You mentioned that economics is constantly updating and can get outdated, while the law is more stable. For the law to change, you need new regulation or a new court decision. When considering the role of economics in antitrust cases, should courts stay up to date and take into account the newest econometric methods or economic theories in their decision? Or do you think that new economics theories only create even more instability in antitrust laws?
I think that courts do have to consider new theories, but should look at them skeptically. To give you an example, the way the law has evolved to deal with this problem is to lay down some scope conditions. Consider a merger case. The courts and the regulators will not look at the effect of the merger on all markets, but only on the markets that are immediately impacted by the merger. The Illinois Brick doctrine says that if anticompetitive behavior occurs, only the people who are immediately affected by it have a claim and not people who are farther away along the supply chain. There is value in having a narrower approach, as it is important to keep the legal inquiry within bounds.
There are some economists who, unhappy with the consumer welfare standard, suggest other approaches such as structural presumptions. You also mentioned before that efficiencies have rarely been accepted as defense in merger cases. Do you see a trend towards less reliance on economics in enforcing antitrust laws? What are your thoughts on this? Is there value in simplifying antitrust laws by focusing less on economics? Or is the problem more complex?
I think there are two intertwined issues here. One issue is just how big a role economic principles should play in antitrust law. Some people want to bring in questions of fairness - which economists rarely consider - or distributional issues. I do not think courts are going to go in that direction. On the other hand, there is a question about how rule-bound or formalistic antitrust law is. Even some economists actually think that antitrust laws should be highly rule-bound. For example, some of the early Chicago School academics like Robert Bork were very formalistic and did not want courts to do economic analysis of mergers. They just wanted courts to look at concentration ratios and, if they were high, to block the merger. They thought efficiency analysis would have been too complicated and courts should not be in charge of conducting that kind of analysis. Simple rules were a better way to advance efficiency. The kind of formalism that was best exemplified by Bork was then abandoned by both people on the right and on the left. In the Reagan administration, the Justice Department became more interested in efficiency as a standard. In following Democratic administrations, the Justice Department and the FTC moved away from efficiency while still using very complicated economic analysis to make predictions about prices. Their approach has not been formalistic at all. So this made things very complicated, and I think in the end, it complicated the enforcement of antitrust laws. So I keep those two things separate, and I am not sure everybody always does.
Before we move to labor markets, which you have written extensively about, I wanted to discuss platforms, which you mentioned earlier. I am curious to hear your thoughts on platform competition since many markets, if stretched, could be seen as a platforms nowadays. Do you think the current regulation is enough to deal with platforms, or is the law flexible enough for judges to have all the tools they need?
I think platforms have to be regulated in some way - like price regulation or regulation to prevent conflicts of interest. So the way I would put it is that, as you actually mentioned, we have always had platforms, and in the old days we would call them exchanges, and they have always posed a problem for antitrust law. There are a lot of famous antitrust cases going back to the Chicago Board of Trade involving exchanges. The basic problem with an exchange is that you have an institution that is creating a market, by bringing together buyers and sellers. This is positive. We should encourage it, and we cannot consider them per se illegal. However, they often reach high market concentration due to network effects. You are going to have either a monopolist or a firm with a lot of market power. The hard part is to detect whether their conduct is anticompetitive, or a necessity for their platforms to exist and their transactions to take place. Sometimes their rules enhance the market power of either the exchange itself, or of particular large agents that operate on the exchange. As of now, we have to review platforms through the complicated rule of reason. What has gradually evolved is the supervision by regulatory agencies, which issue supplemental policies. Authorities like the SEC and the CFTC are doing this in a wide range of settings. I imagine something similar has to happen with a lot of these internet platforms. It seems like we are moving in that direction for advertising, for example.
Moving on to labor markets, why should they be examined under antitrust laws, and why do you think they have been overlooked in the past?
Labor markets are markets like product markets in economics. When we think about antitrust cases, the first thing we usually think of is output markets. Firms are producing goods or services and people are buying them. We are worried that firms might collude to raise prices or become monopolies, which causes harm. But economics and antitrust law have always recognized that harm can occur on the other side of the market. If you have buyers that buy goods and services from suppliers, and they form a buying cartel to push down prices, that causes economic harm. If those suppliers are workers, it is the same type of harm. Law and economics have recognized this for a long time. It is just a mirror image of product markets.
In economics, you want competitive markets because they produce efficiencies. So if buyers get together - whether they are employers or just buyers of goods and services - they will create welfare losses. While that has been clear for a long time, there have been very few of these labor market cases. This, however, does not mean that the law is unclear. There is a famous Supreme Court case from 1926 called Anderson v Shipowners Association (272 U.S. 359), where the Court acknowledged that antitrust laws apply to an employer cartel. At the time, a group of employers fixed wages and agreed not to poach each other's workers. There are other famous cases involving sports leagues like the NCAA and the NFL. We have not seen many cases until recently, and no merger cases. I started writing about this about six or seven years ago because economists were researching the topic. They were not focused on the legal aspect, but they began realizing that labor markets were often highly concentrated and that the issue was slipping under the radar. In addition, literature emerged on non-competes and no-poach clauses, and how common they were. So there was abundant empirical research showing that labor markets were both very concentrated and a lot of anticompetitive behavior was taking place. This heightened the question of why there were so few antitrust cases against employers. This is what prompted me to start writing about it, though the answer is still unclear. My argument has been that more attention should be directed to labor markets because anticompetitive conduct causes harm both to workers and has spillover effects on consumers. We have started to see more litigation and more government enforcement recently, so it is positive that things are moving in that direction.
It might be common for new generations to always think that they are in a transitional moment in history. Considering remote work and new developments in artificial intelligence, we could expect these to really impact market definition in labor market cases. Is the fact that people are working remotely or that artificial intelligence is changing the workplace being used in cases to broaden geographic market definition? In your opinion, is this a transitional moment or is this no different from other technological developments other generations experienced before?
I think the word transitional is too strong at the moment. I think remote work is an important factor, but it is just too early to tell. My guess is that, ultimately, it is going to be marginal, as many firms are reconsidering their remote working policies. Even if it remains a permanent feature of labor markets, it will only impact certain types of jobs that can be fully remote, as hybrid workers will still have to be located near their offices. If this assumption is true, labor market definition will not be affected. So if we are to see a transitional moment with respect to remote work and labor market definition, it has not arrived yet.
It is also too early to assess the impact of artificial intelligence on antitrust. It is more interesting to speculate about that. There are lots of possible future outcomes if everything is automated as fewer labor markets will exist. If artificial intelligence just becomes a tool - like a calculator or a laptop - it will simply make certain jobs more productive, not affecting labor market definition. It might affect the definition of certain jobs. For example, if AI could enable nurses to write prescriptions, that would affect occupational definition or how you analyze certain types of markets. But I do not think it will affect the geographic definition of the market for nurses, because nurses will still need to examine patients. I could see a merger in which the number of doctors is reduced but not the number of nurses, ultimately softening the negative effect on consumers. A lot can change in the next few years. But that strikes me as a marginal change rather than a transitional moment.
I am personally very interested in artificial intelligence, especially when it comes to algorithmic collusion. While it is still at a speculative stage, are you more worried about the potential of collusion between firms, or are you more worried about the concentration of resources? AI requires huge amounts of data for training, and reinforcing mechanisms could lead to a highly concentrated market.
About the first, definitely. I do not know enough to be able to comment on the second. I had heard years ago that only large companies with huge datasets could do image recognition.
I think algorithmic collusion is a serious problem. It seems to me more of a problem of detection and enforcement than a conceptual issue. If a firm simply bought or developed an algorithm that maximized profits, it would be hard to say that that is an antitrust violation. If other firms did the same thing, and these algorithms ended up colluding in a more sophisticated way than what humans could do through parallel pricing, that would be a problem. This should be addressed through regulation. Regulating AI and detecting potential collusive behavior are two challenges we will have to deal with.