Previous article

Google Shopping: a shot in the arm for the EC’s enforcement effort, but how much will it matter?

Introduction [1]

The judgment issued by the EU General Court (“the Court”) on 10 November 2021 [2], dismissing Google’s appeal of the EC’s 2017 decision Google Search (Shopping) (Case AT.39740), [3] is a milestone and a lifeline for DG Competition. Without this endorsement, the whole edifice of Art 102 enforcement (the key European unilateral conduct tool) in the digital space would have been critically undermined. The Court clarifies and even strengthens some of the reasoning in the EC decision, but more importantly it hands the EC a major political victory, appearing to vindicate its approach and record at a time when questions have been raised globally about what has really been achieved by Europe’s antitrust exertions. Having preened itself with being an early starter in the enforcement stakes – while US antitrust had been covered in permafrost since the Microsoft case in 2000 – Europe’s effort had been appearing increasingly hollow of late because in the end it had nothing to show: a handful of decisions against Google, all under lengthy appeal; heavily negotiated remedies on Google’s terms with no impact on the ground; cases against Apple [4] and Amazon [5] taking years to move forward; nothing to show on Facebook. [6]

This article makes three points. First, the Court adopts and therefore establishes the notion that “discriminating in one’s favour” can be anticompetitive, and in what circumstances. And does away with some of the “go-to” defences that had been stalking the land and holding things back in digital enforcement (refusal to supply, product improvement). Does the analysis make economic sense too? Yes, in the circumstances of this case – where the “discrimination in one’s favour” was essentially the means through which Google implemented “input foreclosure”. Second, major discretion is handed to DG Comp (and the national agencies) to find an abuse, which should embolden their antitrust effort in digital. Indeed, I expect it to be used expansively – including by national agencies and regardless of business model – to target anything and everything that can be in any way described as “favouring oneself” (and there are signs of this already [7]). Third, “self-preferencing” is also a major focus of the oncoming European regulatory effort (the Digital Markets Act, “DMA”), [8] though there the formulation of the relevant rule (Art.6.1.d) is quite general. Does the characterization of anticompetitive self-preferencing in the judgment give us something extra to fill that box with? Perhaps, but not much.

Overall, the judgment is good news as we would be in a much worse place if the Court had gone rogue and not handed to DG Comp this essential victory. It contains good reasoning (in places, clearer and tighter than the decision itself) that strips off some cobwebs and is ultimately close to where the economic analysis in the case pointed to. It has some major concessions to a looser standard of evidence, which means defendants in current and future digital cases will be concerned this can open the door to freewheeling theorizing on the part of the Commission (and sister agencies). But in reality, the length of current processes and the difficulty of crafting effective remedies remain the main millstone around European antitrust enforcement and this judgement does not change this. Will regulation move the dial instead?

1. Highlights of the Google Shopping judgement, and does it make economic sense?

Distilling key messages from a Court judgment is a different exercise for an economist than a lawyer. [9] We are unincumbered by the need to constantly refer to precedent and “place” the judgment into the case law (there is very little of this below), and care (or should care) only about whether the reasoning makes economic sense – by which we mean (or should mean) whether the Court’s findings support competition, investment and innovation (to the ultimate benefit of consumers). Through this lens, the judgment maps reasonably well into economic analysis, although it is also fuzzy in places and this creates room for additional discretion which may be good (we need to power up enforcement and need more creative theories of harm) and not-so-good at the same time (giving the regulator licence to come up with feeble narratives without much supporting evidence and calling them “theories of harm”). In my view, there are a few key takeaways.

1.1 Expanding the leveraging toolbox

It is notable that the Court starts with a lengthy discussion of “leveraging”. It makes clear that “leveraging” (using strong assets in one market to create a position in an adjacent market) can happen in various ways, and it confirms the Commission has quite a bit of leeway to come up with various mechanisms through which leveraging can occur: tying, margin squeeze, rebates, bundling, and indeed also discrimination. The Court in effect hands to the Commission more flexibility (away from refusal to supply and indispensability criteria) to frame leveraging narratives in multiple ways, to make sure that every type of anticompetitive conduct by a dominant firm can be potentially and effectively caught under Art. 102. This is good and makes sense. We want theories of harm not to be limited to a finite small number of tried and tested categories, but to be able to formulate and articulate appropriate theories in each case, even if there is no exact precedent to fit the facts.

The Court also confirms that leveraging “per se” is not an abuse, an extension of power could be “on the merits”, and more is needed for a finding that conduct is anticompetitive – even if it leads to disappearance or marginalisation of rivals (para. 162, which is nothing new in fact – already established in Intel and Post Danmark I). Leveraging is recognized as a generic term that can come about as a result of various practices, and only becomes anticompetitive when certain conditions are met (para. 163-164). Fine, but what are these conditions?

Here the judgment gets somewhat fuzzy. The Court essentially says that for a finding of abuse, the conduct must depart from what one would expect to see in conditions of “normal competition”. If the conduct is “abnormal” relative to what one would expect with competition “on the merits”, then it can be found to be anticompetitive. In Shopping, Google’s “self-preferencing” conduct is found to involve abnormality because the Court says that a search engine exposed to competition should be in principle “open” to results from external sources (indeed this is how it generates value) and limiting the scope of results to its own is not necessarily rational. That is, the Court argues that a search engine that was not so dominant would not have promoted its own specialised results over those of rivals, and this “abnormality” is the litmus test for anticompetitive behaviour.

But what is “normal” and what is “abnormal”? {{}}This is hard to say: to an economist, what makes sense is to think in terms of incentives. Is the particular conduct in question profitable? More so than an alternative that was more benign to rivals? And if not, is foreclosure and creating or defending market power the justification for it? Incentives remain the natural way of thinking about how to interpret conduct, and the exercise of asking whether there were profit incentives to adopt a specific conduct is much more fruitful than trying to benchmark it against some hypothetical, subjective “normal”.

Another criterion which the Court cites as relevant to establishing that some conduct can be anticompetitive is whether there was a “change of conduct” associated with the start of the adverse effects. This makes more sense. In this case, the Court finds the negative impact on rivals was all the more obvious as “it followed a change of conduct” on the part of Google: the way Google displayed search results changed when it developed its own comparison shopping service; in addition, Google implemented changes to its Panda algorithm that demoted rival services. These “changes” are difficult to reconcile with “competition on the merit”.

1.2 “Self-preferencing” really means input foreclosure in the Shopping case

Multiple seminars since the 2017 EC decision involved assorted speakers (particularly US-side) who saluted Shopping with skepticism bordering on derision: you silly little Europeans, you don’t know your economics, you don’t know what you are doing. “What does self-preferencing even mean, this EC case is all about fairness and protecting competitors, it is outside of the consumer welfare standard which is the lodestar of all of antitrust”. My response was always that the case was in fact about good old fashioned input foreclosure: diverting to oneself traffic which cannot be replaced inevitably weakens rivals, to the point of marginalisation and exclusion.

Now some (mainly academic lawyers) have pedantically noted there is in fact no actual mention in the Court judgment of “self-preferencing” as such. As if that vanquishes the case. This is just semantics: the Court variously mentions “internal discrimination”, “favouring”, “unjustified difference in treatment”, “promoting” and “demoting”. The theory of harm the Commission pursued was that Google “favoured itself” by directing internet traffic to its own version of comparison shopping services (“CSSs”), and simultaneously reducing traffic to rival services, which could not be replaced with other sources of traffic and led to those rivals’ foreclosure. There is no possible ambiguity here around what was meant by “self-preferencing”. It was input foreclosure. That was the essence of it. And the reason this self-preferencing was anticompetitive is that it was “abnormal”, with a clear “trigger moment” at one point in time.

Again, a discussion of incentives would have fitted better here. Contemporary analyses including by the Commission in the case showed that – because it monetizes its value through ads on its own traffic and not through prices and commissions on third party sales – Google had not just the ability but also the incentive to foreclose. There were incentives for conduct that directed traffic to itself, away from other CSSs, both of a “static” nature (monetizing that traffic through “clicks”) and of a “dynamic” nature (pre-empting future possible challenges). An analysis of “incentives” is apt because it directly maps into business models: the way platform owners monetize the surplus created in these ecosystems matters because it means they have different ways of extracting the surplus created on their platform, and their incentives for conduct vis-à-vis third parties will accordingly differ. [10] Compared to incentives, a benchmark of “normality” is easier to bend to one’s own subjective view of what “normal” means.

1.3 Putting away Bronner’s “refusal to supply” standard in digital

Google had staked much of their appeal in Shopping on the argument that the Commission’s case was a “refusal to supply” case at heart (in particular, refusal to supply rival CSSs access to its own product box and search results), but refusal to supply can only be a competition violation in Europe if it meets the Bronner criteria, and the Commission had not shown that at all. The Bronner precedent [11] has stalked the land for years, as it sets a high bar for refusal to supply (because it is intended to foster infrastructure competition) in a way that is both irrelevant and practically impossible to meet in digital cases (a bit like the Trinko go-to precedent in the US for refusal to deal).

What the Court says here is very significant: (a) Shopping could not be a refusal to supply case because not all access cases attract that treatment, and in this case there was no formal request to supply that could be refused (somewhat legalistic as an argument, not so interesting); more importantly (b) the case is about “internal discrimination” and “difference in treatment” which is different from “refusal to supply” (237, 239); (c) there is no need for the tight Bronner “essential facility” criteria to be met because the objective here is not to preserve the incentives to invest in essential infrastructure; and (d) in any event the Shopping case could not just be a refusal to supply case because let us not forget the Commission’s case also had another limb, i.e. it was not only about self-promotion but also deliberate demotion of CSSs as a result of a change in the algorithm that pushed rival sites “below the fold” – which the Court explicitly calls “relegation” (245). Of these, (c) is especially significant. In cases concerning “refusal to supply” an essential facility, the standard needs to be high because the aim is to preserve incentives to invest in infrastructure, to preserve infrastructure competition, but here no one could realistically be expected to replicate Google Search as a source of traffic. This is a key statement because for the Commission to be nailed to the Bronner standard in each case where the defendant digital giant could argue “I don’t have to supply a service (interoperability, APIs, input) on equivalent terms because I am not caught by Bronner” would mean paralysis.

That said, the judgment also recognizes that Google is a sort of “infrastructure”, a “quasi utility” and de facto “akin to an essential facility” (224-6) as there are no comparable sources of traffic that are “effectively replaceable” as alternatives and “economically viable” (227). Thus, while the Court is emphatic the Commission did not need to show that the Bronner criteria are formally met, the facts of the case get us quite a way in the direction of Google Search being essential and indispensable.

1.4 Putting away the “product improvement” defence

The Court also significantly puts away the “product improvement” defence – the idea that there can be no antitrust violation if a company however dominant simply introduces new features and functionalities within the perimeter of its own product. This was always more of an obstacle to enforcement in the US than in Europe, however it was a defence that Google had relied upon in its Shopping appeal. The Court gives it short shrift: it deals with it under “objective justification” and dispatches it on grounds that even if one is dealing with a product improvement (indeed the Commission did not dispute the procompetitive rationale for developing product universals, 252), this is incapable of offsetting bad conduct. In any event Google should have demonstrated objective necessity or efficiency gains, and it did not. Furthermore, Google did not show how demotions resulted in efficiency gains (575-578), or that alternatives to the conduct were technically impossible.

1.5 Is showing “potential effects” enough?

The Commission conducted an extensive effects analysis in the Shopping case, reported in the decision, which the Court largely approved of – including a very detailed analysis of the impact of Google’s conduct on traffic flowing to rival CSSs (see 388, which considers the comparison of rival CSSs to Google Shopping traffic and shows a decrease in the proportion of traffic to competing CSSs); as well as an analysis of the anticompetitive effects resulting from the reduction of traffic. The Court is satisfied the Commission did a decent job, and did show the effects.

That said, the real interest here is the standard the Court sets for effects analysis. A significant impact of the “more economic approach” (as it has been described) on antitrust enforcement over the past two decades had been an emphasis on the need to document actual foreclosing effects – not just some “shift in market share” which can be consistent with competition on the merits. Economists did push the line that – for instance – price discrimination (or “applying dissimilar conditions to equivalent transactions”) was not a problem in the main, indeed was likely output expanding and pro-competitive, unless it could be established that it brought about foreclosure and exclusion. Or – to give another example – that bundling practices and exclusive dealings are mainly pro-competitive unless one can show material foreclosure effects. Some potential “shift in share” in favour of the defendant was not enough unless one could show a practice was undermining rivals to the point that they were unable to continue to compete and invest for the future.

In truth, multiple decisions and judgments have been leery to endorse this standard. And in Shopping, the Court pulls back quite unequivocally on this. It confirms one would need to show some effects, but there is no requirement to show actual higher prices or less innovation: it is enough to show some “weakening of competition”, or indeed a potential restriction of competition (e.g. 518). It is not necessary to show that rivals are “as efficient” either, or to develop a fully- fledged counterfactual without the conduct.

This is not new, but it remains significant. It makes particular sense in the digital context because we do not need to wait to show full exclusion and “dead bodies” – by which time it is too late for enforcement to have any role. If there is anything that the Shopping case has taught us, it is this: if our concern is underenforcement, and it is, there can be no question that a major cause has been the delay associated with the need to gather ever growing evidence that rivals were imperiled (in addition to major process failures). On the other hand, hopefully this will not mean a licence to be cavalier about effects analysis (on top of incentives), using just a “smell test” and escaping all burden of proof. I discuss this below.

2. Implications for future (and ongoing) digital enforcement cases

The Shopping judgement is not going to make any difference to CSS players and to the rump comparison shopping industry (other than possibly help their future damages claims). As ever in exclusionary conduct cases where the abuse has been effective, and the remedies useless, the victims cannot be revived and regain a decade of lost ground whatever happens: yes, Google won the Shopping war. And this is more important than one may think: CSSs may look in hindsight as a business model that did not get traction, but look at how hybrid e-commerce models are evolving today. It is hard to conclude CSSs may not have been more relevant without the abuse. Still, what will be the broader legacy of this case?

2.1 Strengthening the EC’s enforcement posture?

The first question is what the judgment will do to the enforcement posture of the Commission (and the national authorities). It does away with the nightmare scenario in which the Court would have endorsed Google (there was a realistic possibility this might happen, it was difficult to call ex ante) as a result of which the whole edifice of Art 102 enforcement against digital giants would have been undermined. Having been a well-meaning early mover in the enforcement stakes, the efficacy of the Commission’s enforcement has been (rightly) called into question given the failure to move the dial and to dent the position of Google and others.

The judgment now strengthens the Commission’s hand and should give it confidence on the scope and standards of its enforcement in digital. Most immediately, the judgment arguably provides a blueprint for other search-related cases that have been languishing in the wings for years while “waiting to see what happens to Shopping”. Cases like local search in particular, but other verticals too (travel, hotels, jobs), have seen years of meetings, submissions, more meetings, without any progress – and yet the foreclosure mechanism is arguably the very same (directing traffic to Google’s services and away from others). Let’s see what happens to these, but early signs are the answer may well be “we now will have regulation, the DMA is imminently entering into force, we have other priorities, we have “done” Search”. In other words, a “pass”.

What about other cases? There is quite a bit in the pipeline, though moving slowly. At a high level, as mentioned the Court removes the “refusal to supply” straightjacket that had been a pervasive defence, and encourages an expansive interpretation of Art 102: any type of conduct can be caught – tying, bundling, rebates, discrimination, anything – even practices that do not have a label yet, if one can point to some “abnormality” of the conduct. When combined with the “effects” piece (no need to show actual effects, potential is enough) the implications of the “abnormality” standard can be powerful in policy terms.

In a world where enforcement in the digital space has been weak and feeble, also because of the burden of having to meet high standards and the paralyzing fear of losing on appeal, this is good news. It is important to create room for articulating more relevant theories of harm, around partial foreclosure, raising rivals’ costs, but also exploitation, data extraction, and undermining innovation. Throwing grit in the growth path of potential competitors is a known way of preventing them from acquiring scale and preserving one’s power in digital markets, and a reason we just cannot wait for actual foreclosure (never mind of “as efficient competitors”), or all enforcement will be vacuous and irrelevant.

But foreclosure is (or should be) nothing new, it’s a conventional theory of harm. Multiple other concerns in the digital space fall short of foreclosure: for instance, think of a “hybrid” platform also supplying its own services and generating recommendations, ranking options, determining the visibility of third party offers in ways that may exploit its special knowledge, its access to data, to gain an unfair advantage. Or think of the growth of voice activated devices, with ads and marketing running on them: with consumer impatience and consumer convenience, whichever option is ranked and read out first will get the most clicks. Think of concerns about extracting data from counterparties (compounded by systematic data protection violations) as a condition for operating on the platform, or about forcing the use of certain ancillary services and owning the customer relationship.

Can any of this potentially “exploitative” conduct be also better caught in the wake of this assertive “self-preferencing” ruling?

2.2 Is everything going to be “self-preferencing” now?

The judgment does not imply that every time (for instance) a hybrid platform serves up a recommendation for its own products it is engaging in “self-preferencing”. The “abnormality” benchmark the Court sets out for concern (fuzzy and subjective as it is) essentially says recommendations of one’s own products may well be just fine – if for example the offer from the integrated owner is cheaper/higher quality, or even if it generates more profit for the platform: it is then right that it should dominate third party offers (that would be “normal competition”). But where the owner of the platform puts a “thumb on the scale” by manipulating an invisible algorithm, such that its own offer is featured more prominently each time without being the better one, then there is a problem, and there is a problem even if no one is actually excluded – actually or potentially. It can be exploitation and we should stop pretending it is exclusion. Still a problem. How should one detect and deal with this?

One level of analysis is to gauge “abnormality” with reference to “incentives”. This maps into the business models discussion. There is a growing economic literature on both e-commerce platforms and app stores, for instance, which does look at incentives to foreclose third parties – by charging commissions that are too high, appropriating data/customer relationships, imposing ancillary services, or indeed favouring one’s own version of a product. [12] Some of this work has shown that models with different monetization strategies do not have the same incentives to foreclose rivals. There are multiple externalities that an e-commerce platform needs to internalize, for instance, and business optimization decisions that generate recommendations may reflect a legitimate effort to do so rather than necessarily undermining third party sellers. A platform that monetizes by earning commission revenues on third party sellers does not necessarily always want to sell its own products over those of sellers as it does not necessarily make more money that way. Nor does it want to necessarily undermine the innovation of sellers – to the extent that a bigger assortment and seller innovation increase engagement with the platform in the future. That said, this will also depend on assumptions on entry possibilities, multihoming and so on. There can be no unique presumption either way. There needs to be an understanding of incentives, which must be integrated into the competitive analysis, not just dealt with summarily as an “objective justification” argument that is relegated to the back because it “does not offset bad conduct”.

But we cannot rely on incentive analysis alone. Nor can a regulator be expected to rely just on vast empirical studies of outturns to infer the inner workings of the algorithm that generates the recommendations, for instance. This circumstantial evidence is difficult to generate, expensive to evaluate, and not dispositive. We are moving to a world where agencies will need to examine recommendation algorithms directly. Where there is scope for manipulation in multiple ways, it is hard to rely on something only the defendant has seen. There need to be ways in which the functioning of an algorithm can be explained and its output monitored on a regular basis. Even without full access to the algorithm, there can be options for sandboxed A/B testing. Agencies need to be equipped with these capabilities. Defendants cannot obfuscate and rely on indirect circumstantial evidence. Confidentiality issues and business secrets can be handled. I expect this will become a frequent mainstream discussion, and if this is what it comes to, I would expect transparency with regulators will eventually be seen as preferable to more invasive intervention. We are getting through the stages of denial, but we will get there.

2.3 What about remedies?

Has anything been learned on remedies? The legacy of the Shopping case is that a “cease and desist” decision has translated into multiple “bites of the remedies’ cherry” by Google, with Commissioner Vestager acknowledging that remedies should have been "restorative" and DG Comp should have been more aggressive, while at the same time claiming that the remedies were actually going “well”. [13]

The Court judgment only talks briefly about remedies, suggesting they should have preserved in principle the CSSs’ business model, and not leave them feudally dependent on Google (which is what happened). But the reality of enforcement is that in multiple cases (Google Android most obviously), well-meaning investigations with robust theories of harm have floundered at the last hurdle: designing remedies that could have any effect, being outrun by the defendant’s effort to slow down and hollow out the process. This has been the most persistent and dramatic failure of enforcement. There is a bitter irony in the Shopping case, where the theory of harm was good and the evidence was also good, yet it led to naught because the remedies were incapable of biting. Google, as some have said, won this war.

3. Implications for the Great Regulation Pivot

Will regulation do better? A key reason for the recent major pivot to an ex ante regulatory regime for digital has been the the hope that it could be more effective than antitrust, even “self-executing” – that is, inducing “gatekeepers” to proactively comply with the rules, and thus obviate the need for demanding ex post enforcement and remedies. [14]

“Self-preferencing” is a major theme across all regulatory efforts currently underway. The draft of the DMA provisionally approved by the European Parliament on 23 November 2021 [15] contains specific self-preferencing provisions: Art 6.1.d establishes that a designated gatekeeper must “not treat more favourably in ranking or other settings, (its) services and products (…) compared to similar services or products of third parties, and (must) apply transparent, fair and non-discriminatory conditions to such third party services or products”. As a parallel in the US, the American Innovation and Choice Online Act (also known as the “Self-Preferencing” Bill [16]), introduced recently by Senators Klobuchar and Grassley, also states that it would be unlawful for a “covered platform” to “unfairly preference” its own products and services.

Discrimination is thus unlawful under the new rules if broadly “unfair”. What constitutes “unfair” will however need to be established in each case, and for different business models. The main text of the DMA does include a fleeting reference to the need to reflect business models – Recital 33 contains new language (credit to the IMCO committee) to the effect that “the obligations laid out in the Regulation should take into account the nature of the core platform services provided and the presence of different business models(emphasis added). This is a small nod in the direction of the need to figure out and understand incentives in each case, and I expect it will be relied on significantly by gatekeepers in the implementation of the rules.

Does the Shopping judgment provide a more specific sense of direction for what should be deemed unlawful self-preferencing under the DMA? My sense is that it will be used by gatekeepers to argue that while self-preferencing was found to be problematic in that particular case, the Court is also clear it is only “abnormal” conduct that can give rise to anticompetitive self-preferencing. And even “treating oneself more favourably in ranking” needs not be ‘abnormal’ (if for example it reflects better performance). Regulation is going to be a milestone, but implementation will involve a lot of compliance reports saying “Nothing here, nothing to see” and the “abnormality” benchmark in the Shopping judgment will be an area of contention.

Conclusions

The Shopping judgment is seminal along various dimensions: it sweeps away a number of cobwebs, it contains assertive reasoning, it recognizes “self-preferencing” as nothing else than input foreclosure in the specific case, and above all is a “shot in the arm” for the European enforcement effort – without which we would be in a very difficult place with what is left of the European antitrust “early mover advantage” in digital. That said, the judgment moves no real dial: an entire industry that could have been evolving into new business models is in effect defunct, and the useless remedies agreed with Google do not look as if they will be improved upon. In other words, Google wins here despite the crushing defeat in Court.

DG Comp should now be coming out blinking into the light, forget false modesty (“we must be humble”) and push the enforcement agenda (which is still painfully slow) more decisively (the Google Ad Tech case has been opened in 2021 after everyone else had done one, the Facebook case is still way behind the curve, other cases are years in). Which should also mean that DG Comp should take the latitude the Court is giving it to articulate better theories of harm that fit the concern. Hopefully exploitation can be brought to life, and not remain confined to excess pricing. At the same time, it is to be hoped that – just as not everything in the world needed to look like a Microsoft tying case – not everything in the world that can be vaguely described as “self-preferencing” will be pursued as anticompetitive under the Shopping banner. Sometimes promoting one’s product is rational, fair and good for consumers, not “abnormal”. It depends. Some “case by case” analysis is inevitable.

Finally, the pivot to regulation that owes so much to the failures of antitrust to move the dial suffers from inevitable lack of specificity in the rules. Even with the Shopping judgment, it is not going to be easy to fill provisions like Art. 6.1.d with clear “self-executing” meaning, and in practice we will not be looking at “self-execution” but at enforcement actions for the most egregious cases, following a flurry of “we are compliant!” reports. It is more likely that the intended targets of the regulation will drag their feet in an endless compliance game, rather than adopting pre-emptive adjustments to conduct. [17] That said, the main value of the regulation will be that it is a “signal” to agencies and the courts that the law is now explicitly supportive of cases like Shopping, and of a more expansive use of the tools.

Note from the Editors: although the e-Competitions editors are doing their best to build a comprehensive set of the leading EU and national antitrust cases, the completeness of the database cannot be guaranteed. The present foreword seeks to provide readers with a view of the existing trends based primarily on cases reported in e-Competitions. Readers are welcome to bring any other relevant cases to the attention of the editors.

PDF Version

Author

Quotation

Cristina Caffarra, Google Shopping: a shot in the arm for the EC’s enforcement effort, but how much will it matter?, 13 December 2021, e-Competitions Big Tech & Dominance, Art. N° 104053

Visites 1783

All issues

  • Latest News issue 
  • All News issues
  • Latest Special issue 
  • All Special issues