The unfortunate return of the "strange, red-haired, bearded, one-eyed, man with a limp"

Geoffrey Manne —  29 July 2008

The DC Circuit has reversed the district court in the Whole Foods case.  The opinion is here.  [HT:  Danny Sokol]

As regular readers know, we have covered this case extensively on this blog, including most recently this great, lengthy post from Thom on the proper standard of review.  I wouldn’t be surprised if Thom is disappointed with the standard adopted by the DC Circuit in its appeal, and I look forward to his thoughts.

I’m sure all of us will have more to say about the case in due course.  For now I want to highlight one incredible aspect of the decision:  The claim that relevant market definition can turn on average or “core” customer effects.  Yes, throw out 30-odd years of antitrust economics and the very concept of the marginal consumer.  Here’s the court’s theory:  If a bunch of inframarginal consumers really like products X and Y but not similar-but-slightly-different-and-cheaper product Z, a merger between X and Y would enable the combined firm to gouge the inframarginal consumers, regardless of the effect on the marginal folks.  Sure the marginal guys will shop at Safeway if the price rises too much–but who cares?  There’s always a “core” of super premium and organic supermaket addicts to take advantage of–folks who just can’t resist all those free samples and faux-wood floors! 

OK.  Except–correct me if I’m wrong–if price discrimination were possible before the merger as well as after (and there’s no reason why this would change), the core folks were already being gouged!  Effective product differentiation may make the market price higher for a subset of goods–but this higher market price would prevail even with lots of competition. 

The court here claims that the real damage will occur in prices of perishables, because organic produce bought at a Whole Foods is qualitatively different than organic produce bought at Safeway, and this is what the core shoppers bought at the premium stores.  But if that were true post-merger, it would be tru pre-merger, as well, right?  There aren’t any merger-specific effects that I am aware of.  Again–correct me if I’m wrong–but I don’t think any of the evidence in the case suggested that there could be a significant non-transitory effect on the price of “PNO” produce.  Yes, some evidence suggests produce prices are higher at PNOS’s than at typical grocery stores, but is there evidence that the price could be even higher if Wild Oats exited a market?  The concurrence points to the FTC’s supposition that enough people would be diverted from Wild Oats to Whole Foods to sustain a 5% price increase, but I’m not sure that there was any evidence to support this claim (particularly since the FTC’s economist didn’t look at prices, but rather at profit margins).

Manwhile, if price discrimination is not possible, way more than 30 years of economics tells us that the seller will price to the marginal consumer–the ones who would decamp to Product Z if prices rise too much.  The claim here, however, is that, even if evidence indicates that Whole Foods and Wild Oats compete not only with each other but also many other stores, they are each other’s only competitor for some consumers.  Yes, Virginia, there are inframarginal consumers.  So what?  Did the court forget its Econ 101?  Profit is maximized where marginal revenue equals marginal cost.  Absent price discrimination, it’s a bummer the seller can’t capture the surplus from all those inframarginal shoppers, but profit is still maximized by pricing to the marignal consumer.  There is no reason (other than a new ability to price discriminate) that the post-merger entity would sacrifice profits by jettisoning the marginal shopper just to gouge its regular clientelle.   

Perhaps even more fundamentally–as we have pointed out on this blog before–the differential effect via price discrimination story (if true) suggests a far, far simpler solution:  Better relevant market definition.  If the inframarginal consumers at the grocery store are consuming radically different products than the marginal consumers, perhaps the relevant market is not “premium natural and organic supermarkets” but rather “organic foods,” for example.  In other words, perhaps the relevant market is the products being consumed, not the channel of distribution.  But I would guess that this market definition would have condemned the case from the start given Walmart’s extensive entry into this market, so it wasn’t in the cards. 

“But wait!,” you say.  Some people have idiosyncratic preferences.  They preferer buying organic tomatoes, zucchini and grapes from premium natural stores–it’s a combination, you see, not only of the food being consumed but also the channel of distribution.  These poor sots will be gouged without competition between Whole Foods and Wild Oats, because they don’t want to shop for produce at Safeway.  And Whole Foods without Wild Oats would easily overcharge these 17 or 18 people in any given market.

Yes, indeed.  One can always define a market by focusing on idiosyncratic preferences or product variations.  This is what Justice Fortas decried in his dissent in Grinnell:  

The trial court’s definition of the “product” market even more dramatically demonstrates that its action has been Procrustean – that it has tailored the market to the dimensions of the defendants. It recognizes that a person seeking protective services has many alternative sources. It lists “watchmen, watchdogs, automatic proprietary systems confined to one site, (often, but not always,) alarm systems connected with some local police or fire station, often unaccredited CSPS [central station protective services], and often accredited CSPS.” The court finds that even in the same city a single customer seeking protection for several premises may “exercise its option” differently for different locations. It may choose accredited CSPS for one of its locations and a different type of service for another.

But the court isolates from all of these alternatives only those services in which defendants engage. It eliminates all of the alternative sources despite its conscientious enumeration of them. Its definition of the “relevant market” is not merely confined to “central station” protective services, but to those central station protective services which are “accredited” by insurance companies.

There is no pretense that these furnish peculiar services for which there is no alternative in the market place, on either a price or a functional basis. The court relies solely upon its finding that the services offered by accredited central stations are of better quality, and upon its conclusion that the insurance companies tend to give “noticeably larger” discounts to policyholders who use accredited central station protective services. This Court now approves this strange red-haired, bearded, one-eyed man-with-a-limp classification.

Tailoring market definition to inframarginal consumers who may be willing to pay more than market prices for certain product characteristics is neither sound economics nor sound antitrust doctrine.

Stay tuned.

UPDATE:  A great post from Manfred Gabriel at Antitrust Review makes much the same case.  A couple of really good parts:

Maybe I shouldn’t be surprised by this. But the passage does more than accept that, given the difficulties of economic analysis, we can supplement it with some common-sense heuristic insights to come up with markets. This approach seems to undermine the acceptance of the theoretical foundations of economic analysis. 

* * *

I read this to mean that the “practical indicia” have become the trump card to beat economic analysis.

* * *

The only economic evidence of price discrimination between core and marginal customers cited in the opinion is the fact that Whole Foods stores enjoyed lower margins (not prices!) in cities where they competed with Wild Oats stores. One would have hoped for a discussion of how Whole Foods identifies core customers (the ones in Birkenstocks, perhaps?) and manages to charge them higher prices than the marginal customers (in wingtips and high heels).

Geoffrey Manne


2 responses to The unfortunate return of the "strange, red-haired, bearded, one-eyed, man with a limp"


    Terrific post, Geoff. The implications of the majority’s willingness to define a market based on the likely conduct of core customers is truly astounding. Given that market definition is a key part of the analysis in many antitrust disputes, this bad precedent has the potential to reach far beyond the merger context.

    Consider monopolization, where one of the elements is monopoly power (generally determined based on market share). If we can define markets based on the degree to which core customers would refrain from switching to other products in response to a price increase, then basically every brand with a devoted following is a monopoly! The market consists of it alone, since core customers would not switch in response to a higher price, and the producer is the single seller within that market — a textbook monopolist.

    Take Starbucks, where I’m sitting as I write this. There are many, many people (I am not one of them, by the way), who have a MUCH stronger preference for Starbucks coffee (and perhaps the whole Starbucks experience) than the cofee (and coffee experience) at any other coffeeshop. Starbucks would not lose their business if it raised its prices substantially. Does that make Starbucks a monopolist? Surely not.

    If inframarginal preferences define markets, leading to an abundance of “market power” for antitrust purposes, then antitrust prosecutors and plaintiffs will become the all-purpose policemen of the business world. That’s because any firm with monopoly power (and there will be lots of such firms under the D.C. Circuit’s approach), runs afoul of the antitrust laws if it engages in unreasonably exclusionary conduct, which is notoriously difficult to define. Freed from having to make any sort of rigorous showing of market power, antitrusters can go around attacking all sorts of vigorous competition, labeling it “exclusionary conduct.” Given the vagueness of that concept, they’re likely to have a number of successes. To borrow another classic description, antitrust will once again be, as Robert Bork once described it, “in the good old American tradition of the sheriff of a frontier town: he did not sift evidence, distinguish between suspects, and solve crimes, but merely walked down main street and every so often pistol-whipped a few people.”

Trackbacks and Pingbacks:

  1. TRUTH ON THE MARKET » Varney on the Merger Guidelines - January 27, 2010

    […] Taken literally, what Varney is saying is that an ad hoc (ok, fine–an “integrated, fact-driven”) determination that some customers (”vulnerable” ones, whatever that means) may be made worse off by a merger lessens the need for a more comprehensive assessment of overall competitive dynamics within a relevant market.  But I don’t know what this means, frankly.  In the first place, how is the agency supposed to know that some customers are likely to be harmed if it hasn’t assessed the availability of substitutes and the extent of diversion?  One can certainly criticize the method by which this assessment is made, but a conclusion of harm absent this assessment seems absurd.  Moreover, if Varney really means that all that is required to condemn a merger is that any customers may be harmed, no matter how many are also benefited, at a minimum it sounds like she’s writing the efficiencies defense out of the Guidelines, but she may even be justifying condemnation of any and all mergers–after all, how many actions in the marketplace impose a cost on literally no one?  If, as seems likely, it is inframarginal consumers who are “likely” “vulnerable” to price increases (where “vulnerable” may be a synonym for “having inelastic demand”), then this test is a repudiation of the entire economic edifice of modern merger analysis (parallel to my discussion of the DC Circuit’s Whole Foods decision here). […]