[This series of posts dissects the threatened FTC antitrust case against Google and concludes that a monopolization prosecution by the federal government would be a very bad idea. We divide the topic into five parts, one policy and four legal. Check out Part I, Part II, Part III, Part IV, Part V. Check out the whole series as one long post.]
Section 2 of the 1890 Sherman Act (15 U.S.C. § 2) makes “monopolization” unlawful. As every antitrust practitioner can recite by heart, this means that being a monopoly is not illegal, rather it is illegal to obtain or maintain monopoly power in a “relevant market” by exclusionary or anticompetitive means.
The most famous articulation of this basic principle comes from the case of United States v. Grinnell Corp. (“Grinnell“), 384 U.S. 563 (1966), in which the U.S. Supreme Court explained that a monopoly position reached as a result of a “superior product, business acumen or historic accident” is different from one achieved by the “willful acquisition or maintenance of that power.” That slightly schizophrenic approach reflects the basic conflict within antitrust itself. The law encourages, and permits, firms with market power (typically a synonym for monopoly power, although economists disagree at the margins) to compete aggressively on the merits, and even to eliminate competitors. Yet to tame the results of unbridled capitalism, Section 2 constrains companies from creating or defending monopoly power with anticompetitive practices.
2. Internet Search and Search Advertising are Not Relevant Antitrust Markets
The starting point for every antitrust case is market definition — outlining the contours of a market, in which the defendant participates, in order to assess whether the firm possesses monopoly power in that market. In defining the relevant antitrust market, courts determine which products compete with the defendant’s product and thus limit or prevent the exercise of market power. Typically, this process involves examining substitutability of products (both from a demand and a supply perspective) to find whether consumers and rivals could switch to another source (or sources) if the defendant firm were to raise price or restrict output. For example, in the 1950s chemical innovator duPont was charged with monopolizing the cellophane market, a product it invented, but the courts ruled that the relevant antitrust market could not be so narrowly limited because cellophane was interchangeable with other food wrapping materials. The “great sensitivity of customers in the flexible packaging markets to price or quality changes” prevented duPont from exerting monopoly control over price.
The more broadly the relevant antitrust market is defined, the less likely it is the defendant has the ability to exercise monopoly power in that market. As a corollary, if the targeted firm does not have monopoly power in the relevant market, there generally cannot be Section 2 liability. Many recent antitrust cases, including the FTC’s controversial attempt to block Whole Foods’ acquisition of Wild Oats and the Justice Department’s challenge to the Oracle-PeopleSoft merger, have turned on market definition.
With that background, let’s look at the purported “Internet search” market. That’s obviously the core proposition in any attack on Google for unlawful monopolization, because the necessary premise is that Google’s dominant share — estimated at from 65 to 80% — of Web searches is the foundation of its alleged monopoly. But here the antitrust analysis begins to break down. Internet search is a free product in which the consumers (Internet users) are charged nothing, with the service supported by advertising revenues. Since monopoly power is the “power to control price or exclude competition,” one must necessarily ask whether Google’s high “market share” reflects any market power at all. More importantly, search users are just like broadcast television viewers; they are an input into a different product — search advertising — in which consumers themselves are effectively sold by virtue of advertising rates based largely on impressions and click-throughs. Just as NBC, ABC, CBS and Fox compete for television eyeballs in order to sell more advertising (hence profiting) to sponsors, so too do Internet search engines monetize the service by selling eyeballs to advertisers.
Google’s share of search by itself is therefore almost meaningless. Even if the relevant market is confined to search, moreover, there is nothing that enables Google to prevent users from switching, instantaneously, to another of the scores of search engine providers on the Internet. (It should go without saying that even the government does not contend that Google displaced Yahoo!, Alta Vista, Ask.com and the many former search giants that dominated the Internet in the 1990s with anything other than better, more useful, search results, a consequence of better algorithms — the epitome of Grinnell’s “superior product.”) So the relevant market analysis must therefore focus on the area where Google in fact competes with other search engine providers, namely in the sale of search advertising. We all know that the links displayed alongside so-called “organic” search results are paid, listed conspicuously as “sponsored” results. Without search advertising, in today’s Internet economy there would be no free search engine services.
Whatever Google’s share of Internet search advertising, the relevant market cannot be so limited. First, Web search ads compete with display (e.g., banner) ads. The choice of whether to advertise with a graphical display ad or a text search ad depends on relative prices, budgets, reach and attractiveness. Because advertisers are looking and pay for users who click through the ads to their sites, both represent alternative and plainly substitutable ways to reach customers; if Google somehow raised prices for search ads, its own customers would and could switch more of their advertising dollars to display ads. And the display ad segment is something in which Google has for a “long” time lagged well behind the current leader, Facebook — interestingly a company that just four years ago had a minuscule presence in display advertising compared to Yahoo!, AOL and MySpace. Something on the order of 28% of all Internet display advertising took place within Facebook as of 2011. Google is fighting back, which in itself indicates how interchangeable search advertising and display advertising are on the Web today.
Second, both search and display ads increasingly compete against mobile search ads. Here the nascent but very rapidly growing market is radically different, with searches designed to retrieve more immediate, targeted results and in which a developing majority of searches are performed within smartphone and tablet apps like Yelp!, FourSquare, OpenTable and others.
Third, all advertising-supported Internet services in reality compete with traditional media (television, radio, newspapers and magazines) for advertising revenues. The New York Times and other papers, for instance, have lost huge swaths of advertising revenues — especially, but hardly just, classified ads to Craigslist, eBay and Match.com listings — that threaten their very existence, but are making money in digital Web advertising. Nearly 1/3 of the Times total revenue came from online advertising as far back as 2010. The point is not whether digital technology and Internet-centric business models can save the newspaper industry as we know it from disruptive innovation or disintermediation, but rather that the search ads Google sells are substitutes for traditional advertising. As Peter Juke of The Daily Beast observed just days ago:
For a 60-year period, which peaked in 2000 with income of roughly $60 billion a year, print advertising effectively funded journalism. But last year, revenues dropped to around $19 billion. Where has that $40 billion gone? Here’s a suggestion: last year Google generated $37.9 billion in revenue, 96% of which came from advertising. The bulk of these ad dollars come from traditional finance, retail, travel and education companies that — presumably — would have otherwise advertised in the dead-tree press. So far, so good for Google: if it can deliver advertising more efficiently and more directly to the consumer, it deserves to win this race.
Taken together, all of these factors suggest strongly that the relevant antitrust market for assessing Google’s alleged monopoly power cannot properly be narrowed to Internet search advertising, and likely not even to Internet advertising to the exclusion of legacy advertising media. In an “Internet advertising” market Google’s share is almost certainly well below the 70-80% required as the minimum from which to infer monopoly power. (The analytical steps from market share to a finding of monopoly power depend on other factors, which we examine in Part III). In an “advertising” market Google undoubtedly has a share hardly worth worrying about.
All said, if as it seems the FTC’s monopolization case against Google were premised on an Internet search advertising market, that market definition — assuming the actual data, difficult to assemble without subpoenas, are consistent with the metrics cited here — will probably receive the same negative reaction from the courts so many government-proposed relevant antitrust markets have in the past.