Jean Tirole is an intellectual giant in the economics world. The Frenchman is the foremost thinker on market power and regulation, and won the Nobel prize in 2014 for his work in this area.
His insights are particularly relevant today, as large tech firms grow ever larger and more powerful. Advances in technology has mostly made our lives better, but as privacy concerns rise and fake news spreads, we are starting to see the downside of giving tech companies mostly unchecked power. In the past, regulators could deal with this by breaking up firms or making them public utilities. That hasn’t happened with the tech giants, even though many people and policymakers feel like something should be done—but what?
Tirole’s recent book, Economics for the Common Good, offers some answers. The final third is a handbook on how to think about the ways technology is changing the economy, and what we can do about it. Quartz asked him some of the more pressing questions of the day.
Quartz: The early days of tech promised a ruthlessly competitive market place where even small players could reach billions at little cost. Instead, it seems we ended up with less competition. What happened?
Tirole: There is a sense in which tech has delivered. Small firms have been empowered in many ways. They can avail themselves of cheap back-office and cloud services; they can easily connect with consumers; they can fine-tune their advertising rather than engage in blind mass advertising; their access to borrowing is facilitated by AI-driven lenders, as is the case for the more than 7 million Chinese small and medium-size firms financed by Ant Financial. And, importantly, they can more easily build their own reputation. A taxi driver relied on the taxi company’s reputation; today, through ratings, the driver can have his or her own reputation on a ride-hailing platform.
But at the platform level, competition confronts the existence of large returns to scale and/or network externalities, leading to natural monopoly situations and a winner-take-all scenario. Network externalities can be direct: I am on Facebook or Twitter because you also are; I will use Uber or Lyft if many drivers do so. Network externalities can also be indirect: We may not care directly about the presence of other users on the platform, but that presence leads to improved services, as in the case of many apps or delivery services. For example, I want to use Google’s search engine or Waze if you also use them, as the quality of predictions improves with the number of users.
Natural monopoly situations lead to widespread market power, and a concomitant willingness to lose money for a long time to “buy” the prospect of a future monopoly position—think of Amazon or Uber.
Are tech firms like Google, Amazon, and Facebook monopolies?
Here we need to distinguish between statics and dynamics, or between a transient monopoly and a permanent one. Large economies of scale as well as substantial network externalities imply that we often have monopolies or tight oligopolies in the new economy. The key issue is that of “contestability.” Monopolies are not ideal, but they deliver value to the consumers as long as potential competition keeps them on their toes. They will then be forced to innovate and possibly even to charge low prices so as to preserve a large installed base and try to make it difficult for the entrants to dislodge them.
But for such competition to operate, two conditions are necessary: Efficient rivals must, first, be able to enter and, second, enter when able to. In practice, they may find it difficult to enter a market. And if they successfully enter, they may find it more profitable to be swallowed up by the incumbent rather than to compete with it. In economics parlance, such “entries for buyout” create very little social value as they are mainly a mechanism for the entrant to appropriate a piece of the dominant firm’s rent.
Ten years ago it seemed like Walmart had monopoly power when it came to retail, but the market brought us Amazon. Is it possible that today’s tech monopolies will also face stiff competition one day?
Yes, and let’s not forget that Google replaced AltaVista in the search engine market and Facebook dislodged MySpace in the social network segment.
Walmart and Amazon efficiently exploit returns to scale associated with purchases, logistics, and delivery. It does not mean that they are necessary monopolies, but it is no surprise that large firms emerge in this industry. Walmart took over thanks to intense cost-cutting; Amazon has been disruptive because its internet model further delivered customer convenience. This is a good illustration that industries must be analyzed from a dynamic perspective, and not only a static one. Again, the key is the ease of entry by an entrant who creates value for the consumer.
So, then, what are the potential barriers to entry?
Entry usually concerns a niche segment. Recall that Amazon began as online bookstore and Google as a mere search engine. Later, platforms may build a complete product line and expend to compete head to head with dominant platforms. But the initial niche entry may be hindered by the incumbent using technological or marketing tie-ins, like exhibiting a preference for its own services, or loyalty rebates, or else preying on the entrant. Entry is also made more difficult by the impossibility of user “multi-homing.” Switching to a new social network is easier if the user can easily port his or her content from one network to another. Entry into ride-hailing services is easier if the incumbent platform does not require exclusivity.
Do you think tech monopolies cause more harm than good? Who pays the price for any harm they cause? It seems like, unlike monopolies in the past, tech consumers face low or zero prices. Does that change how we think about or define monopoly power?
Yes, on the whole consumers tend to get a good deal, because we use wonderful services—like Google’s search engine, Gmail, YouTube, and Waze—for free. To be certain, we are not paid for the valuable data we provide to the platforms, as for example Eric Posner and Glen Weyl remind us in their recent book Radical Markets. But on the whole, our living standards have substantially improved thanks to the digital revolution. However, we should remember that these are two-sided markets. For example, on the other are advertisers, who pay very large amounts for the ability to target their offers to customers. So just looking at the consumer side is an incomplete analysis.
Does market power justify regulation? If so, can a traditional approach to regulation offer a solution?
Old-style regulation has a hard time finding its footing. Consider first public utility regulation, which for a century dominated the regulation of electricity, telecoms, and railroad companies. Such regulation tries to regulate profit in industries characterized by natural monopoly conditions.
For example, cost-of-service regulation looks at realized cost and sets prices so as to enable the firm to recoup its cost. It is very hard to implement in tech industries, despite the fact that they are high-fixed-cost industries like utilities (indeed, with a vengeance, since the marginal cost of supplying services to the end user is often nil). To enable cost recovery, one must estimate and factor in the estimated low and, more importantly, unobserved probability of success. Because most platforms fail, a non-negligible profit is needed to recoup costs, but one has little information about how much is needed. A good analogy is here provided by the case of drugs: also a high fixed cost, low probability of success, low marginal cost activity.
This difficulty is compounded by measurement issues. On the one hand, there are many potential future dominant firms, and regulators cannot monitor their expenditures at the start-up stage. The second measurement issue is on the revenue side, associated with the international nature of the platforms’ activity. Platforms already choose the location of their intangibles—patents, data, and so on—to minimize taxes. They could do so to derail utility-style regulation as well. Overall, public utility regulation does not seem an option.
What about just breaking up the tech giants?
There is nothing wrong per se about breaking them up. But breaking up firms only for the sake of reducing their power may fail to accomplish our goals. For example, breaking up Facebook into five Facebooks would do little to address privacy concerns.
In the past, we have broken up Standard Oil, AT&T, railroad, and electricity systems. Regarding internet platforms, we need to give it more thought. First, it takes time to implement divestitures. Railroads and electricity, and to a large extent telecoms in 1984, were simple and stable technologies. By contrast, the current platforms are rapidly evolving. We must make sure that the intervention is not obsolete by the time it is implemented.
Second, we need to apply economic reasoning. To break up a firm, we must identify the essential facility—characterized by natural monopoly features—that separates it from potentially competitive segments, and make sure that the essential facility does not succeed in monopolizing back these potentially competitive segments. This can happen either through a line-of-business restriction or the monitoring of fair access to the essential facility. An electricity company can be broken up in relatively clear segments, like generation, transmission, and distribution, with the transmission grid clearly being the essential facility. Similarly, the railroad tracks and stations are obviously facilities that cannot easily be duplicated by rivals.
Let us assume that we identify Google’s search engine as the essential facility and sever it from YouTube, Waze, and Gmail. One issue is whether the search engine would be as efficient at answering our requests if deprived of the data gleaned from other services. Overall, if structural remedies should not just be swept aside, much more thought needs to be given before using them.
So how should antitrust evolve?
First, we need to reconsider our burden of proof in antitrust decisions. This is a delicate matter.
Consider the acquisition of WhatsApp and Instagram by Facebook. They were social networks, just like Facebook. They could have become Facebook competitors. But is there any evidence for that? Not really, as this is just a guess on what the future would have looked at in the absence of acquisition. The suppression of competition in the absence of data is hard to prove. My guess is that we should err on the side of competition, while recognizing that we will make mistakes in the process.
Second, economists must help antitrust authorities to identify harmful behaviors and design simple remedies. For example, best-price guarantees, also called most-favored-nation or price parity clauses, guarantee that the consumer will benefit from the lowest price on a good or service when using the platform. As I explain in Economics for the Common Good, economists have shown that this apparently benign behavior can allow platforms to collect substantial merchant fees from sellers who need them to reach their unique consumers. And there are a number of topics that we must study in more detail, such as data ownership and data barriers to entry.
Third, jurisdictional issues have become more acute with the digital economy. We must insist on a level playing field and not impose different regulations on different competitors on the basis of an arbitrary industry classification and industry-specific regulations. For instance, traditional media are more constrained in editorial responsibility and advertising than social media. To create a level playing field as well as to increase efficiency, we must also insist on international harmonization of intellectual property rights and taxation, and make sure that global companies are not exposed to a large number of heterogeneous and incoherent regional regulations.
Finally, we must make heavier use of more reactive processes. Drawbacks of classical approaches are well-known: self-regulation tends to be self-serving; competition policy is often too slow; public utility regulation, as we discussed, is mostly infeasible (and it is sometimes captured). We must develop what I would call “participative antitrust,” in which the industry or other parties propose possible regulations and the antitrust authorities issue some opinion, creating some legal certainty without casting the rules in stone.
The possibility of error must be accepted, and so the regulatory innovations must evolve as the authorities learn by doing and slowly incorporate them into guidelines. Such adaptive policies have enabled the comeback of patent pools (through business review letters), and include regulatory sandboxes, which are testing grounds for new business models that are not protected by current regulation, or supervised by regulatory institutions.