Don’t Punish Companies for Being the Best: Part Deux
According to press reports, both the U.S Attorney General and most state attorneys general will file one or more antitrust cases against Google sometime this summer. If true, the United States would be going down a path already well-trod by the European Commission, which has conducted three antitrust cases against the company. The speculation is that the U.S. suits will focus on Google’s monopoly and anticompetitive actions in the market for search and/or for Internet advertising.
In May the Omidyar Network released a report by Fiona Scott Morton and David Dinielli, both of whom served as antitrust authorities in the Obama administration, that looked at the case regarding Internet advertising. While the report listed several Google practices that might have harmed competitors, ITIF’s analysis concluded that these practices did not clearly demonstrate any harm to consumers.
The authors recently have followed up with a new report looking at the search market. It also concludes that Google has a monopoly in the relevant markets and has maintained and benefitted from this dominance while harming other companies, partly through actions that the report alleges are anticompetitive.
The authors start with two important caveats. The first is that they rely on information from a few sources, the main one being an interim report by the UK Competition & Markets Authority (CMA) which is investigating Google’s practices in the UK. They also rely on the European Commission’s 2017 ruling against Google regarding how its search engine handled comparison-shopping sites. The authors admit that, to the extent that either the U.S. market or Google’s activity differ from that in the UK, their analysis will differ. They also clearly endorse the need for a detailed economic study of the specific market in question focusing on harm to customers.
Continued vigilance is necessary in order to ensure markets remain competitive but antitrust authorities should base their decisions on a clear showing of consumer harm. It is impossible to judge whether any of the conduct discussed in the report amounts to an antitrust violation until a thorough analysis is performed. However, regulators should reject the notion that they should break up companies just because they are big. Doing so would often prevent firms from gaining the scale they need in order to maximize efficiency and innovation and reduce prices. These concerns are especially relevant to digital platform models, which exhibit high fixed costs, rapidly falling marginal prices, and significant network effects. As the Obama administration Council of Economic Advisors wrote, “Markets in which network effects are important… may come to be dominated by one firm, because the ‘network externalities’ in these markets tip to one provider of the network product or service.” In this case, a key question is would consumers be better off with the current market structure or one where there are 3 major search providers with just one-third of the market each. Unlike some industries, such as autos and banking, where scale and network effects are nowhere near as large and where an industry with such a large market concentration would almost certainly be anti-consumer, it’s not at all clear that the current structure is not the optimal one.
Internet search is perhaps the most valuable platform service most people use. In a recent poll, Americans said that they would need to get paid $17,530 to give up search engines for one year. Instead, they get it for free. Search engines are constantly refining their techniques in order to improve the results they show users. The accuracy of results is closely tied to the popularity of search engines, and therefore the number of users they have and the amount of revenue they can derive from advertisers.
Morton and Dinielli point out that search engines benefit from scale. The more searches they make, the more they can use information about which websites are chosen to improve their results over time. However, the exact size of this advantage is unknown. A recent study showed that significantly reducing the period for which searches are kept (in one example from 13 months to 3 months) had no noticeable effect on the quality of future search results. Still, the Omidyar paper notes that search engines may have an incentive to deny their competitors volume, thus preventing them from gaining whatever threshold of market share is needed to effectively compete. It should also be noted that in a fair portion of cases prior searches are of limited use. The study estimates 15 percent of searches are unique.
The report describes four practices Google allegedly used to obtain and keep its dominant position:
- Obtaining default positions, some of them exclusive, on home screens and browsers.
- Requiring manufacturers who use its Android operating system to disadvantage or exclude rival search engines.
- Acquiring companies providing inputs to its rivals that might have allowed those rivals to attract users.
- Manipulating its search results to harm specialized search engines such as Bookings.com and Yelp by lowering their search results.
To be clear, these are assertions that any investigation would try to validate. The authors do not allege that these practices resulted in Google’s monopoly position. Each of these activities would be legal if done by a company that lacked market power. Antitrust law is also clear that companies are allowed to acquire market power through legitimate competition. Morton and Dinielli argue that, once Google acquired its dominant position, it was forbidden to continue these practices in order to maintain its control over the market.
Morton and Dinielli document Google’s dominance in general search. Google currently has over 90 percent of the UK market. The only company Google identified as a competitor to it, Microsoft’s Bing, has less than 10 percent. Google’s share of the mobile search market is 96 percent. Much of this advantage comes from investments in extending its reach. Google developed the Android operating system to extend its presence in the mobile market. It has convinced 88 percent of UK websites to carry its “tags,” such as Google Analytics, which allow it to gather information about visitors. It also developed or acquired tools such as Google Maps and Waze to feed it location data, all of which improve its search results. But at least one of these acquisitions dramatically boosted innovation and consumer welfare, taking what was a limited and costly service and making it universal and free.
Google also faces competition from other ways of finding a webpage. Regulators sometimes focus too much on market share for desktop or mobile search. But search isn’t just one thing. People use many tools to find local businesses. Some are doing this on their phone (or even in their car by using specific apps including Facebook, Yelp, TripAdvisor, Apple Maps, Alexa, Siri, FourSquare, etc. or even an in-car infotainment system. When searching for movies, rather than go to Google, users might just start on Netflix or Amazon or YouTube, etc. There are lots of different types of search, and regulators can focus too much on “traditional search engine” without understanding what percent of search that actually represents or how that market shifts in comparison to other search markets.
The report discusses remedies only briefly. However, it does suggest a few changes that might form the basis for a negotiated agreement. These changes might result in more competition, even if they do not allow a competitor to overcome Google’s quality advantage. For instance, in auctions for its ad space, prices are determined by the second highest bid. However, Google makes a “quality adjustment” to prevent companies from gaming the system. Google makes these adjustments on its own without explaining them to advertisers. They recommend making the process more transparent. They also propose giving advertisers more information about who is seeing their ads. Also, if it is actually true that Google has changed the results of its organic search based on the type of website or whether it buys ads from Google, then regulators could take action. They also propose that regulators prevent Google from insisting on a default or exclusive position with a traffic generator in favor of showing users a screen prompting them to make a choice (users can still download a search engine from an app store). They also propose obtaining an agreement not to use software updates to undo a user’s decision to set another engine as the default search choice.
But in evaluating each of these changes, antitrust officials should first ask whether doing so would benefit consumers. The primary purpose should not be to protect other companies from legitimate and fair conduct. Morton and Dinielli argue that any action, including actions that normally represent fair competition, that hurts other search engines, handset manufacturers, advertisers, content providers, or specialized search engines necessarily hurts consumers because it results in higher prices, less choice, and slower innovation. Regulators need to check that this is true.
Left undetermined is whether any of these changes would produce a better search engine. Giving Bing and new entrants more volume may not matter very much if the true source of advantage is that Google just has better algorithms and a better learning capacity. When the authors state that “consumers view Google’s search results, on the whole, to be more relevant than Bing’s, which reflects Google’s tremendous scale advantage,” they may assign too much importance to scale. On the other hand, shifting meaningful market share over to competitors could reduce Google’s ability or incentive to make the investments needed to maintain its quality advantage. This would definitely hurt users.
One thing that should not matter is Google’s size. The Omidyar report states that Google’s scale is a barrier to entry. This is like saying that Ford’s size is a barrier to entering the car market. It is, but only because technology requires large volumes in order to lower the average price of making a car. It is not too surprising that, up to a limit, the same is true of search engines. And as the Obama administration noted, search engines also benefit from network effects. Regulators should respect that.
