Being a “glass is half-full” type of guy, I figured there was no way the D.C. Circuit’s decision on Whole Foods’ petition for rehearing en banc could turn out poorly: Either the court would grant the motion and correct the panel’s mistakes, or the court would deny the motion, setting up an attractive opportunity for the Supreme Court, which hasn’t decided a significant merger case since 1974 and badly needs to update its doctrine (see, e.g., Brown Shoe, upon which the D.C. Circuit heavily relied in Whole Foods).
It seems my optimism was unwarranted. On November 21, the July 29 panel decision was amended and reissued so that Judge Tatel no longer joins Judge Brown’s former majority opinion. Although the two judges agree that the FTC did present evidence that could establish a separate antitrust market of “premium natural and organic supermarkets” (“PNOS”) and that the district court’s denial of the FTC’s preliminary injunction of the Whole Foods/Wild Oats merger must therefore be reversed (and remanded for a weighing of equities), their grounds for reaching those conclusions differ. There is thus no majority opinion from the panel (the third judge, Judge Kavanaugh, dissented), and the other D.C. Circuit judges apparently reasoned that rehearing en banc would therefore be unwarranted. (See the statement of Judges Sentelle and Ginsburg here.)
The good news is that neither of the two opinions in favor of the FTC commanded a majority of the panel, so neither stands as authoritative precedent. That’s cold comfort, though, for the panel holding has precedential effect, and, under the Marks rule, any common principles contained in the Brown and Tatel opinions and necessary to their conclusions will have such effect. Moreover, so few published decisions address merger challenges that those opinions — regardless of precedential effect — are likely to have substantial influence. That’s highly unfortunate, for both opinions contain fundamental mistakes, as I elaborate below the fold.
Problems With Judge Brown’s Opinion
Putting aside the toothless standard of review her opinion endorses — a standard also adopted by Judge Tatel and thus established as precedent under the Marks rule — Judge Brown’s decision veers off course by reasoning that preferences of core customers could define the contours of antitrust markets even without price discrimination. Specifically, Judge Brown stated that
a core group of particularly dedicated, ‘distinct customers,’ paying ‘distinct prices,’ may constitute a recognizable submarket, whether they are dedicated because they need a complete ‘cluster of products,’ because their particular circumstances dictate that a product ‘is the only realistic choice,’ or because they find a particular product ‘uniquely attractive.’
A couple of initial points on this statement:
First, it’s literally nonsensical. An antitrust market must consist of products or services, not customers. Presumably, what Judge Brown meant was that a grouping of products for which there is “a core group of particularly dedicated, distinct customers, paying distinct prices, may constitute a recognizable submarket….” This is maybe a little picky, but it’s important to emphasize that antitrust markets consist of groups of products. (Market power emerges, then, when a seller controls so many of those products that its output decision affects the market-clearing price. If the seller can so affect price by varying its output, then it may be able to enhance its profits by reducing its output and raising price.)
A second initial point is that the existence of a particularly dedicated, core group of consumers should never, by itself, give rise to a separate antitrust market. Practically every brand of product has some dedicated consumers who would not switch away from the brand in response to even a sizable price increase. Some Starbucks devotees, for example, simply won’t go elsewhere and would be willing to pay much more than current prices for their Starbucks fix. Does the existence of a “core group of particularly dedicated, distinct customers” mean there is a distinct antitrust market consisting of Starbucks coffee? Of course not.
If the existence of a core group of “high-reservation price” consumers (i.e., those with a high willingness-to-pay) were enough to give rise to a separate market, then antitrust markets would proliferate, and monopoly power would be ubiquitous. That would transform antitrust prosecutors and plaintiffs into the all-purpose policemen of the business world, for any firm with monopoly power runs afoul of the Sherman Act if it engages in difficult-to-define “exclusionary conduct.” Freed from having to make any sort of rigorous showing of market power, antitrusters could go around attacking all sorts of vigorous competition, labeling it exclusionary conduct. Given the vagueness of that concept, they’d likely achieve a number of successes.
To her credit, Judge Brown did not say that the existence of a core group of consumers of a particular product is sufficient to make that product a separate market. In stating that those consumers must “pay[] distinct prices,” she apparently recognized that the seller must be able to price discriminate against those core consumers — that is, to identify them in advance and to charge them a higher price (reflective of their greater willingness-to-pay), while preventing arbitrage.
That’s the theory that underlies the Horizontal Merger Guidelines’ recognition of antitrust markets consisting of products attractive to core consumers. According to the Guidelines:
If a hypothetical monopolist can identify and price differently to those buyers (“targeted buyers”) who would not defeat the targeted price increase by substituting to other products in response to a “small but significant and nontransitory” price increase for the relevant product, and if other buyers likely would not purchase the relevant product and resell to targeted buyers, then a hypothetical monopolist would profitably impose a discriminatory price increase on sales to targeted buyers. This is true regardless of whether a general increase in price would cause such significant substitution that the price increase would not be profitable. The Agency will consider additional relevant product markets consisting of a particular use or uses by groups of buyers of the product for which a hypothetical monopolist would profitably and separately impose at least a “small but significant and nontransitory” increase in price.
So in order for “products uniquely attractive to core consumers” to constitute a distinct market, the seller must be able to price-discriminate against those consumers. How could that work in the purported PNOS market?
Judge Brown reasoned that such price discrimination could occur because Whole Foods priced its products in such a way that high-reservation price consumers ultimately paid a higher price than its low-reservation price consumers. (At least, I think that’s the reasoning…Judge Brown’s opinion is nearly inscrutable on this point.) Here’s the purported evidence of such price discrimination:
Whole Foods and Wild Oats charge higher mark-ups on perishables than on dry groceries.
Some consumers (presumably those with the highest reservation prices) buy a greater proportion of perishables from Whole Foods and Wild Oats than do other consumers.
Thus, the former group of “core consumers” end up paying higher prices for their Whole Foods/Wild Oats “experience” than do non-core, marginal consumers.
Voila — price discrimination!
This is silly. The price discrimination needed to support the sort of market power exercise envisioned by the Merger Guidelines requires charging different prices (technically, different increments above marginal cost) to different consumers for the same good. The “price discrimination” to which Judge Brown refers involves charging different prices for different baskets of goods. Price-disadvantaged Whole Foods customers, those who buy a higher percentage of perishables, are paying higher prices because they’re getting something different. Moreover, there’s no reason to suppose that the percentage of perishables a consumer purchases corresponds to his willingness-to-pay for the Whole Foods experience. Many high reservation price consumers may buy low percentages of fruits and vegetables.
The “price discrimination” to which Judge Brown refers — the only sort of “price discrimination” supported in the record, as far as I can tell — is akin to the sort of “metering” price discrimination to which Josh refers in this post: a seller with market power over one product (the tying product — e.g., a copy machine) charges something below his profit-maximizing price for that single product but somehow ties in a second complementary product (the tied product — e.g., ink) and charges an above-cost price for that product, thus earning higher margins from heavy users (presumably high reservation price consumers) than from light ones (presumably low reservation price consumers). That’s a sort of price discrimination, but it’s not what was happening at Whole Foods. No one was required to buy any produce in order to enter (and/or in order to buy dry grocery items from) Whole Foods. Thus, the record included no evidence of price discrimination, and Judge Brown therefore erred in concluding that preferences of core consumers could define the market at issue.
(Geoff has a great post discussing additional problems with Judge Brown’s opinion.)
Problems With Judge Tatel’s Opinion
Judge Tatel originally joined Judge Brown’s opinion, but his amended opinion issued in late November quite correctly declines to endorse her misguided core consumer analysis. Instead, Judge Tatel reasons that reversal is appropriate because “the district court overlooked or mistakenly rejected evidence supporting the FTC’s view that Whole Foods and Wild Oats occupy a separate market of ‘premium natural and organic supermarkets.'” Judge Tatel has greatly overstated the strength of the FTC’s evidence in favor of a separate PNOS market, and his willingness to rely on non-pricing evidence — particularly “hot” business documents — sets a troubling precedent.
Here’s the evidence Judge Tatel cited and an explanation for why each piece fails to support a separate PNOS market:
Evidence: Judge Tatel first referenced a report by the FTC’s expert showing that when a Whole Foods opened hear a Wild Oats, sales at the Wild Oats dropped substantially more than they did when a conventional supermarket opened nearby. He then pointed to an internal Whole Foods study predicting that the closure of Wild Oats stores near existing Whole Foods markets would drive most former Wild Oats customers to the Whole Foods outlets rather than to conventional grocery stores.
Problems: This evidence establishes merely that Whole Foods and Wild Oats are particularly close substitutes for each other; it in no way suggests that conventional grocery stores aren’t also viable substitutes whose existence constrains Whole Foods’ and Wild Oats’ pricing, and that’s the key question when it comes to market definition. The fact that a higher percentage of Toyota Camry drivers would switch to Honda Accords than to Nissan Altimas or Volkswagen Jettas does not mean that Camrys and Accords occupy a separate antitrust market from Altimas and Jettas.
Evidence: Judge Tatel then pointed to instances where marketing experts and/or consultants (1) “distinguish[ed] between ‘traditional’ or ‘conventional’ grocery stores on the one hand and ‘natural food’ or ‘organic’ stores on the other”; (2) predicted that conventional grocers’ (e.g., Wal-Mart’s) entry into organic would not substantially enhance competition with Whole Foods; and (3) observed that, while most Whole Foods/Wild Oats consumers also “cross-shop” at conventional grocery stores, “they tend to shop at each for different things (e.g., Wild Oats for fresh and specialty items, Safeway for canned and packaged goods).”
Problems: These sorts of qualitative statements by marketing types, unsupported by pricing evidence, say almost nothing about the question of whether Whole Foods and Wild Oats are so unique that they constitute a separate economic market. (1) The fact that marketing folks distinguish between conventional and natural/organic grocery stores is unsurprising; the entire marketing enterprise is focused on discovering ways for businesses to distinguish themselves from their rivals (and thereby to win the loyalty of core customers). Moreover, the Food Marketing Institute, which Judge Tatel cites as distinguishing between conventional and organic stores, sometimes lumps them together, as in this report, which includes Whole Foods (#16) on the list of top U.S. Supermarket and Grocery Chains (along with Safeway, Albertson’s, Winn-Dixie, etc.). Categorizations by marketers are simply not probative of cross-elasticities of demand. (2) The prediction that Wal-Mart’s and Costco’s entry into organics wouldn’t enhance competition with Whole Foods doesn’t say anything about the competition that already existed. More importantly, this is simply a before-the-fact prediction, not a fact supported by data concerning actual competition (especially pricing data). (3) The fact that cross-shoppers buy different things at PNOS stores versus conventional grocers says nothing about the ability of a PNOS monopolist (i.e., a combined Whole Foods/Wild Oats) to raise prices above competitive levels. Given the existence of widespread cross-shopping, which Judge Tatel concedes, such a price increase would likely prove unprofitable as cross-shopping consumers would just start buying the price-enhanced item on their conventional grocery store trip rather than on their PNOS store visit.
Evidence: Next, Judge Tatel referenced various remarks that, he concluded, evince “industry or public recognition” of a separate market for PNOS stores. Specifically, he pointed to (1) an email to Whole Foods’ board from its CEO stating that “[Wild Oats] is the only existing company that has the brand and the number of stores to be a meaningful springboard for another player to get into this space”; (2) a statement by Wild Oats’ former CEO that “there’s really only two players of any substance in the organic and all natural [market], and that’s Whole Foods and Wild Oats…. There’s really nobody else in that particular space”; and (3) statements from executives of conventional grocery stores that Whole Foods and Wild Oats are not “conventional supermarkets” because “they focus on a premium organic-type customer” and “don’t sell a lot of the things that … a lot of people buy.”
Problems: While the Supreme Court’s unfortunate Brown Shoe decision did state that such “practical indicia” as “industry or public recognition” can define a market, a point Judge Tatel notes, this is awfully flimsy evidence of market definition. (1) The email from CEO Mackey, sent to the board just before the vote on the Wild Oats merger, was designed to drum up support for the merger and was in no way a thoughtful, reflective statement about market contours. This is the classic sort of “hot doc” whose limitation Geoff highlights in his terrific article, Hot Docs vs. Cold Economics. (2) Similarly, the Wild Oats’ CEO’s reference to the “space” uniquely occupied by Whole Foods and Wild Oats (the term “market” was inserted by Judge Tatel, not the CEO himself) doesn’t suggest that Whole Foods and Wild Oats really occupy a separate economic market. “Space,” the term also used by CEO Mackey in the aforementioned email, does not equate to “market.” (3) The fact that Whole Foods and Wild Oats have chosen to “focus on a premium organic-type consumer” just shows that they are pursuing a different market segment, not that they’re in a separate market. And the fact that their selection is more limited than conventional grocers couldn’t prevent the wider-selection conventionals from competing with them.
Evidence: Judge Tatel then pointed to a “peculiar characteristic” of Whole Foods and Wild Oats that, he said, must remove them from the market in which conventional supermarkets compete: “[E]verything Whole Foods sells is natural and/or organic, while many of the things sold by traditional grocery stores are not.”
Problems: This is one of the points raised in the third piece of evidence highlighted in the last section, and the same deficiency applies: The fact that PNOS stores have a more limited selection than conventional supermarkets does not preclude conventionals from competing with (i.e., constraining the pricing of) PNOS stores, though it could prevent PNOS stores from constraining the conventionals’ pricing on products not carried by the PNOS stores.
Evidence: Finally, after going through all this non-probative and/or unpersuasive qualitative evidence, Judge Tatel comes around to the stuff that matters — pricing evidence. He’s OK with relegating the pricing data to the back burner because Brown Shoe listed pricing evidence as “only one of a non-exhaustive list of seven ‘practical indicia’ that may be examined to determine whether a separate market exists.” He points to the fact that the opening of PNOS “Earth Fare” stores near Whole Foods stores in North Carolina caused the nearby Whole Foods stores’ prices to drop about 5% below the prices at other North Carolina Whole Foods. This fact, he says, shows that PNOS stores can constrain Whole Foods’ pricing in a way that conventional grocers cannot do, indicating that PNOS stores constitute a separate market.
Problems: While this pricing evidence is at least probative (unlike the other evidence cited by Judge Tatel), it’s less persuasive than he suggests. As Judge Kavanaugh’s dissenting opinion details, Whole Foods’ price cuts following the entry of Earth Fare stores in North Carolina were short-lived: “[S]oon after those entries, Whole Foods’ prices returned to normal levels. So the record hardly shows the sort of ‘nontransitory’ price changes that are the touchstone of product-market definition.” (Under the prevailing “SSNIP” test, a market is defined to include the smallest grouping of products for which a hypothetical single seller could profitably impose a “small but significant non-transitory increase in price.”) Judge Kavanaugh also observes that “the entry of a Safeway store in Boulder, Colorado, had a similar short-term impact on Whole Foods, indicating that whatever inference should be drawn from the Earth Fare entries cannot be limited to so-called organic supermarkets but rather applies to conventional supermarkets.”
At base, Judge Tatel’s reasoning is infirm because he, like the FTC, relied primarily on non-pricing evidence — hot emails, cherry-picked marketing studies, inapposite diversion ratios, etc. — to define the contours of the market. Sadly, there’s some “official” — though largely discredited — Supreme Court teaching that permits this sort of squishy analysis. I’m speaking, of course, of the highly unfortunate Brown Shoe decision, which both Judge Tatel and Judge Brown cited extensively. In that decision, the Supreme Court stated that antitrust “submarkets” could be defined, in part, according to such “practical indicia” as “industry or public recognition of the []market as a separate economic entity, the product’s peculiar characteristics and uses, unique production facilities, distinct customers, distinct prices, sensitivity to price changes, and specialized vendors.” This unfortunate statement has invited litigants and regulators to rely on common hunches (especially with respect to distinct distribution channels for widely available products) and to scour business documents for market characterizations that suit their end.
Unfortunately, it’s really difficult to get a merger challenge up to the Supreme Court. When regulators lose, they generally don’t appeal because mergers usually close shortly after the district court rules, and most consummated mergers are quite difficult to undo. When the parties to a merger agreement lose, they usually give up because they know they can’t hold the merger agreement together for the duration of an appeal. The result has been a dearth of Supreme Court merger decisions (as noted, the last significant one was decided in 1974). With Whole Foods, the FTC went ahead and appealed because it concluded that the combined company’s decision to operate Wild Oats stores separately for some period of time would avoid the need to engage in a messy disentangling of the merged company should the agency ultimately prevail.
When Judge Brown’s crazy “core consumer” argument was endorsed by the panel majority in July, I was pretty excited. The Supreme Court would have to grant cert to correct the nutty reasoning in Judge Brown’s opinion, I figured, and it could take the opportunity to clean up the mess made by Brown Shoe‘s invitation to define markets on the basis of “practical indicia” — i.e., whatever evidence supports your preferred view.
Alas, Supreme Court adjudication now seems highly unlikely. I’m disappointed.
But, again, I’m a “glass half-full” type, so I’ll just think about how fun it’ll be to teach this case in a few years, given that most of the so-called “conventional” grocery chains (who have realized that they can’t compete with Wal-Mart on price) are moving headlong in the direction of a Whole Foods/Wild Oats format. (See, for example, this Safeway store and this article, which, ironically, appeared in the Wall Street Journal the same day as the article announcing the FTC’s decision to challenge the Whole Foods/Wild Oats merger). A separate PNOS market, my ass.