Search

Opinion

How to Spot a Monopolistic Search Engine

Shortly after the Federal Trade Commission (FTC) closed its investigation of Google’s use of monopoly power three years ago (deciding at the time not to take action against Google), the European Commission ramped up its parallel inquiry (and has now decided to take action against Google). The issue at hand is whether Google should be allowed to use its monopoly power in search to favor its other products such as Google Shopping, Google Reviews, and Google Flights. If this was helpful to customers, then regulators generally think that Google should be allowed to do this. However, if harm is being done to customers, then regulators might want to limit Google’s use of its monopoly status.

The problem is that understanding “harm” is difficult in the digital economy. In contrast with standard monopoly cases where regulators can look for signals like high prices, the price tag is not directly paid by consumers in this context. This complicates antitrust analyses. But the lack of price tag doesn’t mean that consumers aren’t harmed. After all, Google is making a lot of money through their search product, and harm can come in the form of an inferior recommendation for customers.

In recent years, new evidence and tools have emerged helping economists to better understand the impact of Google’s practices on consumer welfare. Research that several of my colleagues and I have done sheds light on Google’s exclusionary practices relating to Google Reviews (which is Google’s proprietary review content that competes with platforms such as TripAdvisor) and Google Flights (which is Google’s proprietary flight search products and competes with platforms such as Kayak).

My coauthors (including Columbia Professor Tim Wu, who was firmly in Google’s camp until he saw the results of our experiments) and I recently ran a series of experiments to understand whether Google is intentionally degrading the quality of its search results to promote Google Reviews. It turns out they are. The company’s practice is to show its own reviews toward the top of the page while excluding reviews from other platforms such as ZocDoc and TripAdvisor. Our experiments show that users are harmed by this practice, and would be happier if Google were to display reviews from all companies – using Google’s organic algorithm – as opposed to intentionally excluding other companies. This suggests that at best, users have to waste time scrolling through suboptimal results.  At worst, they are stuck with poor products and services whose recommendations are based on an intentionally incomplete set of results.

My Harvard Business School colleagues Ben Edelman and Zhenyu Lai focused on understanding Google Flights, another service that Google ties to its organic search. Google claims that Google Flights is the best product to offer consumers. Edelman and Lai came up with a test to check out that claim after they noticed that some searches (such as “flights to Orlando”) led Google to display its Google Flights service while very similar searches (such as “flights to Orlando FL”) did not. Notice that these are essentially the same search, and it was only a quirk in Google’s algorithm that led to the difference (and to my colleagues’ ability to learn from it). They found that Google Flights leads users to click on paid advertisements more often, while reducing interactions with unpaid content (which is good for Google, but bad for others). In short, Google was delivering the same content but making more money by making the organic content a little bit harder for users to get to.

While economic theory will always be an important tool for antitrust, these types of results highlight the fact that we are at the beginning of a new wave of data-driven insights that can aid in antitrust cases. Data and a firm understanding of causation are becoming increasingly common – and powerful – tools in this context.  For those evaluating the performance of search engines and similar online companies, these new tools augment the traditional models used by antitrust economists, helping them replace assumptions with data.

The European Commission has now decided to move ahead with formal charges against Google, alleging that it puts its own products ahead of rivals in comparison shopping searches. According to reports, charges may be expanded into additional areas of Google’s search and mobile business lines as soon as this week. This doesn’t mean that the FTC will reconsider its decision; however, there is compelling reason to do so.  In 2013, the FTC said that Google’s conduct “improved…the experience of its users.”   These empirical studies demonstrate the opposite. Regulators should start to use these new and more accurate tools for antitrust analysis in the digital age.

 

Michael Luca is an assistant professor of business administration at Harvard Business School