The Commission sends Amazon an SO: the rise of common carrier antitrust
As most of you will have seen already, the Commission has sent Amazon a Statement of Objections concerning the use of third party sellers’ data. According to the press release, the authority has come to the preliminary conclusion that this practice amounts to a breach of Article 102 TFEU.
In addition, the Commission has decided to open an investigation concerning the conditions under which third-party retailers gain access to some advantages (namely the so-called ‘buy box’ and Amazon Prime customers).
The two cases are variations on the theme of self-preferencing. In the case of the former, Amazon’s own retail arm would have a competitive advantage. In the case of the latter, third parties using Amazon’s ancillary services would gain an edge over rivals.
These investigations signal the rise of common carrier antitrust. It is an approach to the enforcement of Article 102 TFEU (and EU competition law more generally) that represents, in several key respects, a break from past practice.
The distinctive features of common carrier antitrust are twofold.
- First, the idea that there is something improper, or inherently anticompetitive, in self-preferencing.
- Second, the setting of a low threshold of anticompetitive effects, which would be straightforward to establish in virtually every instance.
Is access to non-public data an expression of competition on the merits?
In its statement of objections, the Commission expresses its preliminary view that Amazon’s use of third-party sellers’ data amounts to an abuse of a dominant position.
The data, which is not publicly available, relates to issues such as ‘the number of ordered and shipped units of products, the sellers’ revenues on the marketplace, the number of visits to sellers’ offers, data relating to shipping, to sellers’ past performance, and other consumer claims on products, including the activated guarantees‘.
The essence of the Commission’s argument is that, by using this data, Amazon would be exploiting its dual role as (i) a marketplace providing services to third-party sellers and (ii) an online retailer.
In particular, the data would give its retail arm a competitive advantage that it would be able to use against rivals.
The press release is remarkable in that it suggests that there is something inherenly anticompetitive in this practice. From this perspective, the use of non-public data from rivals would be an ‘improper’ way of competing. More precisely, the Commission claims that the practice allows Amazon to ‘avoid the normal risks of retail competition‘.
This wording, which is (most intriguingly) vaguely inspired from the definition of concerted practice, signals a new approach to the enforcement of Article 102 TFEU for a number of reasons.
First (and most obviously) it is at least plausible that Amazon’s use of this data improves the conditions of competition (in the marketplace and/or on adjacent markets).
This is so, in particular, if Amazon uses this strategy to challenge the position of well-established players. Why would a practice that is capable of placing competitive pressure on incumbents be an ‘improper’ way of competing?
Second, it is unclear what would turn Amazon’s conduct into an ‘improper’ method. Finding business opportunities by replicating what others are doing well is as old as doing business. Absent a breach of intellectual property, it looks like a most natural expression of competition on the merits.
What is more, it is a commonplace practice in the retail sector (these are long-standing complaints against supermarkets, including the use of non-public data).
What is distinctive, or unique, about Amazon’s behaviour, against this background? Is it the scale? Is it the fact that Amazon is more effective in the gathering and use of third-party data? These questions seem key to making sense of the case.
Finally, one should not forget that all firms (not only Amazon) exploit their advantages. There is nothing inherently anticompetitive in doing so. On the contrary.
In fact, the whole purpose of competition law has never been to create a level playing field in which firms compete with the exact same forces and assets. The point of competition law is instead to ensure that firms retain the ability and incentive to make the most of what they have.
Against this background, the question that comes to mind is whether it is possible to distinguish, in a meaningful way, between competitive advantages that can and cannot be exploited or between competitive advantages that are proper and improper.
What about anticompetitive effects?
The press release suggests that the use of non-public data would allow Amazon to ‘leverage its dominance in the market for the provision of marketplace services‘.
Remarkably, however, the press release is not explicit about the markets that would be affected by the practice. Insofar as Amazon’s conduct is capable of improving the conditions of competition and of injecting rivalry, one would expect the issue of anticompetitive effects to feature prominently as the crucial inquiry in the case.
The wording of the press release raises the question of how the analysis of anticompetitive effects will be conducted by the Commission. A second, related one, is whether the analysis will be relaxed relative to the case law.
More precisely: will the Commission equate a competitive advantage with anticompetitive effects? The case law suggests that this tendency is not unusual in competition authorities. This same case law makes clear that an advantage does not amount, in and of itself, to an anticompetitive effect.
In addition: will the threshold be set at the level of plausibility (as opposed to likelihood)? If so, the analysis of effects would become a mere formality (potentially abusive conduct is most of the time a plausible source of anticompetitive effects).
My impression is that common carrier antitrust tends to lead to a relaxation of this assessment. Under this sui generis approach to Article 102 TFEU, any distortions of competition that come from self-preferencing are deemed problematic, even when they intensify rivalry.
It will be fascinating to see whether this is indeed the path followed.
I very much look forward to your comments. If there was any doubt: I have nothing to disclose.
Chillin'Competition 
It looks to me that self-preferencing is never the main business model. Only peripheral. Just another opportunity to earn extra-penny, as brick and mortar supermarkets do with private labelling. Otherwise, the platform risks to lose its objectively-neutral status. Of course, being an incumbent, you can go far enough, but only up to a point. You go a step further, changing the equilibrium of tolerance – and your customers (not consumers but customers – retailers and advertisers) would begin shopping around. Or would begin thinking of shopping around. Or would begin feeling uncomfortable and mumbling about good old days when the grass was greener and GAFAM were just great and generic – and again this would be the first step for shopping around. So on one hand, self-preferencing can indeed manifest the features of abusive conduct, but on the other, it is a Pandora box of cannibalisation, and incumbents are well-aware of it. On the “third” hand yet even such a peripheral activity in such a tough game is likely to bring extra billion or two of fines (aka costs) and a huge negative publicity: one stuck the neck out – and got the punch.
Oles
10 November 2020 at 5:57 pm
Thanks for the interesting take on this. I recall when the probe was originally opened in July 2019 the EC stated that it was looking at these practices under both Article 101 and 102. I would be interested in your thoughts on why the case is now being brought solely as a 102 case.
Charlie
10 November 2020 at 7:24 pm
Thanks! I suspect that the implications of following the Article 101 TFEU route would have been even more significant. Just imagine the consequences for supermarkets, which tend to have access to the same sort of information by means of an agreement with suppliers
Pablo Ibanez Colomo
11 November 2020 at 12:15 pm
The reason for not going down the Article 101 TFEU route may be the lack of agreement of the third party selling on Amazon’s (A) platform to A using data in respect of its products sold on A’s platform to compete with it. Cf. CJEU’s judgement in Bayer / Adalat (Joined cases C-201 P and C-301 P).
Joachim
19 November 2020 at 11:37 am
Thanks, Joachim! I understand from the publicly available information that online retailers agree to supply this data to Amazon. Article 101 TFEU would thus be fully applicable.
Pablo Ibanez Colomo
19 November 2020 at 2:13 pm
I suppose in the end it has to do with Amazon’s dominant position: first, because it will lead to a gradual disappearance of competitors as private labels win it all. Second, because those same competitors have no choice but to stay with Amazon.
Javi
11 November 2020 at 9:11 am
Thanks so much for the comment!
If the case had been framed in such a way, it would have been far less controversial. It would follow the well-trodden path of indispensability + exclusionary effects. I guess we will have to wait and see
Pablo Ibanez Colomo
11 November 2020 at 12:17 pm
At first glance this looks like a step even beyond Google Shopping. I could understand their case if Amazon were using it’s control of the platform to advantage its own products in terms of search results and product positioning, as this is a clear competitive advantage over rivals.
However, it seems Amazon’s “advantage” is just in knowing which products sell well and which do not. This benefits Amazon because it has more precise knowledge of which products to offer. But where is the competitive advantage over rivals? Once Amazon has made the decision to offer its own private label, it will be selling on the same terms as third-parties which already produce that product. Rivals may exit or be harmed due to increased competition, but unless Amazon is securing dissillimar selling conditions for itself (ala G Shopping), there’s a missing link in the theory of harm.
Tom
11 November 2020 at 11:20 am
Thanks, Tom! I guess the advantage comes from the fact that Amazon can piggyback on third parties’ efforts to figure out what works and sells well.
Pablo Ibanez Colomo
11 November 2020 at 12:19 pm
Thanks Pablo for a very enlightening post, as always. The comparison with supermarkets is interesting to understand that the use of such data is a relatively normal practice, but its relevance likely remains limited because Delhaize, Carrefour or Auchan do not have a dominant position, let alone a super-dominant one. On the latter point, it might be that the common carrier trend you rightly identify applies to super-dominant companies which (once the market has tipped, and assuming of course that the market is actually prone to such tipping) are in a situation which — as a matter of fact — amounts to the granting of special rights under Article 106 TFEU. If the theory of harm is sufficiently constrained along these lines, its consequences may not be so dramatic. The law on 102 TFEU does not have to be binary, i.e., it does not have to be an all-or-nothing game depending on whether one has (or not) a dominant position.
Bino
12 November 2020 at 10:21 am
Thanks, Bino for these comments. I am also grateful, to you and the rest of you commenting, for the polite and constructive exchange of ideas.
Your comments, Bino, identify the key aspects of the discussion.
On supermarkets, I agree with you that a supermarket chain is rarely ever dominant. The interest point, however, is that the press release appears to claim that Amazon’s behaviour is inherently anticompetitive. If it is inherently anticompetitive, the degree of market power would not be decisive (unless one claims that a practice is only inherently anticompetitive when implemented by a dominant firm, which comes across as unpersuasive).
The reference to Article 106 TFEU and exclusive rights is really spot on. It is very much in line with my reference to common carriers. The question, I guess, is how we draw the line and which criteria we use to identify dominant firms with common carrier status. Indispensability used to be the criterion. No longer (or so it seems). I wonder whether a meaningful alternative can be found
Pablo Ibanez Colomo
12 November 2020 at 11:10 am
There would be a rather big assumption on super-dominance. Shopify, which offers a somewhat different business proposition to sellers, is currently outgrowing Amazon by a significant margin. Network effects tip both ways, and first mover advantages diminish over time. https://www.forbes.com/sites/walterloeb/2020/10/26/shopify-is-a-challenger-to-amazon-every-day/
Kay
12 November 2020 at 1:55 pm
Great post and very important to have this discussion on the new “fairness” obligation DG COMP is toying with. Two points:
There are tons of marketing services companies that advise retailers/manufacturers on product development / pricing using non-public data, and even some that focus those services to sellers that use Amazon’s marketplace services. Why is it unfair for Amazon to do the same?
If it’s because this is data that Amazon has gathered through sellers’ use of its platform, why shouldn’t Amazon get to use this data (with consent)? Even the Crémer Report recognises that use of such aggregated data makes markets more competitive (pg. 68). Should competition law be trying to remove competitive advantages to make markets “fair”?
Kay
12 November 2020 at 1:59 pm
Pablo, it seems that the supermarket/Amazon comparison is becoming trendy in the US and in the EU. In the US, Stapp and Evans raise the analogy to oppose platform regulation. In the EU, the Expert Group’s self-preferencing report claims that platforms are more dangerous than supermarkets in order to pave the way for regulation. The Experts overlook that supermarkets’ UTPs are regulated in most MS as competitive bottlenecks and Directive 2019/633 goes in the same direction. I stand by my comments to your previous piece “On the Amazon probe: neutrality everywhere…”. I believe we are in the realm of two-sided competitive bottlenecks (as per Mark Armstrong’s seminal paper, several examples, including the old CRS Regulation, discussed in my paper: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2401723) and competition law is of little or no use (absent highly “imaginative/unorthodox” market definitions). We are dealing with fair competition concepts that may distort the competitive process (Section 5 FTC, New Competition Tool, UK Market investigations, etc.). By the way, to set the record straight: Amazon’s alleged abuse is a joke compared to life in grocery retailing: suppliers share with national leading retailers their most precious trade secrets (innovation launches, including product features, packaging, pricing, marketing, testing, etc.) months before the launch with the hope of being listed and not a single retailer in Europe (and the world?) operates any Chinese wall or contractual commitment not to use it in favour of its own competing brand. Often, the innovations have been copycat and not even listed. Amazon has not replaced these offline bottlenecks, which also lead online grocery retailing, it has or will become one of them (see Japan’s competition authority’s UTP claims against Amazon retailing arm vis-à-vis its suppliers, settled by the company).
J&B
15 November 2020 at 9:48 am
Thanks, J&B! Your comments are always appreciated (and have been missed!)
Both comments are incredibly helpful to put this discussion in perspective.
It is very important that we bear in mind that, as you mention, supermarkets have long gone very far in terms of obtaining non-public data. If Amazon’s practices are deemed inherently anticompetitive, so should the supermarket chains’. It is just not clear that, from a competition law perspective, this is really something that is inherently against competition on the merits.
On the other hand, you mention that there is sector-specific regulation applying to these practices. It is certainly legitimate for democratically elected bodies want to go further and impose duties where competition law would impose none. The fact that such sector-specific regulation exists does not necessarily mean there is a competition law issue, however (and often suggests the opposite).
Pablo Ibanez Colomo
15 November 2020 at 6:04 pm
Too broaden the perspective : there is some case-law from NCAs finding that the use by a (former) monopolist of data gathered in a (formerly) reserved market to compete in that market (once opened up for competition) or in a neighbouring market amounts to abuse of a dominant position (e.g. France : 100 million euro fine imposed on Engie in 2017; Belgium : 1.2 million euro fine imposed on the Belgian national lottery in 2015 – both settlement decisions). I could imagine the EC developing a similar reasoning, provided it can demonstrate that Amazon is ‘super’ dominant. To echo comments of Oles and Kay, it is difficult to understand why an online marketplace would let go its objective / neutral status and risk losing customers as a result. Unless, of course, it is ‘super’ dominant.
Joachim
19 November 2020 at 12:19 pm
Thanks again, Joachim!
My impression is that there are plausible reasons why vertically-integrated firms choose not to be neutral or objective. As mentioned by several commentators here, supermarkets are not neutral and they are not dominant, let alone superdominant. Still, they engage in similar tactics to favour their private labels and piggyback on the efforts made by suppliers.
Superdominance, by the way, is not a concept recognised by the Court of Justice. One interesting question is whether Amazon is an indispensable platform, in which case the precedents on former monopolies would be clearly relevant. It does not appear, however, that claims of indispensability are made (nor does it seem that they would be credible if made).
Pablo Ibanez Colomo
19 November 2020 at 2:22 pm
And what about remedy? Amazon must offer marketplace services to sellers who don’t want Amazon to use aggregated data created from those sellers’ use of the marketplace services (surely at higher cost as compared to the sellers that agree to use of the data)? Amazon must offer the aggregated marketing data services on FRAND terms (essential facility)? Looking forward to Part II!
Kay
17 November 2020 at 8:50 pm
Pablo/Joachin, in my humble opinion, your comments exhibit what I would call an endogeneity bias, you assume that competition law-defined dominance is the parameter of abuse and conclude that a a non-dominant company would not self-prefer its products because competition is sufficiently strong. However, the competition law-defined horizontal dominance does not cover oligopolistic scenarios and/or vertical dominance, i.e., where a few horizontally non-dominant platforms may enjoy bargaining power over all suppliers/business users. Real life shows that in this scenario leading competitors have all the incentives to vertically integrate for a variety of reasons such as reducing horizontal competition (differentiation/consumer lock-in) or simply avoid being bypassed by their suppliers (see “A primer on vertical foreclosure” for different justifications). Long ago, airlines integrated into CRS for the same reason and once self-preference was prohibited the economic incentive for vertical integration disappeared (non-mandated divestitures ensued). Leading credit card networks are not dominant and they can abuse their merchants (for the benefit or not of card users is a different question) and could self-prefer were them to be vertically integrated. Supermarkets did not grow up relying on vertical foreclosure but “discovered” this line of business once they were vertically dominant and do not go for an all or nothing bet but adopt a creeping foreclosure approach brand by brand (including trial and error). From a fairness standpoint the question is whether a platform that has achieved a vertical dominance status based on a given business model may then ABUSE a dual role. Note that I do not question vertical integration or meritocratic competition but the abuse of the dual role (indeed, what is an unfair use of the dual role is the key as Kai mentioned, but this issue arises in any regulation on competitive bottlenecks). I would assume that in your reasoning, internet neutrality would also not be justified because non-dominant ISPs have no incentive to discriminate and dominant ones however we define then) are prevented from doing it (i.e., the US FCC approach under R-Ajit Pai), is that correct?
Disclaimer: I have no professional link with the online sector (platform/users) and I represent an association of consumer goods suppliers in Spain. As Pablo knows well, I first developed my views and exposed them in a report whilst working at a competition authority, (“Distribution of daily consumer goods: competition, oligopoly and tacit collusion”, 2009, available at the web of the Basque Competition Authority, http://www.competencia.euskadi.eus), long before my attorney-client bias 😉
PS – It is very enriching and satisfactory to be able to exchange views with all of you, great minds!
J&B
20 November 2020 at 12:02 pm
Thanks so much again, J&B. Great thoughts as usual – and an effective way of identifying the key issues.
I fully agree with you that self-preferencing does not magically disappear below the threshold of dominance. As discussed with Joachim, supermarket chains are rarely ever dominant within the meaning of Article 102 TFEU, yet they engage in similar (arguably more far-reaching tactics).
The example of net neutrality is ideal. The issue with net neutrality is not that ISPs lack the incentive to discriminate because they are not dominant. I accept that they may well have an incentive to discriminate event when they are not. The question is rather what justifies introducing strict non-discrimination obligations that do not exist elsewhere in the economy and whether there is robust and reliable evidence suggesting that such obligations improve the functioning of markets.
Pablo Ibanez Colomo
23 November 2020 at 1:03 pm
Thanks J&B and Pablo for the fascinating discussion.
I think we agree that any unilateral conduct by a vertically integrated firm that is not dominant on any upstream or downstream market in which it operates is legal under Article 101 TFEU whereas such conduct may amount to an abuse under Article 102 TFEU if the said firm were dominant in one (or several) of the above-mentioned markets.
That being said, as you both pointed out, there are situations where non-dominant companies may enjoy bargaining power over their suppliers and may be tempted to ‘abuse’ that power. This has led various Member States to prohibit (certain forms of) abuse of economic dependency (France, Germany, etc.), the scope of which varies between MS. As J&B pointed out, Directive 2019/633 goes in the same direction. I would hope that in adopting such legislation, each legislator did ask itself – to echo Pablo’s question – whether there is robust and reliable evidence suggesting that such obligations / prohibition improve the functioning of markets and ultimately benefit consumers ?
Coming back to Amazon (A) and Article 102 TFEU : one should not overlook the fact that, unlike supermarkets, third-party merchants sell their goods directly on A’s marketplace. Put differently, A does not purchase third-party merchants’ products to sell them on to consumers. To me, this difference is fundamental: unlike supermarkets, A’s retail arm does not need to have access to information about the sales by third-party merchants on A’s marketplace.
My take: the EC is trying to demonstrate (i) that A’s retail arm access to – and use of – third-party merchants’ non-public information re. said merchants’ sales on A’s platform is a means other than those that come within the scope of competition on the merits as well as (ii) the existence of sufficiently appreciable potential or real effects on competition as a result of said practice.
I look forward to your comments.
Joachim
23 November 2020 at 5:14 pm
Joachim/Pablo,
Re: exceptional intervention. The regulation of dual role conflicts (including, but not limited to the vertical foreclosure-consumer harm angle) is not exceptional as you assume but rather common in concentrated (oligopolistic?) and two-sided competitive bottleneck settings. I mentioned CRS (dating back to Regulation 2299/89, I like the reference to “Code of Conduct”…exactly the same term applied to the UK UTP regulation on supermarkets), Internet neutrality, credit card networks (note the mandatory separation of card scheme and processing functions) and Supermarkets (note the prohibition of the misuse of trade secrets). In the financial sector, FRAND rules apply in different areas (Regulation 2016/1011 on benchmarks, Directive 2014/65 on markets), the separation of investment and commercial banking (e.g., Glass-Steagall in the US), rules on financial advice (MiFID II in the EU). in Spain, vertical integration restrictions apply in the pharma (pharma industry cannot own pharmacies and oil sectors (oil wholesalers cannot own >30% of retail oil stations). Relatedly, network access has been mandated in sectors such as media (“must carry” rules), mobile communications (note the “collective dominance” assessment in order to impose MVNO access, even in the presence of 4 MNOs). Even DG COMP has leveraged sometimes its quasi-regulatory functions in the block exemption field to mandate “guest beer” and “guest car” in beer and car distribution networks.
Re: Need for “robust and reliable evidence”, I am afraid back in 1989 there wasn´t any as far as CRS vertical integration was concerned. The same goes for all the regulated sectors (Indeed, the Commission has conceded that self-preferencing by a dominant CRS could not be considered even abusive in this area (tip for Google lawyers ;-). When you approach this issue from an unfair (method of) competition standpoint you admit you are making a value judgment on what is fair/unfair and meritocratic/unmeritocratic competition certainly. But I am a bit tired of the competition POLICY cheap talk about elusive goals (competitive process/consumer welfare) which are not measurable and reflect values as much as fair competition rules. The moral superiority of the competition law community in this field reminds me of the naked emperor. My words are worth little or nothing, but I would refer you to the words of the late Harold Demsetz “economics has no antitrust relevant theory of competition” (quoted in Josh Wright’s “Antitrust, Multi-Dimensional Competition, and Innovation: Do We Have an Antitrust-Relevant Theory of Competition Now?”). To me the most enlightened reflection on competition policy I have read in my 23 career-years. He made good Benedetti’s saying “a pessimist is only a well-informed optimist” 😉
Re: marketplace/retailer duality v. retailer/brand owner duality. I see no difference. Indeed, Amazon also exhibits the second duality in its retail arm and the conflict is the same or even more acute in the retail arm. The fact that a retailer leverages all the information gathered from suppliers (many months prior to the commercial launch of the product) to benefit its own brand is a deliberate decision not necessary to run its retail arm. In the marketplace retail pricing is under control of the merchant (even though access fees may certainly impact on it), in the retail arm, a competitor controls the retail pricing of all other competitors. Leaving aside the specificities of the case, my take is that it is an attempt to show that competition law is relevant in the platform world (an uphill battle) but my personal reading is that it is a further evidence of the shaky foundations of market definition, dominance and competitive process/consumer welfare analysis in competition law. The investigation focuses on two Member States only because of the market definition/market share (dominance) hurdle. My question is whether the Commission considers Amazon’s conduct to be pro-competitive in the remaining Member States or whether it consideres it abusive/unfair but outside the competition law formalistic boundaries.
All in all, I wonder if the Commission is taking due stock of its own experience dealing with competitive bottlenecks in many sectors and adopting a holistic approach or it is just embarked on an ad hoc response to a perceived new paradigm (online gatekeepers). Interestingly, the Commission’s Green paper on UTPs in the food and non-food supply chain (2013), though highly influenced by the supermarket bottleneck issue, could have been the first attempt to address competitive bottlenecks across-sectors. However, it never progressed further.
J&B
26 November 2020 at 1:26 pm