Below, we have provided the full transcript of our panel discussion Self-Preferencing and Online Platforms: A Universal Theme?. Read below to see the timely discussion where a panel of experts deepened the discussion regarding this topic, and how it specifically relates to South Korea.
Frederic JENNY:
Well, good morning, good afternoon, and good evening, and welcome to this CPI panel discussion on Self-preferencing and Online Platforms: The Universal Theme? This is a very timely discussion, as the Google Shopping judgment of the European General Court was delivered last week. And as the discussions on the relative merits of ex-ante regulation and ex-post competition law enforcement to reign in the alleged abusive practices of large tech firms is going on in many jurisdictions.
We have a wonderful panel, and let me very briefly introduce the panelists.
Reiko Aoki, who’s commissioner of the Japan Fair Trade Commission since 2016, and was Professor of Economics at Hitotsubashi University.
Marcus Bezzi, who’s the Executive General Manager of the ACCC in Australia since 2009. Tsuyoshi Ikeda who is a very prominent antitrust competition law petitioner in Japan with the law firm of Ikeda & Someya. He was previously an investigator of the Japan Fair Trade Commission, where he participated in many dawn raids, leniency programs, and investigation of cases.
We have Michael Jacobides, who is the Sir Donald Gordon Professor of Entrepreneurship and Innovation and Professor of Strategy at the London Business School.
And we have Madoka Shimada who’s a lawyer with the law firm Nishimura & Asahi, specializes on competitional matters, and also a member of the subcommittee on Unfair Trade Policies and Measures of the Industrial Structure Council at the MITI since 2013.
My name is Frederic Jenny. I’m the chair of the OECD Competition Committee, and a Professor of Economic, at ESSEC Business School.
Now, if we get into the topic, there are sharply divided views among competition authorities, economists, and courts over whether self-preferencing by dominant firms is always unfair and anti-competitive, or whether it can be pro-competitive, and how it should be dealt with. Those different views are partly the result of the fact that competition laws are different in different countries. For example, whereas in Europe, dominant firms have a special responsibility to ensure that they do not act in a way which could impair the competitive structure of markets. Firms with monopoly powers in the U.S., or market powers, do not have a duty to help their competitors, as we were reminded in the Trinko Decision.
Furthermore, while some economists are used to self-preferencing, as a normal business practice, and may in fact promote innovation and increase competition, and they refer, for example, to the self-preferencing of private labels by supermarkets, which is seen as a sign of increased competition in products, others are of the view that self-preferencing by large digital ecosystems is unfair and can deprive consumers from the benefits of innovation.
One question, of course, is whether digital markets and ecosystem are different from other markets, and I’m sure that we will discuss this. Now, with respect to courts, let me mention two cases with strikingly different results. The first case is in the UK, the Streetmap Decision by the High Courts from February, 2016. In this case, Streetmap, who was a very early mover in the online mapping space, accused Google of abusing its dominant position to promote its mapping service within its general search results, thereby having a knock-on effect on the traffic to Streetmaps own service. The argument of Streetmap was that the inclusion by Google of a map at the top of the search results, displaying results from Google Maps, promoted Google’s own maps more favorably than its rivals. And hence, the claim of anti-competitive behavior by Google.
Justice Roth, from the High Court, dismissed Streetmap’s argument, ruling that the inclusion of the Google Map box at the top of the search result, was, and I quote, “Not reasonably likely appreciably to affect competition in the market for online maps,” and that Google’s conduct was objectively justified.
Now, Justice Roth agreed that preferential promotion by a dominant company to its power on the markets where it is dominant of its separate product on a distant market where it is not dominant, may constitute an abuse if it has the effect of strengthening its position on that other market and is not otherwise objectively justified.
However, to do so, and in application of the jurisprudence of Tamra in Europe, all the relevant circumstances of the case have to be taken into consideration. And in this case, the questions were, did Google intend to foreclose competition by introducing Google Maps one-box? Justice Roth concluded that it had not. Its main purpose was to improve its general search engine by remedying its perceived deficiencies in comparison with the offering of its competitors.
The second question was, did Streetmap establish an actual and appreciable effect, rather than merely a potential effect, on the market for online mapping services? Mr. Justice Roth found that they did not.
And the third question was, could Google objectively justify any possible abuse on its part? There were two issues or two sub-question to this. First, whether any exclusionary effect could be outweighed by advantages that also benefit consumers, and was the conduct proportionate? Justice Roth found that, yes, indeed, the Google map system was far superior to its competitors, which meant that consumers benefited from the Google Maps service which was offered. Second, that it was proportionate because any alternative would have been very impractical and overly burdensome.
Now, altogether, the court held that Google Maps was successful in its own rights because of a number of technological innovations, which provided a better product than Streetmap, and therefore, these developments were pro-competitive and demonstrated competition on the merits. Now Google was, of course, very pleased with this and made a statement to the fact that the court made clear that we’re focused on improving the quality of our search results. This decision promotes innovation.
Now it’s interesting to contrast this decision with last week’s decision by the General Court of the EU. When the General Court upheld the Google Shopping Decision of the EU Commission, which found that Google had abused its dominant position by self-preferencing its own specialized comparison-shopping service, Google Shopping, over that, of its competitor on its general search pages. I will only quote two paragraphs of this judgment.
One of them, which is paragraph eight, states that, “It is uncontested that Google began providing internet users with a Comparison Shopping Service in 2002. And that this initiative was in recognition of the fact that the processes that had either to be used by search engines did not necessarily return the most relevant results in responses to specific searches, such as those relating to news or shopping.”
In other words, the court says, “Yes, there was an innovation there. And this is one of the reasons why Google started its shopping service, it’s comparison-shopping Service.” And the results were not the result of the ordinary general search algorithm, but they were specific to try to get the best possible, the most relevant offers possible.
The second paragraph I want to quote is the heart of the reasoning of the General Court’s decision, I think. It’s paragraph 177, and it says the following: “The rationale and value of a general search engine lie in its capacity to be open to results from external third-party sources and to display these multiple and diverse sources on its general results pages, sources which enrich and enhance the credibility of the search engine as far as the general public is concerned. And enable it to benefit from the network effects and economies of scale that are essential for its development and its subsistence in a market in which by their very nature, few infrastructures of that kind can subsist given those network effects. A very large number of users is needed to reach the critical mass capable of compensating for the service being free of charge on one side of the market, and generating advertising income on its other side. Accordingly, for search engine, limiting the scope of its results to its own entails an element of risk and is not necessarily rational, save in a situation as in the present case, where the dominance and barriers to entry are such that no market entry within a sufficiently short period of time is possible in response to that limitation of internet users’ choice.”
So, the court says, basically, that self-preferencing is not a rational behavior, at least for a digital platform, unless the benefit is to restrict competition if there are high barriers to entry and no possibility for competitors to compete. It does find that this is not competition on the merits.
Now, we have two very different visions there, and I think that it’s interesting to ask our panelists to discuss two sets of questions. The first one is, where do they stand on this issue: is self-preferencing by dominant digital platforms always anti-competitive? Or if they believe that it may not always be anti-competitive, under which conditions should self-preferencing be considered to be a competition violation?
Now, sub question to this is, of course, what to do with self-preferencing by dominant digital firms, whether the self-preferencing by dominant digital firms of their services, when those services are more innovative or better quality than the services offered by competition. Is this a relevant issue or not?
Now, let me turn first to Reiko Aoki, as an enforcer. How should we define self-preferencing? And what are the limits of the concept? And once we’ve defined it, what are your views about whether self-preferencing by digital platforms is usually or presumptively a violation of competition law? Reiko.
Reiko AOKI:
Thank you, Fred, for inviting me to participate in this exciting panel.
First of all, I want to say that I am participating in this panel in my private capacity, but I am aware I’m a Commissioner of the JFTC, so let me just first mention that the JFTC has looked into self-preferencing. And, our view, the JFTC’s view, is that it can lead to foreclosure, exclusionary behavior, or unfair competition, is mentioned in two reports of fact-finding surveys that were done on app stores and digital advertising. You can find them by going to the JFTC’s English webpage, which you can find by Googling, or Duck-Duck-Going or accessing directly.
Under quick link in JFTC webpage, you can find “Approaches to Digital Markets” tab. And you’ll find the fact-finding survey reports in English there.
Self-preferencing, as Fred just mentioned, is hard to define. I’m first going to talk about why I think self-preferencing is anti-competitive. And I’m thinking of self-preferencing mentioned in Google Shopping in the UK, also the Amazon Buy Box, and Google’s ad exchange bidding algorithm, which is included in the JFTC’s digital advertising survey mentioned before.
I think self-preferencing is a type of c leveraging akin to tying. It can leverage between vertical affiliates, but also unrelated affiliates of the digital platform. The mechanics of leveraging and its effect is very unique to digital platforms with an ecosystem, particularly when involving data. And self-preferencing may lead to more data, but necessarily no more sales in the short run. So, in the short run, there may be no immediate effect, but will give the platform an advantage in the long run, allowing it to reach into and define future markets. And this avenue of advantage is not available to platforms without an ecosystem, which Google and the two cases that Fred mentioned would be an example.
In the long-run, competition is about innovation, which was also included in the argument. And self-preferencing used for *innovative service* is always going to be a suspect, and I’m almost tempted to say that self-preferencing by a digital platform with an ecosystem should be per se prohibited for this reason. If the new service is so good, it should be introduced under the same condition as rivals. Maybe new service is good or is innovative only because it was introduced through self-preferencing.
I also note that self-preferencing is akin to tying, but it’s different from the old tying known in competition. It’s not uncommon for a once innovative firm, not necessarily digital platforms, that start leveraging their legacy patents or established network base in order to enter or foreclose a new market, typically by tying and bundling, including rebates. An ecosystem can leverage for advantage effect in an affiliate product or market without innovation, and it is going to be detrimental to long-run competition. So traditional tying was used to compensate for lack of innovation in the past, which is relevant in the current market, but self-preferencing by an ecosystem digital platform is going to replace innovation NOW affecting the FUTURE market. And there have been arguments in the past that bundling resulted in a new innovative product. Unbundling (result of antitrust action) led to entry of new innovative firms, and it’s very ironic that some of these new innovative entrants are now resorting to self-preferencing.
I’ll stop there, thank you.
JENNY:
Well, thank you very much, Reiko, for this pretty strong statement. One a very interesting aspect of what you were saying was the fact that you see the danger as being a danger to long-run competition. Now it is sort of- the competition authorities are usually a bit more concerned with short-run price effects and less with a long-run effects, but you come out very strongly by saying, well, presumably it is an anti-competitive practice because it may have long-term effects.
Let me turn, maybe as a contrast, to Madoka Shimada, who, I think, has a more lenient view of the practice. Madoka.
Madoka SHIMADA:
Thank you very much for this great opportunity to be participating on this panel. And thank you very much, Ms. Aoki, for your really interesting and insightful views.
As Fred said, I think that self-preferencing should not be deemed as presumably anti-competitive. And it should be illegal probably only in some kinds of specific cases. There’s a simple idea, as Frederic said in the beginning, that self-preferencing, if we define it as any promoting of one’s own business, it’s a kind of natural thing, natural behavior for every business who sells their products or services in a good way. And of course, it’s related to vertical integration and that could produce enhanced innovation and efficiency. This is a very simple things that I think.
Of course, if we look at the digital platforms, probably there are some like in different factors than if we compare to the non-digital platforms, the basic economic idea should be the same. As Fred said, let’s look at that. The grocery store has various things, and it’s a very natural thing to get their own brand, their own private labeled things on the good shelves.
Now why is this kind of grocery store marketing is not seen as anti-competitive? Probably because there are a lot of other shops out there. You can go even though it is very big chain store like Walmart, you can find another store where you can shop. And this is really a kind of analogy for the marketplace for things like Google Shopping.
Of course, we need to look at that. If there’s no other choice, if there’s only one shop in the world, of course it should be problematic. We need to really carefully see that- really the shop is in a dominant position. In Japan, probably the legal structure is a little bit different from Europe. We don’t have a kind of dominant position that are basically the same. So, from that point of view, I think that not all the self-preferencing should be allowed, of course. Only for specific cases.
And the next question is what should be those specific cases? Well, as a practitioner, I will look at the JFTC’s recent surveys and fact reports and fact findings. And how they categorize self-preferencing, in specific legal theory. They find that one of the categories of conducts in the Unfair Trade Practices, which is interference with competitors, and that interference with competitors category is, let’s say that this is kind of a basket clause that is used when no particular other theory of harm could be applicable. Well, maybe, this is questionable.
Let’s look at that, the precedents of this interference with other competitors. The cases where the intentional exclusionary conduct against the specific competitors, that case can amount to things, of course, that could be anti-competitive. Also, the second type should be the refusal to trade type of thing. The platform is an essential facility, an indispensable service. In that case, and if there’s refusal of trade, refusal to provide services as a basis of service competitors, the competitors should be in a very difficult position to continue to business.
This is kind of an extreme case, I believe. So well, looking back to the Japan case. Still, there are a lot of cases that the JFTC has already looked into the kind of digital platform mergers, but not pure self-preferencing without these factors, not being kind of seen as kind of violative. I think we need to draw the line in the very simple idea of self-referencing. It’s a really intrinsic motivation of business people. But, of course, only limited to specific cases, maybe some kind of extreme, only those maybe anti-competitive.
Thank you.
JENNY:
Thank you very much, Madoka, for this. A more nuanced view. You associate the possibility of anti-competitive effect, particularly to market power, and you say that in the extreme cases where there’s a very strong market power, nearly an essential facility, there is a case against self-preferencing. But on the other hand, self-preferencing is a normal business practice in many sectors and should not be treated differently in the digital sector.
Now, let me turn to Michael Jacobides, because I think that one of the themes that you want to touch upon is a difference between the digital sector and the non-digital sector when it comes to self-preferencing. So, Michael.
Michael JACOBIDES:
Absolutely and thank you very much. I think that this is also the difference in my mind between what we heard from Reiko and what we heard from Madoka.
Let me just also start from a very personal note, which is that Madoka Shimada told us what is normal in the course of business. And I agree that trying to attain profit and leverage your position is exactly what we would teach in business school in terms of what firms should do.
Now, the problem is that, as someone who has both researched and advised on the corporate side, I think that I got so concerned about the potential for abuse in the digital sector that I started looking at regulation. And there we go to what we heard from Reiko Aoki.
I would also venture to suggest that the difference between the two judgments that Frederick Jenny told us 2016 and 2021, is also that 2021 is not 2016. One of the differences, and I think that that has weighed in how the court decided in 2021, is the understanding of the fundamentally different economics, both in terms of the platforms and in terms of the ecosystems that Reiko Aoki mentioned.
I should also mention by way of blood, perhaps, but really to address to what are the underlying economics of it that there is a special issue coming out in Industrial & Corporate Change that I co-edited with Ioannis Lianos, the chairman of the Hellenic Competition Commission. Frederic has the leading article in that, and many other of the known suspects are contributing to that special issue looking at the specificities of what happens in the digital context.
Again, I think that it is extremely important to consider what are the differences, not only in terms of degree, but also qualitatively in the digital context and what are the economics. I think that the difference between 2016 and 2021 also was 2019, and I’m referring to that because three important reports started outlining the underlying economic issues that emerge in the digital context.
And with due respect, I do not find any correspondence between Walmart cases and Google cases. They’re very different kettles of fish. Also, the idea that self-preferencing, now this is a very important one, we are not speaking about banning Google from making maps. We are not speaking about banning anyone from creating anything that is innovative and pro-competitive. I honestly cannot, for the life of me, see what could be pro-competitive for having either an exclusion or a self-preferencing, i.e, not allowing competition in fair terms.
By the way, when we speak about self-preferencing, if anything I’m concerned that the label is too narrow, because when you look at the digital issue, one of the main concerns is interoperability. The Australian Commission has done a good job in that, speaking about some of these issues in terms of the APIs and comparing that and I’m sure that we’re going to hear from the investigation they did in the function of the app economy in a little while, but I’m just pre-staging it a little bit.
The problems that we have in these digital contexts, which is what Reiko Aoki described, is that the nature of power is wholly different, and what we are also appreciating is the behavioral propensities that lock customers in.
The word pro-competitive is very tricky here; why? Because we are increasingly understanding how behavioral propensities of what customers do are locking them in to default positions.
Richard Taylor, speaking about someone else between 2016 and 2021, got his Nobel Prize in 2017, partly in showing part of these behavioral predispositions. I think that for anyone reading any newspaper, you understand how these are assiduously and often egregiously used by some of the digital platform firms for their own advantage.
If I go back to the comparisons, why is it that there could be something taken in the Walmart case of a pro-competitive effect if Walmart starts having its own label in addition to having things from Proctor Gamble or Unilever? Fundamentally, it is a rebalancing of the power from the power of the FMCG producers and their position in the entire consumption chain, and shifting it a little bit to the distribution channel, so that the power of the brands gets diluted.
What do we have in the digital context? Not only have we gotten a remarkable concentration of power of these digital firms, the other thing that we have is precisely the topics that Reiko Aoki mentioned in terms of the way that those that own the platform have information advantages and the creation of the ecosystem lock-ins that are particularly relevant for a given type of digital firms. Now, within the universe of digital firms—I’m referring to something that I developed in another paper that you can find in the www.devolutionlimited.net website, which is on what is positional power—I think that the problem becomes rather different. You can say, okay, in the digital context, power works in different ways. And there are some firms, very few firms, as I think we will all agree, that are becoming the essential gatekeeper between us and the environment. I think that within this universe, there are characteristics, and we are using in terms of anti-trust the term of gatekeepers in a very loose term.
The European Union created something which I intellectually abhor, which is if they are really big, probably American also, if they are more than 65 billion, or they have X number of customers. I think that all of this reliance on numbers is unhealthy and also academically entirely unjustifiable. I think that what we should do is have some criteria to say, “ah, I’m going to tell you what is the nature and the exercise of power in some firms.” And for them, provided that they have these criteria, I will say that any self-preferencing should be, per se, not allowed. Again, per se, we are not speaking of you not creating a tight connection. We’re speaking about two things. We’re speaking about you not unduly pushing others out.
I cannot, for the life of me, understand how that would ever increase competition and choice for the customer, especially given the lock-in dynamics, and especially given the way that these multi-product and experienced ecosystems lock customers in. And the second thing is that I think that we should really be striving for mandatory, and as common as it can be, interconnection. And again, in the digital economy, I think that the question of how interconnected and interoperable these services are will significantly promote innovation. I do think that in the case of digital platforms, and in particular, when you have gatekeepers, who are the gatekeepers? Just to give you one example, you have gatekeepers when you have firms whereby the competition between two of them does not resolve the issue of the competition within.
I don’t want to abuse the time, perhaps I will need to return to the definition of gatekeepers and self-preferencing in another comment in our talk. Should be mandated to not have preference, not exclusion, this is a much narrower definition. I think self-preferencing should also be mandated to have those that can provide similar service the same right of interconnecting. Why? Because they benefit from inherit economies of network economies and economies of this behavioral capture that reduces competition. And what regulators should be doing is trying to redress all these issues that Reiko mentioned.
A few more things, but let me park it there because I’m afraid of expanding too much in my response.
JENNY:
Thank you very much, Michael, for pointing out that important difference between the digital sector and the other sectors, and the relevance to gatekeepers.
I want to turn to Marcus Bezzi because, as we’ve already heard, the ACCC has conducted a number of market studies on those issues. And I’m under the impression that those markets studies suggested that some forms of self-preferencing might be objectionable, but others not necessarily so in the digital sector.
Marcus, you have the floor.
Marcus BEZZI:
Thanks very much, Fred, and as Michael was talking, I was reflecting on your question at the beginning, on the difference between the two cases. I think Michael does have a point. In 2021 competition agencies, and perhaps also the courts, have a much deeper understanding of the important role of data and network effects in empowering digital platforms.
As you’ve said, we’ve done a number of market studies now. And our market studies power is a very strong one. It enables us to get access to a lot of confidential information that’s held by market participants. We have found that self-preferencing is common, and it’s identified in each of the reports that we’ve done, including reports on app marketplaces—the one that Michael was referring to—choice screens, and ad tech.
There’s been self-preferencing in all of those circumstances. We found what we’re dealing with, and digital platforms are vertically integrated firms with a huge amount of influence on the consumers and businesses that rely upon their services. We’ve found that there’s dominance in ad-tech and in search in Google, Google dominates those markets. In app marketplaces, it’s Apple and Google. In social media, it’s Facebook. For each of these dominant firms, there’s a degree of vertical integration, and there’s evidence of considerable anti-competitive self-preferencing when the firms favor their own interests over downstream rivals in a number of different ways, but that depends a bit on the market. It can include things like pre-installation and default first party services using algorithms to favour first party services. By that I mean services of the business that is engaging in the self-preferencing, or using information collected as a service provider to benefit the services of that service provider over rival services.
We’ve really seen the important role of data, and what Reiko described as the need for long-term access to data flows as a way of entrenching a strong market position, and that, we think, over time is creating significant barriers to entry and expansion.
There’s also a lack of transparency between platform operators and users in the operation of many digital markets that we’ve studied, and the users, being both businesses and consumers, really have no idea about what is going on much of the time. We’ve identified patterns of behavior that we’ve described as dark patterns. Where users are manipulated in many ways to favour the interests of the platform operators. We don’t say that self-preferencing is always a problem. We do conceive that there may be circumstances where self-preferencing by these businesses is not a problem, but we do start from, I suppose, a presumption that if these firms [with market power] are engaging in systematic and widespread self-preferencing in the course of their business, that it is likely to be a problem.
JENNY:
Does that mean that you don’t see a problem if there’s no market power by the firm engaging in the practice, or are there others…
BEZZI:
If the firms are dominant, and we’re looking here at four or five different dominant firms within their particular markets, we do start from the presumption that there is a problem, yes.
JENNY:
But it’s a rebuttable presumption?
BEZZI:
If there’s no market power, if it was a different firm and it had no market power, yes, we fully conceive that there may be no competition issue.
I should add one further qualification, and that is that in our system there may be a justification. We don’t look at it from an efficiency perspective, but we can take the view that there may be a public interest justification in some form of self-preferencing. So, for example, if privacy concerns or security concerns are favored in order to promote a stronger ecosystem for consumers, that may well be an argument for which there is a strong public interest and which in our system, we may recognize.
But as I said, we start from the strong presumption that if these dominant firms within the markets that are identified are engaging in self-preferencing, then it’s likely to be a problem. But we do want to have a closer look.
JENNY:
Okay, very good. Thank you very much.
Let me finish by turning to Tsuyoshi Ikeda. As a former competitional enforcer, and as a lawyer now, how do you look at those issues?
Tsuyoshi IKEDA:
Thank you, Fred. It’s my great honor to be here with the excellent colleagues, and as an ex-official and a current private practitioner, I have to have, maybe as expected, a balanced view.
In Japan, we didn’t have any precedents that specifically the JFTC code mentioned as a self-preferencing case. Yet when I looked back at the precedents, maybe there are some cases which can be classified as the self-preferencing. For example, the mechanical parking lot case. The manufacturer of the mechanical parking lot offered repair parts, but when it came to the process offering the repair parts to third-parties, not the manufacturer itself, it delayed those orders so it would take months for the third-party to get repair parts.
This can be self-preferencing, maybe in a very traditional industry.
Also in the past, the JFTC gave a warning, not a formal cease and desist order, against an internet service provider—the biggest internet service provider—NTT East. In that particular case, the JFTC specifically mentioned that the NTT East could take advantage of information it had obtained through third-party internet service providers. Because the third parties had to negotiate with the NTT East to obtain the internet connection, which could be offered to the end-users, NTT East’s conduct may also be regarded as self-preferencing under recent theory.
These cases are not, of course, judged by the theory of self-preferencing, rather these cases were justified by the traditional, private monopolization or unfair trade practices. So the question would be, especially under the context of Japanese law, whether we should the define self-preferencing as a special antitrust violation apart from the traditional ones.
Of course, I have to admit that self-preferencing in the online market is more serious and more problematic, as other colleagues have already mentioned, the network effects under the work of the data, and also, maybe the speed of business, sometimes makes the situation much worse than in the traditional markets. Of course, competition authorities have to tackle with the difficult situation, but when it comes to Japan, as Madoka mentioned earlier, we have a relatively broad tool, especially recarding interference with competitors’ transactions, which is one category of the unfair trade practices, which is capture regulation, and that covers a lot of the potential illegal conduct. Not even a dominant position is required to apply this provision.
In this context, I would rather emphasize that we have to mind the risk of over-enforcement rather than under-enforcement, because the regulation is so broad. And that we also have to mind the fact that the Japanese law doesn’t have a per se aspect, even for the very obvious price-fixing cartels or bid-riggings. So, we have to prove anti-competitive conduct around the competitive effect in the relevant market. That’s the basic theory under the Japanese Law.
First, I have to admit that it’s very difficult to claim on which occasions, self-preferencing is illegal under the antitrust law, but I would suggest that we use self-preferencing as a kind of tool to evaluate the conduct. When we analyze the interference with competitors’ transaction case, we have to prove the illegal conduct or behavior, and also the anti-competitive effect in the relevant market.
Again, the anti-competitive effect is a case-by-case analysis, not a per se rule applied, so we have to take into account the various factors, including the market share and the size of the network effect and even among the Big Techs, the situation is quite different. For example, for the app store, there are two companies, but when it comes to search engines, Google is dominating the market. In the social networking field, Facebook is, of course, strong, but there may be alternatives. Even among the Big Tech companies, the situation is different, so we have to have a case-by-case analysis for proving the anti-competitive effects.
But when it comes to the illegal conduct part or element, of course we have to consider or take into account the pro-competitive effect toward innovative aspects of the self-preferencing. Self-preferencing is often associated with the offering a new technology, a new service to the consumer, but the fact that the big companies have unequal treatment to third parties in the peripheral treatment of their services, but the treatment against the third-party sites. That may lead to the kind of a presumption that their conduct does not have any justification when we evaluate the conduct element of the Big Tech behavior. In that sense, self-preferencing has a special meaning in the context of the antitrust law enforcement. That is my thinking and suggestions here. Thank you.
JENNY:
Thank you very much, and thank you very much for introducing, in fact, the second part of our discussion.
I’m going to ask each speaker to be rather short because we’re running a bit out of time, but I want to know what the implications of your views on self-preferencing and the dominant views being that there are concerns, at least in some cases, but for some of you in nearly all of the cases where there is self-preferencing by a very large ecosystem.
Is this an argument in favor of ex-ante regulation against those practices, or should they be dealt with as was just suggested a moment ago by Mr. Ikeda? Should that remain a case-by-case analysis within the general context of competition law enforcement? This question is quite interesting because, as we know, there are discussions in many different jurisdictions about whether one needs ex-ante regulation as a complement to competition law enforcement.
Let me turn to Reiko Aoki. First, you came out very strongly against the self-preferencing. Can you explain the reason for that? Does that mean that we should have an extended regulation that would prohibit self-preferencing by other gatekeepers or by dominant firms?
AOKI:
Personally, I don’t think so, because I am very skeptical of ex-ante regulation in general. Economics has taught us that with things like mechanism design and new market design models you can regulate an industry or firm, but it also tells us about the limits. It is very important to have very good information in order to have successful ex-ante regulation. And exactly because digital platforms and ecosystems are very new, I think it’s very difficult to have successful ex-ante regulation. Even with self-preferencing, which is a suspect, we need to apply the existing antitrust laws to keep it from having anti-competitive effects on the market.
Another thing I want to add about regulation, particularly in context of innovation, is that it’s very easy for the regulator to be captured by the incumbent technology, old technology, and that would be another reason against having ex-ante regulation and making self-preferencing per se illegal.
JENNY:
But isn’t there a slight problem? I mean, you explained originally that self-preferencing today may not have any effect on the current market, but will have an effect on the future. That this is a long-term concern that one can have with self-preferencing. Now, it doesn’t that mean that it has to be stopped immediately, which competition law enforcement is not really able to do because we have a quasi-judicial process. From that point of view, isn’t there an advantage from regulation? Regulation prevents the thing right from the beginning, from existing, and therefore, we don’t risk those long-term difficulties that you were concerned about.
AOKI:
Well, that’s a very good point. First of all, antitrust does deal with innovation and long-term effects, so we have the framework to do it. In terms of speed, to have effective regulation, you need to define exactly what the self-preferencing is. As Michael told us, it is a problem when it is done by a gatekeeper, and the definition of a gatekeeper is going to be very difficult, although Michael started to tell us what it was.
If we look at what’s going on in Europe, they had the Google case first, and now the regulation system is just beginning to run. It’s basically taken a similar amount of time precisely because it’s a new industry and a new way of thinking. Thank you.
JENNY:
Thank you very much.
Now, let me turn to Madoka Shimada. For some reason I believe, particularly given what she told us originally, that she’s not so keen on ex-ante regulation. Is that correct?
SHIMADA:
Yes, you’re correct. Really, as Aoki said, ex-ante regulation can be very difficult to craft, design effectively and update that in a very timely manner to catch the ever changing, fast-changing digital platforms businesses. It’s simply very difficult for a government to consider and catch up with the movement in the private sector, digital platform and various technology. That’s one reason.
And the other thing is, with respect to the self-preferencing, as I said before, I am a supporter of the case-by-case competition law enforcement type of approach, not ex-ante. Because that only gets specific cases with specific facts and evidence to be regulated. That’s a simple reason that I support ex-post regulation. Well, not regulation, probably ex-post law enforcement.
And, of course, more industrial, best market practices kind of things. It is a good idea to have people’s hand in pro-regulation and have the industry itself consider what is the best way, as a code of conduct, to make their business fair and better. That is scaled up social responsibility of the subject. The company plays big power, but I don’t think that the government can do that role. It’s too complicated, too fast-changing and too innovative to catch up.
JENNY:
Okay, thank you very much.
Let me turn to Michael Jacobides. You told us that you couldn’t think of any case where self-preferencing would be anything but anti-competitive by a very large ecosystem. Isn’t that a clear argument in favor of ex-ante regulation?
JACOBIDES:
Let me reflect also on the answers that we got so far. I’m a strategy guy who works now on regulation, as opposed to regulation guy who looks at strategy. I was a little bit perplexed by the concern of how difficult it is. And with all due respect, it ain’t that difficult to say that you will have forced interoperability. It really, honestly isn’t. And it doesn’t take someone to redesign the entire industry.
Saying that the interoperability that a company gives to their own subsidiaries will need to be given to other companies as well. I.E., forcing greater openness, is not hard. We would not have Telco if in the UK, NTT East—which was doing mightily nasty things in terms of long versus short-term interchanges—was not forced by the federal government about 70 years ago to open up.
Again, that was not terribly complicated.
So, I think that there are things that are doable. I’m a little concerned that we overintellectualize. The number of the papers that they’ve recently seen, most of them funded by Big Tech, do overintellectualize the argument, so let’s be real. A, can you easily say, will I force interoperability? You absolutely can.
B, can you define, provided that you are a gatekeeper, and I think that there could be some criteria. Let me give you one example or at least the current pitch that I have. I have four to five simple questions. Does the platform have exclusive access to large body of consumers? Is difficult for users to multi-home or switch platforms? Are sellers substitutable? Do users benefit from network effects requiring sellers to multi-home across platforms? And does the platform have an established network or ecosystem of multiple different products that create lock-ins that increase barriers of entry? Now if the answers to all of the above are yes, and again, that is not terribly complicated, then I think that you should be able to have some ex-ante rules.
The two ex-ante rules that I would say are forced interoperability and no self-preferencing. To give you an example of that, consider the differences between large platforms. You can think about Apple, for instance, that is pretty much guessing most of them, and Uber, that is no in most of these. Why is this relevant? Because if you say that I have two platforms where the answers are yes, I have Apple and I have Google, or Android, then the fact that there are two does not mean that there is sufficient competition.
Think about yourself as a dating platform that has networks nowadays, match.com. Now, if match.com is told by Apple, “Give me 30%, or I’m going to go out.” You can’t say, “Oh, well, that’s fine because I’m going to go with Android.” Why? Because you won’t want your customers to miss the dates of the people have only one phone, and it is an Apple.
What does that mean? It means that this is a gatekeeper, unlike Uber and Lyft, where both the drivers and the sellers can move around.
I think that these basic qualitative differences between that different types of platforms mean that there is a subset of platforms for which we should have a higher bar. And this higher bar for me is that we complement the ex-post with the ex-ante.
By the way, when you look at the ex-ante, there are also ways to do it reasonably. How do you do it? You say no self-preferencing. I, as a competition authority, either look at it myself, or I can have others who report to me. If I see self-preferencing, I then have within, 60, 90, 120 days—again, I’m not a regulator, so I wouldn’t be setting any of these limits—to have an explanation of why this is not anti-competitive.
What are the merits, arguments for the maintaining it, or the absolute integrity managed for determining it? And only if the competition authority is satisfied and says fine, I’m going to waive it. But there should be a default, and the default is no, because we know that this is not defensible given the economics of digital platforms.
What do I think? Limited, targeted ex-ante. Ain’t that difficult, let’s not overintellectualize. We’re not speaking about us rethinking business models. And then, this combined with ex-post investigations, which I favored the stuff that the CMA does which look at business models and tries to identify the harm and also allows you to go outside these massive platforms, and then see whether there are platforms that create problems.
I’ll give you an example in Greece. Ioannis Lianos and I put together a new article, which may or may not be voted as a result of corporate pressure, in the new law for competition, which would allow, for instance, to challenge things like, booking.com, like the Germans have done in terms of their request that nobody else has a price that is lower than what is offered there, which is a clear exercise of power.
I do believe that there is a role for focused, ex-ante rules. I honestly cannot think of, having looked at some of the strategies of these firms, that this is rocket science. I think that there should be a presumption with the possibility of the firm making the argument of being exempted. That would be on the discretion of the authority. And then that the authority can say no, and that can be challenged to court. But on the other hand, in such a fast-moving market, you would not have a lock-in. This is a complement, as opposed to a substitute, to the traditional ex-post, which I think needs to be a little bit more subtle and needs to be less rules bound, which is my concern with the DMA Compliance, with what I’ve seen, the Competition Market Authority in the United Kingdom do by way of general example.
I don’t want to speak much about APAC as I’m not a specialist in the area.
JENNY:
So much more of a bespoke system in fact, that’s what–
JACOBIDS:
The bespoke system, but you see, again, we are overintellectualizing the ex-ante versus ex-post. And I think that the ex-ante and the ex-ante with the possibility of some regulatory dialogue, which is a reasonable thing, precisely because the world is changing, is the way to go. I think that once we create that, once we put that in place, we can do the necessary refinements. But there was a lot of scope, I think, for productive innovation in that regards.
JENNY:
So limited and focused ex-ante as a complement to competition.
JACOBIDES:
Absolutely.
JENNY:
Now, let me turn to Marcus Bezzi, because I understand that the ACCC has recommended ex-ante regulation for some of the self-preferencing practices.
What are those practices, and why did you recommend ex-ante regulation?
BEZZI:
I should start, Fred, by saying that the ACCC is a very keen enforcer of our laws. We are enthusiastic enforcers. I personally am a very enthusiastic enforcer, but I understand the limits of ex-post enforcement action, and indeed I think the Google Shopping case is a very strong argument against ex-post enforcement as a way of dealing with market conduct. It’s now really of historical relevance. It deals with conduct that happened in, the complaint, I think, was made in 2010. An [EC] decision was made in 2017. We now get the first level Appeal Court decision in 2021. Who knows when we’ll get a final decision on the matter? It could easily be 15 years after the complaint. What is the value of that to the businesses that made the complaint, who actually don’t exist? A number of them have gone out of business.
You’re right, we at the ACCC do believe that there may need to be carefully crafted ex-ante regulation to deal with specific problems that are identified, and, picking up on Reiko’s point, it should be done very carefully on the basis of thorough evidence which can be gathered.
Competition agencies have got the capacity to make a case and prepare a very sound foundation for it. The UK proposal is very interesting. I think that’s essentially the model that they’re putting in place. Coming up with something limited, fair, targeted on the competition problem, proportionate.
I agree with Michael, it shouldn’t be difficult. We know that as a tool ex-ante, regulation works. We’ve seen it in the telecommunications market.
I remember myself as an enforcer 20 years ago taking cases in the telecommunications market, which were all about trying to break down the power of vertically integrated monopolies. Why would we ignore such an effective tool [as ex ante regulation]?
We have a limited amount of time to deal with the power of these digital platforms. We’re moving onto the next technological development. We’re moving into the Metaverse, we hear, a virtual reality version of the internet. We really need to have proper rules in place to ensure that the very powerful firms that dominate the markets at the moment, don’t completely dominate markets going forward and exclude any possibility for competition, and in the long run, exclude innovation and choice for consumers. That’s really what it’s about at the end of the day.
Look, on my side, one final point: down here in Australia, were a long way from the rest of the world, but we do see one thing very clearly. The greatest possible alignment between jurisdictions is really desirable when it comes to any ex-ante regulation. Now, it may not be complete alignment in terms of the exact regulation, but we need to be aligned about our objectives, aligned about what we’re trying to achieve, to the greatest extent possible. That really means that organizations like the OECD are very, very important in the task that faces us in coming up with some form of [appropriate] ex-ante regulation.
JENNY:
Thank you very much for this.
Let me ask you just a very quick question for a very quick answer, a subsidiary question: who should be in charge of enforcing this ex-ante regulation with all the desirable properties that you’ve mentioned? Should it be the Competition Authority, or should it be a sector specific regulator?
BEZZI:
It might depend on what the regulation is focused on. If at its heart it’s focused on competition, it should be the competition authority, but we’ve done some market studies where we’ve recommended ex-ante regulation, not focused on self-preferencing, focused on other problems where it was entirely appropriate for other regulators to take on that task.
JENNY:
With self-preferencing, would be-
BEZZI:
Self-preferencing, it’s a competition problem, yes.
JENNY:
Okay.
Mr. Ikeda, do you have anything to add? You started this discussion already, but you may have some further comments.
IKEDA:
Yes, thank you.
Again, I am a great supporter of the careful case-by-case analysis as ex-post regulation. But also, at the same time, I have to admit that there are some benefits of ex-ante regulation.
First of all, the enforceability. Speaking of the two cases Fred mentioned at the beginning of our discussion, perhaps Google had the same understanding when they introduced Google map and Google Shopping, but in the latter case, they are perhaps losing billions of Euros in fines, which may be a surprise for them.
Also, as Marcus mentioned, the speed issue. Of course the recent judgment will be a very big contribution to the competition law society, but at the same time, the European Commission needed years to get here, and also they may need some more years until the final decision by the European Court of Justice.
In order to have a proper remedy, the competition authority may take too many years, despite the fact that the online market is so quick and the changing. In that sense, I would echo the conclusion of Michael earlier mentioned; I think it may be a good idea to have a tailored, narrow, limited scope ex-ante regulation, which may enhance enforceability for some Big Tech companies.
Of course, the scope of the company subject to the regulation will be limited to those with the huge market share. Perhaps the those who we call the essential facility in our society. Also, the conduct subject to litigation is limited to some obvious self-preferencing conduct, which is not any part of the debate about the divisions of the conduct. Again, self-preferencing is often associated with the introduction of new technology, so we have to rule out any possibility of the kind of chilling effect caused by the introduction of new technology. Overall, we may introduce some ex-ante regulation in the near future in the competition community, but we cannot expect too much from the ex-ante regulation. That’s my thinking right now.
JENNY:
Well, thank you very much to all of you for what has been a very rich debate in a very compressed time period.
A few things came out, I think, from our discussion. The first one is that we will all learn as we go along. And we know probably know more now about the economics of digital platforms than we did four or five years ago, and people like Michael Jacobides, for example, have contributed hugely to the development of the knowledge of what is specific, what is special, how do ecosystems work, and therefore what are the competition issues which were raised.
The second point was that there was general agreement on the fact that self-preferencing because of the specificities of the digital sector has to be distinguished from the traditional approach to self-preferencing, which is much more threatening in the long-term and could possibly block innovation, which is a very important dimension.
There was a feeling that regulation is useful to complement competition law enforcement because competition law enforcement is rather slow. It’s a very detailed and interesting analysis, but sometimes it is too slow. With the possibilities which were evoked that ex-ante regulation would be finely tailored to a few extreme cases of gatekeepers or central facilities, the possibility that there could be a rebuttal on the part of the firms explaining why in their case, it was absolutely necessary to have self-preferencing, and a question mark on whether it should be competition authorities that should be in charge of assessing whether the rebuttal was correct or not, and justified an exemption from the ex-ante regulation.
I want to add, finally, that Marcus Bezzi’s comment on the fact that all the jurisdictions are faced with exactly the same problem because we are talking about a-territorial behavior by a-territorial firms, and that therefore, there would be a huge benefit to the international community coming to some kind of a basic agreement on what ex-ante regulation would be both something that would promote legal predictability, but also legal and economic convergence, and would be very useful.
And certainly, Marcus referred to the OECD. That’s certainly one of the things that we are interested in at OECD. We have been so far trying to develop our understanding of the issues, and we’ll continue to do so with the hope that we can bring together, at least the OECD members, to agree to some kind of a common understanding of what the solution to the problem could be.
Thank you very much to all of you for participating in this debate, and I hope to see you very soon, either in your country or in some other conferences that I’m sure we will have over the next few years on the same kind of topics.
Thank you very much to all of you.
