive evidence and enter findings sufficient to dispose of the case under two warring doctrines articulated in the corpus of FTC law. As we saw above, see supra text at 1130, the earlier doctrine was the Doubleday rule, first clearly enunciated in Doubleday & Co., 52 F.T.C. 169, involving a publishing company's disparately favorable discounts bestowed upon three large jobber-distributors. To remind the patient reader, the Doubleday rule briefly stated is as follows: no injury to competition is occasioned by a buyer's receipt of a discount from a seller-supplier if that discount simply compensates the buyer for additional useful services it performs for the seller-supplier's benefit.

Doubleday recognized that the world of commerce, like the rest of life, is not simple. As complex distribution systems developed, the lines along the distribution chain became blurred. No longer was there a straightforward, three-tiered structure of manufacturer-wholesaler-retailer. In Doubleday, then-Chairman Howrey commented on this development as follows:

[M]ore complex types of distributors ... were beginning to dominate our market structure--distributors whose functions ranged from only partial performance of the wholesale function to those who were almost wholly integrated, that is, who were both wholesalers and retailers and often consumers as well.

Under these conditions classification of buyers became unprecise and shifting in meaning. Wholesalers and retailers no longer comprised clear-cut separate links between the producer and the ultimate consumer, each responsible for a clearly defined set of duties. Marketing functions became scrambled, with many permutations and combinations.... The number of patterns was legion and diverse.

Id. at 208. This presumably beneficial dynamism in the marketplace was threatened by what Chairman Howrey viewed as "a suspended state of confusion" in the law, which did not recognize that distributive activities traditionally carried out by sellers could often efficiently be shifted to buyers, in return for which the buyers could legitimately secure compensating discounts from the sellers. Id. And thus to bring clarity into this chaotic state of affairs, Chairman Howrey posited the following principle:

In our view, to relate functional discounts solely to the purchaser's method of resale without recognition of his buying function thwarts competition and efficiency in marketing, and inevitably leads to higher prices.... Where a businessman performs various wholesale functions, such as providing storage, traveling salesmen and distribution of catalogues, the law should not forbid his supplier from compensating him for such services.

Id. at 209. Chairman Howrey condemned as arbitrary an exclusionary rule that would forbid an integrated wholesaler from proving that it was indeed performing the wholesaling function. The matter, rather, should "be put to proof." Id. The evidentiary burden on the integrated wholesaler was not, the Chairman warned, to be insubstantial:

[T]he Commission should tolerate no subterfuge. Only to the extent that a buyer actually performs certain functions, assuming all the risks and costs involved, should he qualify for a compensating discount. The amount of the discount ... should not exceed the cost of that part of the function he actually performs on that part of the goods for which he performs it.

Id.

But as befits an imprecise statute which reflects, in Dean Pitofsky's words, a "tension between the philosophies of the Sherman Act and the Robinson-Patman Act--a conflict created by Congress and left unresolved on the FTC's doorstep many years ago," Complaint, Boise Cascade Corp., (dissenting statement of Comm'r Pitofsky), J.A. at 12, Chairman Howrey's suggested approach did not find fertile soil even within the offices of his fellow Commissioners. Foreshadowing what was destined to be the second competing approach, the Mueller rule, Commissioner Secrest in writing separately in Doubleday took issue with the Chairman's thinking. Commissioner Secrest advocated the approach adopted by the hearing examiner, who refused to consider evidence by three large book jobbers that the discount extended to them was offset by the costs of assuming wholesale functions which Doubleday would otherwise have to perform. In Commissioner Secrest's view, a discount is improper to the extent it compensates a buyer for services which, albeit helpful to the seller, also redound to the buyer's own benefit. Neither altruism nor servility were at work in the marketplace; the integrated buyer was out for its own welfare, even if its activities did in fact further the seller's commercial purposes. Calling for simplicity in enforcement, Commissioner Secrest laid down a warning if Chairman Howrey's approach were embraced:

Enforcement of the law would be extremely difficult if not impossible, if, in each 2(a) case, the Commission were required to divide a common service which may benefit both the buyer and the seller. Each case would require an operation as delicate and difficult as the separation of Siamese twins.

52 F.T.C. at 211 (Comm'r Secrest, concurring in result).

This contrary (soon to be called Mueller ) principle was thus informed by both instrumentalist and conceptual concerns: the Doubleday rule would be an immense headache to administer, and the rule would justify a discount where the buyer, by providing such services, was improving its own competitive position with other buyers. And this latter reality raised the spectre of Robinson-Patman's core concern for little fish competing with whales (or at least larger fish). Doubleday's rule, in short, "would undoubtedly give the larger buyer a price advantage in the resale of the seller's goods." Id. This same concern was echoed by Commissioner Mead, who likewise took issue with Chairman Howrey's approach:

Whether or not [the Chairman's view] is good economics, I am not prepared to say.... [But] it is not the law as expressed in the Robinson-Patman Act.... [It] paves the way for the ultimate annihilation of small retail dealers who are unable, by reason of their inability to perform the same marketing functions as their larger dual-functioning competitors to successfully compete with them.

Id. at 211-12 (Comm'r Mead, concurring in result).

The cacophony occasioned by Doubleday was short-lived. For in Mueller Co., 60 F.T.C. 120, the Commission squarely and unanimously rejected Doubleday's treatment of functional discounts. The Doubleday approach, in the Commission's later view, "ignores the fact that the favored buyer can derive substantial benefit to his own business in performing the distributional function paid for by the seller." Id. at 127. In the course of its decision, the Commission overturned the hearing examiner's finding that there was a failure of proof as to competitive injury. The Commission went on to hold in broad terms that a Robinson-Patman Act violation would lie even if the challenged discount (1) had been carefully limited to specific goods that Mueller's favored ("limit") jobbers had actually warehoused and (2) at most merely covered the jobbers' increased costs incurred by virtue of the warehousing services performed on those specified goods:

These findings mean only that [Mueller] has subsidized in whole or in part the "limit" jobbers' warehousing of certain products.... By doing so, [Mueller] has given this class of customers a substantial competitive advantage in the resale of such products. In this connection, the items on which the higher discount is given are ... the smaller, most commonly used products--those which are needed most frequently by the ultimate user, often to meet an emergency. That a jobber who has products of this type on hand is in a more favorable position than the jobber who does not is so obvious as to require little comment.

Id. at 128-29.

The Commission did not stop at this conceptually pure juncture, however. It went on to explore at some length the darker side of Mueller's purportedly benign regime of functional discounts. In contrast to Robinson-Patman's aim at securing a "level playing field" for all players, large and small, Mueller had in fact been bestowing the extra discounts on large jobbers with which it had long-standing relationships.

As so frequently happens in antitrust litigation, a documentary "smoking gun" found its way into evidence in Mueller indicating that the discount pricing system was not founded solely on a cold, economically rational decision to reimburse the favored jobbers for distributive functions performed by them. The tell-tale document in that case was a "get lost" letter from Mueller to one of its second-class ("regular") jobbers who sought to move into the more desirable ranks of Mueller's first-class ("limit") jobbers. The regular jobber had had the effrontery to request the higher discount because the ambitious jobber was planning "to stock several items which [the jobber maintained] should qualify [it] as a stocking distributor [or "limit" jobber]." Id. at 129. Mueller would have none of it.

We cannot see our way clear to change your 15% discount, for we do not have a 20% discount. We do have a greater one, however it applies only to those large stocking jobbers who place hundreds of orders with us throughout the year, totaling thousands of dollars.

Id. To leave no lingering doubts in the mind of the aspiring second-class jobber, the powers-that-be at Mueller administered the follow coup de grace:

These old established jobbers, who have been carrying MUELLER goods in large quantities for a number of years, are entitled to this protection, and until there might be some major change in your State and surrounding states, it will be necessary to continue the same differential that we have been allowing you.

Id. at 130.

Thus, upon analysis, Mueller's two-tiered structure of jobbers smacked of the regime of favoritism that had been condemned by the Supreme Court in Morton Salt. After all was said and done, Mueller was playing favorites based upon such factors as the volume of jobber purchases and whether the jobber was a member of the exclusive club, as it were, of "old established jobbers." Mueller's distribution system had succumbed to calcification.

Examination of this darker side of Mueller appears to bring the case closer to the ultimate holding in Doubleday. For while in the latter case the publishing company had won the battle over the principle of law, the victory proved empty as Doubleday lost the war. The Commission concluded that Doubleday, like Mueller and Morton Salt, was just another in a long line of manufacturers who play favorites among distributors with anticompetitive consequences of the sort that Congress intended in Robinson-Patman to condemn. Doubleday had, upon analysis, stacked the distribution deck in favor of a very limited class, "the so-called 'Big Three' jobbers," 52 F.T.C. at 206, who simply "treated the higher discounts as price reductions and not payments or allowances for services rendered." Id. at 209.

So it is that Mueller and Doubleday, the two warring rules spawned almost a generation ago by the Commission, shared a salient characteristic with Morton Salt and, years later, Falls City. In each instance, a manufacturer or supplier's idiosyncratic, non-economically justified pricing regime favored certain identified distributors (big distributors in Morton Salt, Doubleday and Mueller; home-State, or more precisely, non-Indiana distributors in Falls City ), with nothing but favoritism--the evil sought to be remedied through the Robinson-Patman Act--to explain the disparate treatment.

This case loomed at the Commission as the test of whether Mueller or Doubleday would survive in the present generation. The ALJ, whose thoroughness and care in this proceeding we would be remiss not to acknowledge, was thus presented with the unenviable task of trying two cases--one under Doubleday's teaching and the other under the starkly different approach enunciated by Mueller. The unorthodoxy of this two-pronged directive from Commission headquarters did not escape the Commissioners' notice. "The majority of the Commission today issues a complaint proceeding on two mutually exclusive, inconsistent theories of violation of the Robinson-Patman Act," was the apt description penned by Commissioner Clanton in dissent. Complaint, Boise Cascade Corp. (dissenting statement of Comm'r Clanton), J.A. at 11; see also id. (dissenting statement of Comm'r Pitofsky), J.A. at 21-22.

At day's end, as we have seen, the Commission (affirming the ALJ) embraced Mueller. And thus the case is now heralded as a battle of the Robinson-Patman titans, of Mueller versus Doubleday, with eminent authority summoned to support each of the two competing approaches. But we see a preliminary difficulty that precludes our declaring in this case whether the champions of Mueller or those of Doubleday should be the ultimate winners in the long-lived struggle over the application of Robinson-Patman to dual distributors in modern-day distribution systems. And that has to do with the threshold question whether in this case competitive injury, even as expansively defined under Robinson-Patman, see supra note 12, has been shown. Competitive injury, which no one disputes is the condition precedent for establishing violations of the Act, is a critical element. For, as we have seen, Robinson-Patman is directed to the preservation of competition. Indeed, the terms of the statute itself mandate that, in order for a price differential to be unlawful, it must tend "to injure, destroy, or prevent competition." 15 U.S.C. Sec. 13(a). Cf. also Foremost Pro Color, Inc. v. Kodak Co., 703 F.2d 534, 548 (9th Cir.1983), cert. denied 465 U.S. 1038, 104 S. Ct. 1315, 79 L. Ed. 2d 712 (1984); Whitaker Cable Corp. v. FTC, 239 F.2d 253, 256 (7th Cir.1956), cert. denied 353 U.S. 938, 77 S. Ct. 813, 1 L. Ed. 2d 761 (1957). Injury to competition is thus the name of the Robinson-Patman game.

In its opinion, the FTC waved aside substantial evidence (1) that competition among dealers generally was healthy, (2) that the selected dealers singled out for FTC examination were thriving, and (3) that this happy picture of prosperity was apparently unclouded by instances of diverted sales attributable to the challenged discounts. The Commission flatly concluded that this entire body of evidence was irrelevant. Indeed, in the Commission's view, for reasons that we fail to discern and the FTC failed to articulate, this evidence did "not address the causal connection [between the price differentials and competitive injury] at all." Commission Decision, J.A. at 191. This cannot be. In our view, the Commission's conclusion that Boise's dealer-specific evidence was irrelevant to the inference of competitive injury is wrong as a matter of law.

It is clear that Morton Salt's inference of competitive injury "may be overcome by evidence breaking the causal connection between a price differential and lost sales or profits," Falls City, 460 U.S. at 435, 103 S.Ct. at 1289. In reason, the inference can also be overcome by evidence showing an absence of competitive injury within the meaning of Robinson-Patman. That is to say, a sustained and substantial price discrimination raises an inference, but it manifestly does not create an irrebuttable presumption of competitive injury. Specific, substantial evidence of absence of competitive injury, see supra note 12, is, in our view, sufficient to rebut what is, after all, only an inference. The Commission, in effect, employed the Morton Salt inference to presume competitive injury conclusively in this case, and would only treat as relevant evidence "breaking the causal connection" between that assumed injury and the price discrimination to rebut the inference. This approach defies both logic and the import of Morton Salt that the inference of injury is rebuttable; for if the respondent's evidence demonstrates that there is no competitive injury (or reasonable possibility of competitive injury) to begin with, then evidence breaking the causal connection is obviously impossible to adduce. There is, under those circumstances, no causal connection to break.

As a result of its interpretation of Robinson-Patman, the Commission found it unnecessary to sift and weigh Boise's evidence of absence of injury. All those days of trial before the ALJ were, it seems, largely for naught. In reaching this odd result, the Commission simply failed to determine whether Boise's evidence demonstrated that no injury or "reasonable possibility" of competitive injury existed. See Falls City, 460 U.S. at 434-35, 103 S.Ct. at 1288.

First. Important among the salient facts that the Commission chose to ignore was the ALJ's finding that the selected dealers were not wallowing in a hopeless or deteriorating environment. Quite to the contrary, as we have already recounted, the ALJ found that all selected dealers for which data was available enjoyed an increase in sales and gross profits in excess of 22% annually during the period in question, despite the recessionary condition of the economy. ALJ Finding 431, J.A. at 121-22. Indeed, Judge Parker observed that various indicators showed high profitability and financial health on the part of the selected dealers. Id. 433-34, J.A. at 123-24. Although we need not decide the matter, this tends, if anything, to point to an absence of "lost profits" of the sort described in Falls City as one possible manifestation of injury to competition; accordingly, this evidence merited consideration by the Commission rather than the out-of-hand dismissal that it received.

There is yet another portion of the record relating to profitability left unexamined by the Commission and, in our view, deserving of attention in reaching a reasoned result. As computed by the Commission's accounting expert, Mr. Rowe, the selected dealers' median net profit before taxes as a percent of sales over the years 1976 to 1981 ranged from 2.3% to 3.5%. Id. 430, J.A. at 121. Their net profit was thus apparently lower than those of dealer-members of NOPA, whose net profit ranged from 3.0% to 3.9% from 1967 to 1980. Id. 423-27, J.A. at 120-21. See supra text at 1135. Although these statistics pose obvious comparative problems (in light of the fact that they are spread across different years and were computed by different entities), this apparent discrepancy in profitability obviously deserved consideration. But it received none. Likewise, the Commission failed to analyze the relationship of Boise's higher net profit (which, of course, reflected both its sales as a wholesaler and a dealer) to that of the selected dealers. See supra text at 1135.

In addition, the evidence, as found by the ALJ, failed to demonstrate "displaced sales," another form the Falls City Court indicated injury could take. The ALJ found that switches of accounts from one supplier to another were "not uncommon"; to the contrary, they were the order of the day. Id. 422, J.A. at 120. See supra text at 1135. While on the one hand twenty-one of the twenty-three dealers testified that they had lost accounts to Boise's lower prices or better service in recent years, ALJ Finding 384, J.A. at 112-13; see also id. 385-406, J.A. at 113-17, the phenomenon of lost accounts was very much a two-way street. As the ALJ extensively recounted, Boise also lost accounts and it too was unable to meet competitors' prices in some instances. See ALJ Finding 408-21, J.A. at 117-20. (The list is long, and as we have provided various examples, see supra text at 1136, we will avoid the temptation common in antitrust cases of proliferating the record. Automatic Canteen, 346 U.S. at 65-66 & n. 3, 73 S. Ct. at 1020-21 & n. 3.) Indeed, considered together, the number of accounts that switched from the selected dealers to Boise was quite small. See supra text at 1135-1136. And the proportion of switched accounts to a dealer's total accounts was strikingly low. Whereas a typical dealer might have a thousand accounts (or more), only a handful of accounts were shown to have switched. On top of that, the reasons for the switches were manifold. The following finding by Judge Parker makes the key point:

None of the selected dealers who lost accounts in whole or in part to Boise were able to conclude that the losses were due to the different prices charged them and Boise by the six manufacturers (citations omitted).

Id. 407, J.A. at 117. In short, the switches to Boise apparently cannot be explained as sales diverted through operation of the wholesale discount.15

Second. To explain its refusal to consider Boise's evidence (and, indeed, the countervailing evidence) of absence of actual competitive injury, the Commission asserted that "the competitive injury requirement of Section 2(a) is satisfied by a showing of 'a reasonable possibility that a price difference may harm competition.' " Commission Decision, J.A. at 192 (quoting Falls City, 460 U.S. at 434-35, 103 S.Ct. at 1288). While true, that principle does not justify abdication of the duty to consider evidence indicating that a "reasonable possibility" of harm does not, in fact, exist in the particular industry. Robinson-Patman has not ushered in a bizarre rule of law that exalts theory "no matter what" in the face of hard, cold facts. That curious rule, if it ever was a rule, has been scotched by Falls City's teaching.16

In this case, the Commission is attacking a long-established, widespread practice within a vibrant, dynamic industry. The ALJ specifically found that the wholesale discounts are no new-fangled concoctions. For example, Bates, a distinguished manufacturer whose products appear to enjoy considerable customer loyalty, "has had its wholesale discount since the 1940's, and Boise has been receiving that discount from Bates as long as it has been a customer." ALJ Finding 302, J.A. at 97. Forty years is a long time for evil, anticompetitive practices to have flourished unchallenged either by Washington, D.C. regulators or private entities which find themselves on the short end of the discount stick. Much the same can be said of Master Products. That firm's system of wholesale discounts has been unchanged for almost 40 years. Id. 345, J.A., at 105. So too with Boorum & Pease, whose even-handed pricing structure has been in effect for many years. Id. 263, J.A. at 91.

To be sure, the prevalence and longstanding nature of the discounts does not preclude the possibility that the discount regime posed a threat of injury to competition in the future. In light of the flourishing condition of the industry generally and the selected dealers specifically, however, one would reasonably expect that any potential for injury, if indeed there was one, would have manifested itself by now in the form of some dragging down. As yet, however, there appears to be no such damper.17

Third. In assuming, without analysis, the existence of competitive injury, we believe that the Commission erred in another way as well. It entirely failed to inform its application of Morton Salt's inference of injury with the purposes of the Robinson-Patman Act. As we have previously detailed, see supra text at 1138-1139, it is fairness, as Congress perceives it, that Robinson-Patman is all about. That Act, a legislative monument to smallness in a century which rewards bigness, is aimed at putting a halt to singling out the "Big Fives" and "Big Threes" of the world for disparately favorable treatment (unless, of course, a statutory defense lies).

This is not a case of the "Big Five" retailers (Morton Salt ), the "Big Three" book distributors (Doubleday ), the long-established, large jobbers who enjoyed manufacturer protection from uppity newcomers (Muller ) or the Hoosier beer distributor which happened to be on the "wrong" side of the Indiana-Kentucky state line (Falls City ). Quite to the contrary, there is among the six manufacturers selected in this case no regime of favoritism towards Boise. There is not the slightest hint in the ALJ's opinion, or in the Commission's for that matter, that the manufacturers were singling out Boise (and other big fish) for disparately favorable treatment. Rather, in accordance with a practice prevalent well before Boise even entered the industry, it is undisputed that the six manufacturers in question treated all wholesalers alike and provided the wholesale discount to Boise solely because it met their respective definitions of "wholesaler." There is simply no cabal between Boise and the six manufacturers to give Boise an additional price break.18

To recap briefly, there are five national wholesalers, ALJ Finding 26, 28, J.A. at 50-51, of which Boise is one of the two largest, and a host of smaller wholesalers. The "little fish" are typically "small, privately-owned businesses with annual volumes of under $10 million." Id. 27, J.A. at 51. Judge Parker's findings of fact show beyond doubt that equal treatment of all that qualify as "wholesalers," whether great or small, is the order of the day as to all six selected manufacturers. See, e.g., id. 447, J.A. at 127-28 (Rediform); id. 450, J.A. at 128 (Sheaffer-Eaton); id. 449, J.A. at 128 (Kardex); id. 451, J.A. at 128 (Boorum & Pease); id. 452, J.A. at 128-29 (Bates); id. 453, J.A. at 129 (Master Products).

This evenhandedness might strike Robinson-Patman aficionados as a rather odd choice as a basis for a claim of illegal price discrimination. Indeed, one attuned to such matters might more naturally expect that, if such a discrimination claim were to lie, the charge would be (1) that Boise (and perhaps the other national wholesalers, "the Big Five," as it were) was receiving higher discounts than smaller firms that qualified as "wholesalers," see e.g., Automatic Canteen, 346 U.S. 61, 73 S. Ct. 1017, Morton Salt, 334 U.S. 37, 68 S. Ct. 822, or (2) that Boise had coerced the manufacturers into framing their "wholesaler" definitions in a manner favorable to Boise's method of doing business. In either of these contexts, one could readily understand why the Robinson-Patman complaint would be levelled against Boise as a power buyer exerting its muscle to win additional pricing concessions from amenable, or at least nervous, vendors anxious to please the big-fish buyer. But, again, that is not this case. Boise has not been accused of piggishly demanding more than its equal share at the wholesale discount trough.

Fourth. We note, finally, the irony of the Commission's concept of Robinson-Patman liability, namely that these egalitarian-minded manufacturers with their long-settled commercial practices now stand condemned as price discriminators in violation of federal law. As the Supreme Court has made abundantly clear, a buyer's liability under section 2(f) is entirely derivative in nature. Great A & P Tea Co., 440 U.S. at 76-77, 99 S.Ct. at 931. See supra text at 1129. For there to be a guilty buyer, there must be a guilty seller. And thus it cannot pass unnoticed that under the Commission's approach, Bates, Master Products, Boorum & Pease and the other three manufacturers have run afoul of Robinson-Patman even though they have followed neutral, objective criteria in determining what buyers are to be considered wholesalers. It cannot be clearer that the Commission's sweeping attack, not narrowly focused on some anti-competitive, discriminatory pocket within an industry, carries with it dangers that the FTC is marching rapidly down a road leading to a regime of "price uniformity and rigidity" that has long stood condemned by the Supreme Court as antithetical to our law. This is nothing less than an all-out attack on uniform wholesale prices to dual distributors. Id. at 80, 99 S. Ct. at 933; Automatic Canteen, 346 U.S. at 63, 73 S.Ct. at 1019.

In sum, although the Commission (and the ALJ) correctly took the view that the Morton Salt inference is rebuttable, its rejection on relevancy grounds of Boise's evidence of the absence of competitive injury (or reasonable possibility of competitive injury) was in error.19 The petition for review is therefore granted, and the case is remanded for further proceedings consistent with this opinion.

It is so ordered.

I willingly concur in Judge Starr's excellent opinion and write separately to indicate the extraordinary difficulties facing the Commission if it should attempt to pursue this case on remand. On my reading of the law (which is somewhat less generous to the Commission than that of the majority), the practices complained of cannot be found to be illegal price discriminations.

In 1980, the Commission filed a complaint charging Boise with "knowingly [receiving] a discrimination in price" that may substantially lessen or injure competition in violation of Sec. 2(f) of the Robinson-Patman Act. 15 U.S.C. Sec. 13(f) (1982). The complaint attacked Boise's receipt of "functional discounts"--discounts that its suppliers offered to each and every customer who met certain objective criteria. The most important of these criteria was that the customer resell a minimum percentage of its purchases at wholesale. In the face of powerful evidence that the discounts in no way injured competition, the Commission assumed injury. Instead of evaluating the evidence of competitive injury, it considered only evidence that might "break the causal connection" between the imagined injury and the contested discounts. In support of this odd set of blinkers, it invoked FTC v. Morton Salt Co., 334 U.S. 37, 68 S. Ct. 822, 92 L. Ed. 1196 (1948), which allows an inference of competitive injury from the mere fact of price discrimination.1 As Judge Starr points out so well, Maj. at 1144, Morton Salt clearly cannot justify this complete dismissal of the issue of competitive injury vel non.

In my view, however, the case suffers graver faults than those identified by the court. The Commission's theory seems to me to place the Robinson-Patman Act in hopeless and complete conflict with the other antitrust laws, in violation of innumerable Supreme Court injunctions that it must be construed to avoid such conflict. E.g., Great Atlantic & Pacific Tea Co. v. FTC, 440 U.S. 69, 80-81 & n. 13, 99 S. Ct. 925, 933 & n. 13, 59 L. Ed. 2d 153 (1979); United States v. United States Gypsum Co., 438 U.S. 422, 447-59, 98 S. Ct. 2864, 2879-84, 57 L. Ed. 2d 854 (1978); Automatic Canteen Co. v. FTC, 346 U.S. 61, 63, 74, 73 S. Ct. 1017, 1024, 97 L. Ed. 1454 (1953). As a matter of law the very conditions on which sellers offer the discounts should preclude their being illegal price discriminations.

* * *

* * *

The Robinson-Patman Act has been no favorite among enthusiasts of competition. Price differentials may be found violative of the Act even when they represent the opening wedge of competition in a market that is largely cartelized or in which for some other reason competition is sluggish. See R. BORK, THE ANTITRUST PARADOX 385-91 (1978); R. POSNER, THE ROBINSON-PATMAN ACT 37-38 (1976). But the Act is law, together with the pro-competitive provisions of the antitrust laws. The courts accordingly have struggled to reconcile the two. In practice, they have read the Act to protect certain fairness goals, at the expense of competition. In no case, so far as I know, have they read the Act to stamp out a pro-competitive practice that in no way permits any competitor an unfair advantage.

Price discrimination in an economic sense--and I will shortly address the issue of Robinson-Patman cases that might be seen as extending the Act's reach beyond discrimination in that sense--causes disfavored purchasers to bear a disproportionate share of the seller's total costs of production. If variable production costs per unit are $1 and fixed costs $2, and a firm sells in equal amounts to A at $2.50 and to B at $3.50, B will bear 5/8 of fixed costs while A will bear only 3/8. (E.g., if 100 units are sold to each, sales to B will account for $250 of the $400 in total fixed costs, while those to A will account for only $150.) Such differentials appear to give the favored purchaser an edge over his competitors in the resale market, even where he has no superior efficiency to justify the advantage. The advantage has at least a look of unfairness. Thus it is not altogether surprising that Congress might seek to prevent such unfairness by prohibiting the practice, even though to do so may inhibit competition or injure consumers.2

We deal here, however, with a practice that cannot entail such an unequal allocation of costs. In instances of genuine price discrimination, the seller is able to extract a higher proportion of fixed costs from the disfavored customers because he has segmented the market. Critical to that segmentation is the existence of a barrier preventing resales by favored customers to disfavored ones--resales known as "arbitrage." In the absence of such a barrier, the favored customers would make such resales. Thus they would displace their own supplier and would capture for themselves the revenue that the original supplier sought to secure by his price discrimination.

Accordingly, it is black letter economics that price discrimination cannot occur if the favored customers can resell to the disfavored. See, e.g., P.R.G. LAYARD & A.A. WALTERS, MICRO-ECONOMIC THEORY 240-41 (1978); E. MANSFIELD, MICROECONOMICS: THEORY & APPLICATIONS 286 (2d ed. 1975); W. BAUMOL, ECONOMIC THEORY AND OPERATIONS ANALYSIS 347-48 (3rd ed. 1972); A. ALCHIAN & W. ALLEN, EXCHANGE AND PRODUCTION: THEORY IN USE 136-37 (1969).

Obviously the principle cannot apply if arbitrage is obstructed, and barriers may take a variety of forms. A physical problem is common. If, for example, local gas companies discriminate in favor of residential gas customers and against industrial ones, clearly the residential customers cannot increase their purchases and resell to industrial users: they lack control over the pipes necessary to reship the gas to their hypothetical purchasers. But physical barriers are hardly essential. The law itself may obstruct arbitrage,3 as may firm behavior. For example, a seller can sometimes monitor the favored customers' resales and credibly threaten to cut off their supplies if they resell to disfavored ones. The defendant salt manufacturer in Morton Salt, for example, surely possessed that modest monitoring capability. (The absence of discussion of the subject in that case, cf. Dissent at 1161, is hardly a sturdy basis for an inference that the salt seller could not penalize attempted arbitrage.) Discount recipients could conceivably organize a group boycott of disfavored buyers. Cf. Dissent at 1162 (reference to "price umbrella").

The present case is not merely bereft of evidence of any such barrier, but plainly demonstrates its absence. Far from restricting favored customers' sales to disfavored ones, or relying on any physical or legal barrier to their doing so, Boise's suppliers demand of the favored customers that they resell to disfavored ones. Indeed, the only condition uniformly imposed upon recipients of the discounts is that they make a minimum proportion of their sales to dealers (the disfavored customers). Boise's suppliers seek not to segment the market; they demand its integration. Far from seeking to discourage arbitrage, or relying on any physical or legal barrier to arbitrage, they insist on it. These are not the acts of a price discriminator.

Once the absence of any barrier to arbitrage is clear,4 the exact terms under which wholesalers resell to dealers, whatever they may be, cannot support a claim that manufacturers are offering wholesalers unmerited discounts. Suppose, for example, that wholesalers do not share any of their discounts with retailers to whom they sell. Cf. Dissent at 1162 ("[I]t is not to [dual distributors'] advantage to reduce prices for competing retailers."). If this be so,5 there appear to be two imaginable explanations. (1) There are no unmerited discounts; the costs of selling to wholesalers are lower or they perform services of offsetting value to manufacturers (or some combination). As the wholesalers have already given value for the nominal discounts, there is nothing for them to "pass" through to dealers. (2) The manufacturers are giving wholesalers unmerited discounts and, given the availability of arbitrage, are allowing them to capture the "excess profits" available on sales to the disfavored customers. The principle accepted without dissent by economists is that the second hypothesis is utterly implausible. So it seems to me. Nothing in the record appears to support the notion that office supply manufacturers are engaged in such a bizarre charitable program.

The Dissent suggests that the inferences from arbitrage-availability are undercut by findings that "disfavored retailers performed many of the functions that Boise performed." Dissent at 1155; see also id. at 1161. There is, however, no finding nor even a contention that any single retailer performs all the services provided by Boise or other wholesalers. See J.A. 139-40 (listing of some services provided by some dealers); Maj.Op. at 1134 n. 7. Nor is there any qualitative or quantitative evaluation of the services provided by the two classes of intermediaries. Thus there seems no reason to question the inferences that are normally to be drawn from ease of arbitrage.

I must confess puzzlement at the Dissent's contrasting "realistic appraisals of relevant competitive facts" with "general economic theorems." Dissent at 1161. The theorems relied on, if they should be so labelled, are hardly metaphysical speculations; they are inferred from widely shared propositions about human behavior, e.g., that firms do not gratuitously turn profit opportunities over to other firms for no reason. The Dissent contends that this proposition is not at issue, id. at 1161-1162, evidently on the basis of various theories which, if 100% true, would show only that wholesalers do not pass their discounts on to retailers. But the issue is precisely whether there are any unmerited discounts to pass on.6

Of course the fact that obstruction of arbitrage is a necessary condition of "price discrimination" as economists use the term would not prevent Congress from enacting a prohibition that spoke more broadly. Indeed, the Supreme Court has explicitly said that price discrimination for Robinson-Patman Act purposes means no more than a price differential. See FTC v. Anheuser-Busch, Inc., 363 U.S. 536, 549, 80 S. Ct. 1267, 1274, 4 L. Ed. 2d 1385 (1960). Under this view--which of course is authoritative--the price differential offered by Boise's suppliers to wholesalers is price discrimination. But the Act prohibits only price discrimination whose effect "may be substantially to lessen ... or to injure ... competition." 15 U.S.C. Sec. 13(a). The issue is whether this "price discrimination" can injure competition in any sense that Congress might have intended. Cf. F. ROWE, PRICE DISCRIMINATION UNDER THE ROBINSON-PATMAN ACT 96-97 (1962) (discussing Anheuser-Busch ).

Even Falls City Industries, Inc. v. Vanco Beverage, Inc., 460 U.S. 428, 103 S. Ct. 1282, 75 L. Ed. 2d 174 (1983), which the Commission appears to regard as the Court's most expansive extension of Morton Salt's readiness to infer injury to competition, makes clear that the Act is not intended to reach losses that are consistent with fairness. There a brewer charged an Indiana beer distributor more than it charged a competing Kentucky distributor. The Court held that unlawful price discrimination could be found (and that anticompetitive effect could be inferred from the price discrimination itself) even though the favored buyer was not "extraordinarily large." 460 U.S. at 436, 103 S. Ct. at 1289. Although this holding extended Robinson-Patman liability beyond the core image animating Congress--the buyer that uses sheer size to secure unfair advantages--it in fact did not disconnect Robinson-Patman liability from the economic concept of price discrimination or from a concern over unfair competitive advantages.

The economists' definition of price discrimination requires, to be sure, that the seller have market power (i.e., a power to raise the price above competitive levels without loss of all sales): otherwise the discriminatees would simply buy from alternative sources. Although Falls City does not refer to any finding of market power, as it was not at issue, the record might well have sustained such a finding. The seller was a regional brewer with an established brand name, and brand name loyalties typically enable a seller to raise prices without all customers abandoning it for alternative suppliers.

Moreover, Falls City expressly limited the permissibility of inferring competitive injury. It pointed out that the defendant brewer had identified no "economic reason" why the inference should not be drawn. 460 U.S. at 436, 103 S. Ct. at 1289. So far as appeared, the brewer's pricing practices enabled Kentucky wholesalers to increase their sales at the expense of Indiana ones regardless of the latters' efficiencies. As the Court noted, Indiana and Kentucky retailers competed in a single interstate market, so that the relatively high prices paid by Indiana wholesalers would necessarily impede their sales. Id. The message of the case is that where some inherent aspect of the practice supplies an economic reason why it could not generate such an unfair competitive advantage, the inference cannot be drawn.

The "price discrimination" here cannot lead to the sort of competitive injury evidently feared by Congress--an impaired ability to compete regardless of the losers' efficiency.7 If the disfavored customers bore a disproportionate share of their suppliers' costs, their ability to buy from the favored ones would enable them to escape the burden. The terms of the discounts themselves assure this result.

Because the nature of the challenged practice compels the conclusion that it could confer no competitive advantage of the sort the Robinson-Patman Act sought to eradicate, the practice should be absolved as a matter of law. The Supreme Court's mandate to construe the Robinson-Patman Act consistently with the other antitrust laws demands no less.

I am frankly unable to see how the Commission can resurrect this case, nor any reason beyond bureaucratic momentum why it should try to do so. With luck, however, it will make use of the remand to develop an intellectually coherent policy on functional discounts, one that addresses economic reality.

The majority today tells the Federal Trade Commission ("FTC") that manufacturers who charge different prices to competing retailers cannot be held to have violated price discrimination law--unless the FTC somehow overcomes evidence that competition persists in the affected business sector. Apparently, the majority imposes this new evidentiary burden to avoid a result it finds economically unappealing. Even if the majority is right in its economic appraisal, however, the choice is not ours to make. Though born of good intentions, today's decision disregards the policies that Congress purposefully created more than fifty years ago.

There can be no doubt that Congress' chief aim in enacting Robinson-Patman was to combat discrimination that favored big retailers--to counter "the impact upon secondary-line competition of the burgeoning of mammoth purchasers, notably chain stores." FTC v. Anheuser-Busch, Inc., 363 U.S. 536, 544, 80 S. Ct. 1267, 1271, 4 L. Ed. 2d 1385 (1960) (footnote omitted). Congress thus sought to ensure the survival of small retailers by requiring that they be placed on an equal footing with larger competitors.

Congress had a second objective in enacting Robinson-Patman. As I demonstrate below on the basis of its legislative history, Robinson-Patman was passed by a Congress that believed price discrimination favoring large buyers ultimately harmed competition itself. Congress was convinced that, by protecting small businesses, it was also protecting the operation of a competitive economy.

Two factors thus set Robinson-Patman apart from the rest of antitrust law. First, the fate of small retailers remained the Act's primary concern. Secondly, Congress determined that the proscribed activity--price discrimination between competing buyers (so-called "secondary-line" injury)--tended by its very nature to injure competition. Unlike many Sherman and Clayton Act cases, then, price discrimination suits would not require every court to reestablish the likelihood of market injury. Under Robinson-Patman, Congress had already identified and condemned the harmful consequences of this discrimination.

Today's decision disregards these crucial facts about Robinson-Patman law. It ignores, first, how much this case resembles the paradigm of Robinson-Patman discrimination. A giant, favored buyer here competes with smaller disfavored retailers. Secondly, the majority injects Sherman and Clayton Act doctrine into the distinct jurisprudence of Robinson-Patman's secondary-line cases. The majority in effect holds that, where a party can demonstrate continuing competition in a business sector--the sort of showing that might shield a conventional antitrust defendant from liability for restraint of trade--that party avoids liability for price discrimination, too. This novel proposition contradicts numerous judicial precedents concerning the evidence required to show secondary-line injury. It also threatens to frustrate FTC enforcement of Robinson-Patman. Because "[t]he determination whether to alter the scope of [Robinson-Patman] must be made by Congress, not this Court," Falls City Industries v. Vanco Beverage, Inc., 460 U.S. 428, 436, 103 S. Ct. 1282, 1289, 75 L. Ed. 2d 174 (1983), I dissent.

Evidence concerning the competitive structure of a business sector is precisely the sort that the Supreme Court has said need not be presented by complaint counsel in Sec. 2(a) cases and need not be considered before a court finds unlawful price discrimination. In FTC v. Morton Salt Co., 334 U.S. 37, 68 S. Ct. 822, 92 L. Ed. 1196 (1948), the Supreme Court explained the advisability of dispensing with such evidence. "It would greatly handicap effective enforcement of the Act," the Court held, "to require testimony to show that which we believe to be self-evident, namely, that there is a 'reasonable possibility' that competition may be adversely affected by a practice under which manufacturers and producers sell their goods to some customers substantially cheaper than they sell like goods to the competitors of these customers." Id. at 50, 68 S.Ct. at 830. Of course, not every disparity in the prices charged to competing retailers justifies the inference of an "adverse effect" on competition. But, "where the record indicates a price differential substantial enough to cut into the purchaser's profit margin ... [or one] that, if reflected in a resale price cut, would have a noticeable effect on the decisions of customers in the retail market, an inference of injury may properly be indulged." Foremost Dairies, Inc. v. FTC, 348 F.2d 674, 680 (5th Cir.) (footnote omitted), cert. denied, 382 U.S. 959, 86 S. Ct. 435, 15 L. Ed. 2d 362 (1965).

Once the inference of injury is justified, the burden logically shifts to the defendant to rebut the presumption. The Supreme Court made this evidentiary consequence explicit when it recently reaffirmed the "inference of injury" doctrine in Falls City Industries v. Vanco Beverage, Inc., 460 U.S. 428, 103 S. Ct. 1282, 75 L. Ed. 2d 174 (1983). The effect of invoking the inference, the Court said, is that the Robinson-Patman defendant can then present a specific type of rebuttal evidence. "In the absence of direct evidence of displaced sales, this inference may be overcome by evidence breaking the causal connection between a price differential and lost sales or profits." Id. at 435.

The Falls City Court cited a Fifth Circuit decision as an illustration of the type of evidence that could refute a "causal connection." Id. In Chrysler Credit Corp. v. J. Truett Payne Co., 670 F.2d 575 (5th Cir.), cert. denied, 459 U.S. 908, 103 S. Ct. 212, 74 L. Ed. 2d 169 (1982), the court rejected a car dealer's claim that a manufacturer's bonus program constituted price discrimination. Ultimately, the court found the price differential too small to warrant an inference of injury. But, the court also noted that plaintiff's franchise (unlike its competitors') was not located in an expanding sector of the city, that plaintiff had been unable to obtain financing for its used cars, and that plaintiff had shifted its sales focus from new car purchasers to fleet purchasers. Id. at 581. These factors supplied independent reasons for plaintiff's poorer performance among car dealers and thus could have broken a "causal connection" between the price differential and any decline in relative profits or market share.

As these precedents demonstrate, data concerning the level of competition in a given business sector usually does not fall within the range of rebuttal evidence. Thus, the majority today instructs the FTC to consider evidence that Falls City and Morton Salt say it is not obliged to assess. The majority rebukes the FTC for having "waved aside" evidence that "competition among dealers generally was healthy," maj. op. at 1143, and specifically directs the FTC to consider (1) the relative profit margins of retailers, wholesalers, and dual distributors, maj. op. at 1144-45, (2) the "longstanding nature" of discounts given to Boise, "[i]n light of the flourishing condition of the industry generally and the selected dealers specifically," id. at 1146, (3) the absence of a "cabal between Boise and the six manufacturers," id. at 1147, and (4) the fact that the discriminating sellers "have followed neutral, objective criteria" in establishing price differentials. Id. at 1147.

Though it requires the Commission to assimilate this eclectic array of evidence, the majority insists it is "adher[ing] to traditional principles of Robinson-Patman law." Maj. op. at 1146 n. 16. The majority cannot cite a single precedent, however, that finds this broad range of evidence to be a proper "rebuttal" to Morton Salt's inference of injury. The majority also protests that it has not imposed a "new evidentiary burden on the FTC." Id. But obviously the majority expects the Commission to do more with this evidence than simply accept it into the record. As the majority makes clear with its repeated references to the "intense competition" and the "absence of anti-competitive forces" in the office supplies sector, maj. op. at 1135, the FTC must somehow undercut or overcome such evidence of "competition" before the majority will countenance a finding of competitive injury. The practical effect of today's remand, then, is to force the Commission to "rebut the rebuttal." This does impose a new evidentiary burden on the FTC. Moreover, nothing in Robinson-Patman commands such an inquiry into market dynamics, and the Supreme Court has expressly pronounced it unnecessary.

As I read today's decision, my colleagues offer two justifications for diverging from the Supreme Court's evidentiary guidelines. First, "[t]his is not a case of the 'Big Five' retailers (Morton Salt ), the 'Big Three' book distributors (Doubleday ), the long-established, large jobbers who enjoyed manufacturer protection from uppity newcomers (Mueller ) or the Hoosier beer distributor which happened to be on the 'wrong' side of the Indiana-Kentucky state line (Falls City )." Maj. op. at 1147. The majority goes too far in reading any "big vs. small" element out of this record. It is true that many small wholesalers benefit from the functional discounts at issue. But the only wholesaler charged with using those discounts to compete against disfavored retailers is the largest distributor of office products in the nation: the Boise Cascade Corporation. J.A. at 46.

The majority's second reason for believing that "[t]his case is simply not of the lineage of Morton Salt," maj. op. at 1139 n. 14, is that it involves dual distributors. Applying price discrimination law to a dual distributor, however, is hardly a novel proposition. Indeed, the mass distributor of the '20s and '30s, against whom Robinson-Patman was originally aimed, "invested capital in facilities for performing bulk storage, redelivery, and financing, so as to 'integrate' the retailing and wholesaling functions within his own organization and to eliminate middleman profits by dealing with the manufacturer directly." F. Rowe, Price Discrimination Under the Robinson-Patman Act 4 (1962). See, e.g., Purolator Products, Inc. v. FTC, 352 F.2d 874, 882 (7th Cir.1965), cert. denied, 389 U.S. 1045, 88 S. Ct. 758, 19 L. Ed. 2d 857 (1968) (even discounts that do no more than compensate an integrated competitor for the cost of its integration are proscribed; "some WDs [warehouse distributors] have costs peculiar to their manner of doing business[, which] does not show ... that competitors of these WDs would not be injured if some of those costs are paid by" the manufacturer).

Thus, the majority's true reason for contending that the FTC is "venturing into relatively new territory," maj. op. at 1139 n. 14, is not that the case involves dual distributors. Rather, the majority is impressed by the "powerful and eloquent voices ... railing against" the type of enforcement action presented in this case. Id. Among those voices, my colleagues are particularly persuaded by Dean Pitofsky who, as a Commissioner, dissented from the FTC's complaint against Boise. Id.

Though I share my colleagues' regard for Dean Pitofsky's eloquence, I must point out that his dissent as a Commissioner was--much like today's decision--based more on economic philosophy than on an application of the statute before us. Commissioner Pitofsky conceded that Boise's price advantage "[a]rguably ... leads to an injury to competition, and therefore Section 2(a) of the Robinson-Patman Act is violated." J.A. at 15. The Commissioner nevertheless opposed FTC action in this case because he found "the Mueller rule[, holding that functional discounts are not exempt from Robinson-Patman's proscriptions,] is anticompetitive, anti-consumer and anti-efficient." J.A. at 17. Such economic judgments may well be appropriate when the Commission is deciding upon its enforcement priorities, but this court does not have the same discretion once the FTC has found a Robinson-Patman violation and that finding is brought before us on appeal.

Moreover, Commissioner Pitofsky's reason for opposing the complaint against Boise is now moot. The Commissioner said he would be willing to judge Boise's price advantage by the less stringent Doubleday rule (which permits functional discounts under certain conditions), rather than the Mueller doctrine. But the Commissioner was unsure, on the basis of the initial staff investigation, whether a violation could be proven under Doubleday. J.A. at 21 n.*. That uncertainty has now been resolved. Both the ALJ and the Commission found that the price discrimination favoring Boise was proscribed even under the more accommodating Doubleday rule because disfavored retailers performed many of the functions that Boise performed but Boise alone received a discount. J.A. at 159-61, 200-03. I note that the majority does not challenge that legal conclusion.

In sum, the majority's two reasons for disregarding traditional Robinson-Patman doctrine--that the price differentials in this case benefit small wholesalers and that they encourage dual distribution--are insufficient. These reasons do not negate the harm that such price discrimination may inflict on retail competitors and competition. Applying price discrimination law to dual distributors may well bring with it inefficiencies. But they are the type of inefficiencies that Congress accepted when it enacted Robinson-Patman. As I explore more fully below, Congress in 1936 set out to help independent retailers at the expense of large chain stores. It cannot have escaped Congress' notice that the giant retailers were often more efficient.

This case does implicate, then, some of the same concerns about the economic fate of smaller retailers that gave rise to Robinson-Patman. And, even if this case did not implicate such "originalist" concerns, it has long since been established that the Robinson-Patman Act "is of general applicability and prohibits discriminations generally." FTC v. Sun Oil Co., 371 U.S. 505, 522, 83 S. Ct. 358, 368, 9 L. Ed. 2d 466 (1963). "[N]either the scope nor the intent of the statute was limited to th[e] precise situation" that generated the Robinson-Patman amendments. Id. at 520, 83 S.Ct. at 367.

Just as price discrimination is proscribed regardless of the size of its victims, so the new evidentiary burden imposed by the majority threatens to frustrate Robinson-Patman enforcement in every setting. The majority does not confine today's holding to cases involving dual distributors or even functional discounts. And, by generally instructing the FTC to consider evidence of competitive health in a business sector, the court does much more than expand the record. Even if injury to the structure of competition were the focus of Robinson-Patman (and I indicate below that it is not), the burden of proof that my colleagues impose on the FTC undermines the Supreme Court's objective in Morton Salt. It permits a Robinson-Patman defendant, once he introduces any evidence that competition persists in his business sector, to stall or frustrate FTC action by forcing complaint counsel to assemble a detailed and time-consuming record.

In some cases, such a record will be irrelevant since we may be asking the Commission to prove something that has not yet occurred. As the Supreme Court has often reminded us, "Sec. 2(a) is a prophylactic statute which is violated merely upon a showing that 'the effect of such discrimination may be substantially to lessen competition.' " J. Truett Payne Co. v. Chrysler Motors Corp., 451 U.S. 557, 561, 101 S. Ct. 1923, 1927, 68 L. Ed. 2d 442 (1981) (emphasis in original). Even when the injury is assumed to have already occurred, however, the proof required by the majority will be inappropriate. Where a seller has imposed (or a buyer has benefited from) a substantial price difference that is not cost-justified, placing the burden of proof on that seller (or buyer) to rebut an inferred injury best accords with the equities of the situation. That burden of proof will be meaningless if it can be satisfied (and shifted back to the Commission) by simply showing that competition still exists in the relevant market.

Not only will the majority's evidentiary innovation often prove irrelevant or inappropriate; it will doubtless prove confusing, too. While the majority does specify which evidence the FTC must consider on remand in this case, maj. op. at 1144-47, my colleagues provide a murky standard for evaluating this evidence. The majority tells the Commission to "determine whether Boise's evidence demonstrated that no injury or 'reasonable possibility' of competitive injury existed." Maj. op. at 1144. But this simply begs the question: What constitutes such injury? The majority implies, without clearly stating, that competitive injury is manifested only when competition itself threatens to disappear from a given market. If that is the majority's standard, the majority still needs to tell the Commission how serious that threat must be. But, aside from such questions of degree, the majority's standard also raises a more fundamental legal issue.

3. Distinguishing Robinson-Patman From Other Antitrust Laws

The majority's willingness to increase the evidentiary burden on the FTC reflects, in my view, a misunderstanding of RobinsonPatman doctrine. The majority proclaims that "Robinson-Patman is directed to the preservation of competition." Maj. op. at 1143. This description of the statute's purpose is simply too restrictive. The Congress that enacted Robinson-Patman did not focus on the process of competition to the exclusion of the living players in that process. This point is revealed in the statutory language itself. The majority presents the statute as proscribing only price differentials that tend to "injure, destroy, or prevent competition." Id. But the full text of this crucial passage banishes discrimination that would "injure, destroy, or prevent competition with any person who either grants or knowingly receives the benefit of such discrimination, or with the customers of either of them." 15 U.S.C. Sec. 13(a) (1982) (emphasis added).

Thus, as the Supreme Court long ago realized, "Congress intended to protect a merchant from competitive injury attributable to discriminatory prices...." Morton Salt, 334 U.S. at 49, 68 S.Ct. at 829 (emphasis added). The amendment to the Clayton Act "was intended to justify a finding of injury to competition by a showing of 'injury to the competitor victimized by the discrimination.' " Id. (footnote omitted; emphasis added). On the facts of Morton Salt, the FTC could thus find a Sec. 2(a) violation because "less-than-carload purchasers [of salt] might have been handicapped in competing with the more favored carload purchasers by the differential in price...." Id. at 50, 68 S.Ct. at 830.

The Morton Salt result exemplifies Professor Sullivan's summary of price discrimination case law:

Robinson-Patman ... secondary line cases lack any requirement that injury be shown to competitive structure or process; they rely instead on injury to particular firms in the market.... There is warrant in the history of the statute for construing it so and the courts have regularly done so.

L. Sullivan, Handbook of the Law of Antitrust 694 (1977) (footnote omitted). Often, injury to competitors does involve (or at least foreshadow) some impact on competition. As the Ninth Circuit pointed out very recently, "[c]ompetition does not exist in a vacuum; it consists of rivalry among competitors." Hasbrouck v. Texaco, Inc., 830 F.2d 1513, 1518 (9th Cir.1987). "With respect to price discrimination claims, the significance of harm to competitors is particularly clear." Id.

In Morton Salt, for example, the court found that "the competitive opportunities of certain merchants were injured when they had to pay respondent substantially more for their goods than their competitors had to pay." 334 U.S. at 46-47, 68 S. Ct. at 828 (emphasis added). The Fifth Circuit has similarly held that "a substantial price advantage can afford a favored buyer a material capital advantage by enlarging his profit margin in a highly competitive field or it can enable him to offer customer-attracting services which will give him a substantial advantage over his competition." Foremost Dairies, Inc. v. FTC, 348 F.2d 674, 680 (5th Cir.), cert. denied, 382 U.S. 959, 86 S. Ct. 435, 15 L. Ed. 2d 362 (1965). Indeed, the ALJ in the present case found that even if victimized retailers experienced growth in their own sales this did not foreclose injury to their ability to compete. "[T]heir growth would have been even greater, and their profits would have increased substantially, if they had enjoyed the pricing which the six manufacturers extended to Boise." J.A. at 155.

As these examples illustrate, the relationship between injury to competitors and injury to the structure of competition is highly variable. When the former injury does produce the latter, that market impact can be anything from severe to attenuated. The academic commentators are agreed on this range of market effects. See L. Sullivan, Handbook of the Law of Antitrust 681-82 (1977); H. Hovenkamp, Economics and Federal Antitrust Law 344-45 (1985); F. Scherer, Industrial Market Structure and Economic Performance 323-25 (2d ed. 1980).

Notwithstanding the ambiguous relationship between injury to competitors and injury to competitive structure, Congress saw fit to proscribe price discrimination in secondary-line cases even when only the first of these injuries could be proven. To be sure, not every court has recognized this fact. Indeed, Boise identifies several decisions that renounce injury to competitors as a sufficient basis for Sec. 2(a) liability. See Petitioner's Brief at 31. Most of these cases, however, involve distinguishable (and sometimes unusual) facts. E.g., General Foods Corp., 103 F.T.C. 204 (1984) (primary-line injury); Edward J. Sweeney & Sons v. Texaco, Inc., 637 F.2d 105 (3d Cir.1980),cert. denied, 451 U.S. 911, 101 S. Ct. 1981, 68 L. Ed. 2d 300 (1981) (price difference consisted only of a hauling allowance that was even-handedly applied); American Oil Co. v. FTC, 325 F.2d 101 (7th Cir.1963), cert. denied, 377 U.S. 954, 84 S. Ct. 1631, 12 L. Ed. 2d 498 (1964) (price difference was of very brief duration). The remaining opinions that Boise cites are simply aberrations. See, for example, Foremost Pro Color, Inc. v. Eastman Kodak Co., 703 F.2d 534 (9th Cir.1983), cert. denied, 465 U.S. 1038, 104 S. Ct. 1315, 79 L. Ed. 2d 712 (1984).

If there were any doubt that decisions like Foremost Pro Color stray from the proper reading of Robinson-Patman, it is laid to rest by Robinson-Patman's own legislative history. That record clearly demonstrates that Congress equated any significant injury to competitors with injury to competition. As the House Report explained the need to amend the Clayton Act:

The existing law has in practice been too restrictive in requiring a showing of general injury to competitive conditions in the line of commerce concerned, whereas the more immediately important concern is in injury to the competitor victimized by the discrimination. Only through such injury in fact can the larger, general injury result. Through this broadening of the jurisdiction of the act, a more effective suppression of such injuries is possible and the more effective protection of the public interest at the same time is achieved.

H.R.Rep. No. 2287, 74th Cong., 2d Sess. 8 (1936). See also S.Rep. No. 1502, 74th Cong., 2d Sess. 4 (1936) (similar language). Congress could not have stated more clearly its conviction that, where a competitor was substantially injured, the market would eventually be injured, too. "To catch the weed in the seed," the Senate Report observed, "will keep it from coming to flower." Id.

The reason for limiting Robinson-Patman's proscription to price differences that tend to "injure, destroy or prevent competition" was simply to exclude "from the bill otherwise harmless violations of its letter." Id. In other words, the "injury to competition" test was not intended to leave substantial numbers of price differentials intact. According to the House Report, the bill "prohibit[ed] discriminations in price between purchasers where such discriminations cannot be shown to be justified by differences in the cost...." H.R.Rep. No. 2287, 74th Cong., 2d Sess. 3 (1936) (emphasis added). The same document reiterated Congress' overriding concern "that the survival of independent merchants, manufacturers, and other businessmen is seriously imperiled and that remedial legislation is necessary." Id.

At bottom, today's decision refuses to recognize the inherent tension between the Robinson-Patman Act and the rest of antitrust law. The majority claims the Supreme Court has instructed us not to interpret Robinson-Patman in ways that " 'conflict with the purposes of other antitrust legislation.' " Maj. op. at 1138 (quoting Automatic Canteen Co. v. FTC, 346 U.S. 61, 63, 73 S. Ct. 1017, 1019, 97 L. Ed. 1454 (1953)). But my colleagues have wrenched this quotation from its context; the Supreme Court counsels us to bend price discrimination law to antitrust principles only when FTC enforcement threatens to "extend beyond the prohibitions of the [Robinson-Patman] Act." Id. (emphasis added). Indeed, what the Court in Automatic Canteen (and in two subsequent cases) declined to do was "to read [Robinson-Patman's] ambiguous language" about a seller's meeting-competition defense in ways that would violate "congressional policy as expressed in other antitrust legislation." Id. at 73, 73 S.Ct. at 1024. See also Great Atl. & Pac. Tea Co. v. FTC, 440 U.S. 69, 80 & n. 13, 99 S. Ct. 925, 933 & n. 13, 59 L. Ed. 2d 153 (1979); United States v. United States Gypsum Co., 438 U.S. 422, 458-59, 98 S. Ct. 2864, 2884-85, 57 L. Ed. 2d 854 (1978).

When, however, a case falls within Robinson-Patman's clear prohibitions, the Supreme Court recognizes that a special concern for small business and "fairness" animates Robinson-Patman--and that these values sometimes conflict with the free operation of competitive markets. "Congress intended to assure, to the extent reasonably practicable, that businessmen at the same functional level would start on equal competitive footing so far as price is concerned." FTC v. Sun Oil Co., 371 U.S. 505, 520, 83 S. Ct. 358, 367, 9 L. Ed. 2d 466 (1963). "The Robinson-Patman Act was passed to deprive a large buyer of [certain] advantages," the justices declared in Morton Salt, and its "avowed purpose [was] to protect competition from all price differentials except those based in full on cost savings." 334 U.S. at 43, 44, 68 S. Ct. at 826, 827.

Commentators have framed the conflict with antitrust principles more starkly. As Professor Scherer observes,

it seems clear that the courts and the Federal Trade Commission are more inclined to err on the side of protecting competitors than protecting competition when the two goals are not congruent. And in this respect, the criteria applied when enforcing the Robinson-Patman Act are at odds with the broader pro-competitive objectives of antitrust.

F. Scherer, Industrial Market Structure and Economic Performance 578 (2d ed. 1980). Professor Hovenkamp is even more blunt in making this point: "[W]hile the Robinson-Patman Act is quite hostile toward competition it is nevertheless disguised as an antitrust law." H. Hovenkamp, Economics and Federal Antitrust Law 346 (1985).

Obviously, protecting competitors from this type of price competition does not always maximize "consumer welfare." Yet, the majority insists that Robinson-Patman enforcement must not diverge from "maximization of consumer welfare," maj. op. at 1138, because to do so would violate the underlying principles of antitrust. Such a result, the majority says, would suggest "Congressional schizophrenia." Id. The majority mistakenly equates clearheadedness with singlemindedness. Congress has enacted many laws, under many different circumstances, to serve many disparate purposes. Our role as courts is to apply the laws at issue in a given case without doing violence to related statutes. It is not for us to legislate a consistency that Congress did not enact.

Thus, we may "harmonize" laws so that enforcement of one does not produce violations of another, cf. Automatic Canteen Co. v. FTC, 346 U.S. at 73, 73 S. Ct. at 1024, but we must not entirely subjugate one law to another. I think that the majority has overstepped this fine line by a quantum leap. The court today makes price discrimination law subservient to the rest of antitrust doctrine, when it declines to enforce Robinson-Patman in any way that might undermine "maximization of consumer welfare." As the Supreme Court has said of Robinson-Patman enforcement, "we are not free on the basis of our own economic predilections to make the choice between harm to [two groups] ...; that choice is foreclosed by the determination in the statute itself in favor of equality of [price] treatment." FTC v. Sun Oil Co., 371 U.S. 505, 519, 83 S. Ct. 358, 367, 9 L. Ed. 2d 466 (1963). The majority today has opted for what it thinks is good free-market philosophy--at the expense of faithfully interpreting the law that Congress passed. That is not the role of a court interpreting policy made by others.

4. The Commission's Determination Should be Upheld

Since I decline to impose a new evidentiary burden upon the FTC, the question that I believe is before this court is whether the Commission abused its discretion or was unsupported by substantial evidence in holding that the functional discounts in this case constitute proscribed price discrimination. See Ash Grove Cement Co. v. FTC, 577 F.2d 1368, 1378 (9th Cir.) ("appellate review of [FTC] fact-findings [is confined] to a determination whether they are supported by substantial evidence"), cert. denied, 439 U.S. 982, 99 S. Ct. 571, 58 L. Ed. 2d 653 (1978); Colonial Stores Inc. v. FTC, 450 F.2d 733, 740 n. 14 (5th Cir.1971) ("even when the Commission's findings[ ] are framed in terms of legal conclusions, their presumptive validity is considerable. FTC v. Mary Carter Paint Co., 1965, 382 U.S. 46, 48-49 [86 S. Ct. 219, 221-22, 15 L. Ed. 2d 128 (1965) ].").

Ultimately, the ALJ's and the Commission's finding of competitive injury rested on one essential fact: the size of the functional discounts that Boise received (which ranged from 5 to 33 percent) exceeded the net profit rates of the dealers. These profit margins were quite low because of the intense competition in the office supplies business. "Considering these facts," the ALJ stated, "the only possible inference is that the effect of the substantial and sustained price discriminations favoring Boise may be to destroy or prevent competition with the unfavored dealers." J.A. at 152.

The fact that a substantial price discount exceeds the disfavored dealers' profit margins or occurs in a highly competitive sector has been found to justify the Morton Salt inference of injury in many cases. See, e.g., Corn Products Refining Co. v. FTC, 324 U.S. 726, 742, 65 S. Ct. 961, 969, 89 L. Ed. 1320 (1945); Hasbrouck v. Texaco, Inc., 830 F.2d 1513, 1519 (9th Cir.1987); Kroger Co. v. FTC, 438 F.2d 1372 (6th Cir.), cert. denied, 404 U.S. 871, 92 S. Ct. 59, 30 L. Ed. 2d 115 (1971); Foremost Dairies, Inc. v. FTC, 348 F.2d 674 (5th Cir.), cert. denied, 382 U.S. 959, 86 S. Ct. 435, 15 L. Ed. 2d 362 (1965); United Biscuit Co. v. FTC, 350 F.2d 615 (7th Cir.1965), cert. denied, 383 U.S. 926, 86 S. Ct. 930, 15 L. Ed. 2d 845 (1966); Monroe Auto Equip. Co. v. FTC, 347 F.2d 401 (7th Cir.1965), cert. denied, 382 U.S. 1009, 86 S. Ct. 613, 15 L. Ed. 2d 525 (1966).

In this case, the inference of injury was not rebutted by any evidence that the price differences were somehow offset or otherwise would not harm disfavored retailers. Indeed, there was evidence reinforcing the likelihood of harm: the ALJ found that many of the wholesaling functions that ostensibly justified Boise's discounts were also performed by the retailers who received no discounts. J.A. at 139, 141, 159. Thus, the retailing costs incurred by Boise and its disfavored competitors were likely to be similar. Additional evidence suggested that profit margins were higher for Boise than for its disfavored competitors--although, as the majority points out, these statistics were not strictly comparable. Maj. op. at 1144. Given this evidence and the numerous precedents supporting the Commission's treatment of substantial price differentials, I believe the Commission's result must be upheld. Especially is this so, since "[t]he precise impact of a particular practice on the trade is for the Commission, not the courts, to determine." FTC v. Motion Picture Advertising Service Co., 344 U.S. 392, 396, 73 S. Ct. 361, 364, 97 L. Ed. 426 (1953). See also Corn Products Refining Co. v. FTC, 324 U.S. at 739, 65 S.Ct. at 967.

The advisability of deferring to the Commission's reading of close cases is highlighted, I think, by Judge Williams' concurrence. In counseling the Commission on what it can conclude from this case, Judge Williams relies on abstract economic models to interpret ambiguous facts and solve real market problems. Apparently, my colleague prefers "black letter economics," concurring opinion ("con. op.") at 1149, to black letter law.

The particular economic model that Judge Williams invokes is the process of arbitrage. Judge Williams believes there can be no price discrimination unless there is a "barrier preventing resales by favored customers to disfavored ones." Con. op. at 1149. Since, in this case, favored wholesalers do sell to disfavored retailers, my colleague assures us that "[t]he 'price discrimination' here cannot lead to the sort of competitive injury evidently feared by Congress." Con. op. at 1152 (emphasis added). Despite a record clearly demonstrating that disfavored dealers have paid thousands of dollars more than Boise for the same office supplies, Judge Williams is confident that no injury has occurred for one reason: this record does not exhibit any of the "textbook" barriers to arbitrage. I do not share my colleague's willingness to exalt the model over the record--nor, I imagine, did those who drafted the Robinson-Patman Act.

I indicate below why reliance on a simplified model to infer crucial facts may be misplaced. But it is important to note at the outset that, by focusing upon arbitrage, my colleague transforms Robinson-Patman doctrine in two substantial and related ways. First, Judge Williams is willing to assume that arbitrage (or a viable threat of arbitrage) has done its work if favored purchasers who receive price discounts "perform services of offsetting value to manufacturers." Con. op. at 1150. Such a showing may satisfy Judge Williams, but it does not satisfy Robinson-Patman. Even under the more accommodating Doubleday rule, a price discount is not legal simply because the beneficiary performs "offsetting services." Rather, all buyers who perform similar services must also be eligible for the discount. In this case, the FTC expressly found that disfavored retailers perform services similar to those performed by Boise yet are ineligible for the discount. J.A. at 201-02. Thus, the discounts in this case run afoul of Robinson-Patman, even if they pass the "economic" test for injury that my colleague has newly fashioned.

The second way in which the focus upon arbitrage transforms Robinson-Patman doctrine is that it effectively alters the litigation burden. Just as the majority opinion imposes a new evidentiary duty on the FTC, so the concurring opinion, for its part, creates a new presumption favoring defendants. Despite Morton Salt's inference of injury whenever substantial price differences are imposed on competing buyers, Judge Williams creates an inference of noninjury whenever his model predicts that arbitrage must have occurred. Evidently, if no physical or other "barriers" between favored and disfavored buyers appear in the record, the courts should presume (whether rebuttably or irrebuttably, I am not sure) that arbitrage has eliminated any ill effects of price discrimination.

Thus, Judge Williams' reliance on the concept of arbitrage is not compatible with current Robinson-Patman doctrine. Even if it were compatible, however, I would still question my colleague's willingness to infer that effective arbitrage has occurred (and that competitive injury has been eliminated) on the basis of abstract principles about resale "barriers." Numerous Robinson-Patman precedents demonstrate the defect in Judge Williams' approach. Those precedents counsel reliance on "realistic appraisals of relevant competitive facts," FTC v. Sun Oil Co., 371 U.S. 505, 527, 83 S. Ct. 358, 370-71, 9 L. Ed. 2d 466 (1963), not general economic theorems. The "relevant competitive facts" in Morton Salt, for example, suggested none of the barriers to arbitrage that Judge Williams would make prerequisite to Sec. 2(a) violations. Indeed the five buyers who benefited from the lowest prices for Blue Label salt were precisely the sort of integrated retailers that Boise exemplifies in the present case. 334 U.S. at 41 n. 4, 68 S. Ct. at 826 n. 4. The opinion cited no evidence of barriers--physical, legal, contractual or otherwise--that prevented favored buyers from engaging in arbitrage by reselling the salt to disfavored buyers. Nevertheless, the Court found price discrimination and competitive injury.

Judge Williams' economic model yields only two explanatory "hypotheses" for the facts in Morton Salt. Con. op. at 1150. Since arbitrage is clearly "available" according to the model, either (1) manufacturers are giving "unmerited discounts" to the favored buyers or (2) "there are no unmerited discounts." Id. "The principle accepted without dissent by economists," Judge Williams reminds us, "is that the first hypothesis is utterly implausible." Id. My colleague is right, of course, that economists would not expect "office supply manufacturers [to] engage[ ] in ... a bizarre charitable program." Id. But simply to recite this conclusion is to recognize its emptiness. The real question is whether economists "accept without dissent" the Hobson's choice of explanatory "hypotheses" that Judge Williams offers.

Are there "relevant competitive facts," more compelling than my colleague's "hypotheses," that may explain why integrated retailers like Boise (or like Safeway and A & P, in Morton Salt ) do not engage in arbitrage? I believe there are. Without trespassing on the work of the FTC, whose familiarity with this record and general expertise in analyzing trade restraints leave it best equipped to evaluate Boise's price advantage, I would at least point out some aspects of this case that may have thwarted the process of arbitrage. These factors operate on the two levels at which the arbitrage sales might occur: retail and wholesale.

Consider, first, the possibility of arbitrage sales from favored to disfavored retailers. Although retail sales account for only forty-two percent of Boise's office supplies business, J.A. at 57, this average is misleading. Boise's proportion of retail business in particular sales regions is usually much lower or much higher. In seven of the eight distribution centers that the FTC examined in this case, Boise's proportion of retail sales ranged between eighty and ninety percent. J.A. at 58. When a dual distributor chiefly functions in a retail capacity (but purchases its inventory at wholesale rates), it is not to the distributor's advantage to reduce prices for competing retailers by engaging in arbitrage. Boise's profit margins are higher if it takes the same items it could sell to competing retailers and, instead, sells them directly to end users. Moreover, if Boise undercuts the prevailing price to the other retailers, those retailers can then lower their own prices to customers, which in turn would adversely affect Boise's direct sales to end users.

The logic of this economic calculus undoubtedly informed the strategies of mass retailers in the '20s and '30s. Indeed, the FTC investigation that spawned the Robinson-Patman Act concluded that "lower selling prices are a very substantial, if not the chief, factor in the growth of chain-store merchandising, and ... lower buying prices than are available to independent [stores] are a most substantial, if not the chief, factor in these lower selling prices." Federal Trade Commission, Final Report on the Chain-Store Investigation, S.Doc. No. 4, 74th Cong., 1st Sess. 53 (1935). Judge Williams' conclusion that price discrimination could not occur in such circumstances (because there were no barriers to arbitrage) is unpersuasive. It cannot explain why Robinson-Patman was enacted or why disfavored retailers complained so bitterly before Congress stepped in.

Notwithstanding the fact that the competitive injury in this case is inflicted by dual distributors acting in their retail capacity, Judge Williams expects the injury to be mitigated by arbitrage occurring in a different context: sales by favored wholesalers to disfavored retailers. Con. op. at 1150. One difficulty with this view is that, even if arbitrage effectively eliminated discrimination between wholesalers and disfavored retailers, it still might not mitigate such injury inflicted by favored retailers. This would be true, for example, if wholesalers who engage in arbitrage cannot cut their resale prices as low as favored retailers can because the wholesalers perform more costly "offsetting services."

Even assuming, however, that arbitrage between wholesalers and retailers could eliminate all competitive injury, there are factors in this record that might interfere with that process. To begin with, the record shows that wholesalers try very hard to stay informed of the prices that manufacturers charge to retailers. Such information largely determines wholesalers' own pricing toward retailers. J.A. at 51-54. This suggests that manufacturers may be exerting price leadership. If they have indeed established some sort of price umbrella, wholesalers may be forsaking arbitrage for the predictable profits of somewhat higher prices.

Effective arbitrage also entails some threshold costs of identifying disfavored buyers of substantial quantities of each good that is discriminatorily priced and of fashioning the appropriate discount on each of those items for each of those buyers. Price competition on office supplies may be especially difficult--and unrewarding--because wholesalers like Boise carry more than 8,000 products in their catalogue, J.A. at 51, and many retailers may purchase only a small number of items or from only a few manufacturers' product lines. The record supports this conclusion. Favored wholesalers more frequently compete for retail accounts on the basis of their overall service or their salespersons than by selectively discounting their catalogue prices. J.A. at 54, 61, 123a-124.

The record also reveals that net profit rates for Boise and other wholesalers are significantly higher than among disfavored retailers. J.A. at 62, 121. This suggests that whatever price competition occurs among wholesalers to win retailer accounts is constrained. In other words, wholesalers will not cut their prices below a certain level that preserves their higher profits. Once again, the fact that wholesalers more often compete on the basis of their service than on the basis of discounts supports this conclusion that price competition is limited.

Merely describing these factors that may hinder the process of arbitrage illustrates why a close understanding of a particular market is crucial for determining whether price discrimination has occurred. It is precisely that detailed familiarity with the facts that the FTC possesses and that this court lacks. Absent some indication that the Commission has erred in the law or not supported its conclusions with the evidence, I would refrain from foisting this court's own theory of the case upon the FTC. The operation vel non of a process of arbitrage is one of those matters on which "the inferences to be drawn from [the facts proven] are for the Commission to determine, not the courts." Corn Products Refining Co. v. FTC, 324 U.S. at 739, 65 S.Ct. at 967-68.

Implicit in Judge Williams' instructions to the FTC is a fear that Robinson-Patman doctrine will not accommodate economic realities like arbitrage unless this court forces the issue. But the defense of availability, which has been repeatedly recognized by the courts, allows a defendant to defeat a charge of price discrimination precisely by proving that the lower prices were available to the disfavored purchasers from some other source--i.e., from an arbitrager. See Tri-Valley Packing Ass'n v. FTC, 329 F.2d 694 (9th Cir.1964). In this case, Boise raised the availability defense, and the FTC found it was unsupported by the evidence. J.A. at 124-29.

The majority today rewrites a controversial law with its decision to send the case back to the FTC. I may share my colleagues' doubts about whether the price discrimination enjoined in this case produces such pernicious effects. But I am unwilling to create new doctrine in order to accommodate my policy views. Just as "[t]he Fourteenth Amendment does not enact Mr. Herbert Spencer's Social Statics," Lochner v. New York, 198 U.S. 45, 75, 25 S. Ct. 539, 546, 49 L. Ed. 937 (1905) (Holmes, J., dissenting), so Congress did not mandate a "turbulent sea of pro-competitive efficiency and maximization of consumer welfare," maj. op. at 1138, when it enacted Robinson-Patman. That Act has been held repeatedly to proscribe price differences of the substantial size and duration that are exhibited here. I would uphold the Commission's injunction.

I dissent.