Like most competition lawyers working in the tech space, I eagerly read the Digital Markets Act (DMA) proposal when it came out on 15 December 2020. My first reaction was positive. As a proponent of the idea that some ex ante regulation is needed to address the concerns created by digital gatekeepers, I was pleased to see that the Commission meant business. After reading the proposal a couple of additional times, my reaction is still largely positive, but with some caveats. Some excellent colleagues have also published their preliminary assessment of the proposal (see, e.g., the piece by C. Caffarra and F. Scott Morton and the paper prepared by Alexandre de Streel and his CERRE colleagues), which offer useful insights.
The question that I would like to address in this blog post is whether the Commission’s proposal can form a sound basis for discussion or whether it needs to be re-conceptualized as it contains too many flaws. In my view, the proposal is a strong starting point and a sound basis for further discussion as it goes through the legislative process. The Commission should be commended for putting together such a comprehensive proposal, while they were under time pressure to deliver, in an area that is complex and where reasonable people may have widely different views.
This does not mean that all elements in the proposal are perfect. The latter contains some weaknesses, which can be addressed with some additional work. My observations focus on three issues: (i) the designation process; (ii) the obligations contained in Articles 5 and 6; and (iii) the implementation and enforcement process.
The designation process
Article 3 of the proposal relies on a combination of qualitative and quantitative criteria to designate digital gatekeepers. In fact, the three overarching qualitative criteria contained in Article 3.1 are presumed to be met when a provider of one or several core platform services (“CPS”) satisfies the various quantitative thresholds set out in Article 3.2. I have not done precise calculations, but it seems that based on these criteria about 10 to 15 digital market players could fall under the scope of the DMA, which is a larger number than I had anticipated.
My concern with this approach is that it focuses on size rather than on whether a given provider of CPS is a gatekeeper, which has been described by Caffarra and Scott Morton as “an intermediary who essentially controls access to critical constituencies on either side of a platform that cannot be reached otherwise, and as a result can engage in conduct and impose rules that counterparties cannot avoid.” Aware of this problem, the proposal provides in Article 3.4 that a provider can escape designation as a gatekeeper if it presents “sufficiently substantiated arguments” to rebut the presumption and shows that one or several qualitative criteria in Article 3.1 are not met in the specific case, account being taken of the elements listed in Article 3.6.
This approach is acceptable provided there is real possibility (not only a theoretical one) to rebut the presumption established in Article 3.2. Otherwise, the DMA could end up regulating companies that have no gatekeeping function, but just happen to be successful and/or big, as indicated by their market capitalisation, turnover and number of users. Conversely, Article 3.6 allows the Commission to designate as a gatekeeper a provider of CPS that meets the qualitative criteria contained in Article 3.1, but does not satisfy all the thresholds of Article 3.2.
As to the list of “elements” contained in points (a) to (f) of Article 3.6, some improvements could be made. For instance, no explicit reference is made to the existence (or absence) of multi-homing (even though one may consider that multi-homing could be examined under element (e), which refers to “business user or end user lock-in“). As I have noted elsewhere, it is hard to see how a platform characterised by multi-homing on both of its sides could be designated as a gatekeeper. When each side enjoys alternatives to deal with the other side, the platform does not have a gatekeeping function. This should definitely be a relevant element.
The obligations contained in Articles 5 and 6
Articles 5 and 6 contain a list of 18 obligations imposed on designated gatekeepers. The difference between the seven obligations contained in Article 5 and the eleven contained in Article 6 is that the latter are “susceptible of being specified further” following a “regulatory dialogue” between the gatekeeper and the Commission.
The above-mentioned CERRE assessment paper refers to the list of obligations contained in Article 5 as a “blacklist” and those contained in Article 6 as a “grey list”. Personally, I see two blacklists considering that the 18 obligations these Articles contain are binding on all designated gatekeepers. It is true that before the proposal was made public there were rumours that the proposal would contain a blacklist and a grey list (as well as a whitelist), but my understanding was that the grey list would contain obligations that would not necessarily apply to all gatekeepers given the existence of different business models.
As clearly shown in the Caffarra/Scott Morton paper, while some obligations try to capture conducts that have been pursued by Apple, others are more clearly linked to Google, and so on and so forth. Thus, it would be preferable that the obligations contained in Article 6 be considered as a menu from which the Commission could select the obligations that would apply to a given gatekeeper. Alternatively, different obligations could be applied to different categories of gatekeepers depending on the CPS offered (e.g., app stores, marketplaces, search engines, operating systems). Applying the exact same obligations to gatekeepers that operate different business models does not seem to make sense. The proposed modification would not require much additional work and would give greater flexibility to both the Commission and the gatekeepers. In this respect, the Commission could learn from the Advice given by the Competition and Markets Authority’s Digital Markets Taskforce to the UK government, which proposes a principle-based approach with principles being targeted to certain activities of SMS firms (for a discussion see here).
Note that Article 8 contains a safety valve. Pursuant to Article 8, the Commission may “exceptionally suspend” a specific obligation laid down in Articles 5 and 6 “where the gatekeeper demonstrates that compliance with that specific obligation would endanger, due to exceptional circumstances beyond the control of the gatekeeper, the economic viability of the operation of the gatekeeper in the Union, and only to the extent necessary to address such a threat to its viability.” While I understand the logic of this suspension procedure, it does not address the concern expressed above. If the imposition of an obligation to a gatekeeper does not make sense given the business model of that gatekeeper, it is better for the Commission not to impose it to that gatekeeper. Otherwise, there is a risk that the Commission will be flooded with suspension applications, which will increase bureaucracy (as they need to be reviewed on a yearly basis).
In my view, the better approach is to turn the Article 6 blacklist into a proper grey list offering more flexibility to the Commission. Alternatively, Article 6 could specify that following the “regulatory dialogue” the Commission could decide that a given obligation does not apply to a given gatekeeper. The idea here is not to weaken the list of the obligations (as I fully support the need to impose obligations on gatekeepers), but to make sure the approach is conceptually sound and to give it greater legitimacy.
Implementation and enforcement
This is of course a critical part of the proposal and perhaps the most worrying. The proposal centralizes implementation and enforcement at the EU level. That makes sense considering that digital gatekeepers are large companies that typically operate across borders. Thus, the Commission will typically be better placed to tackle problematic conducts and impose EU-wide remedies. But this approach can only work if sufficient resources are devoted to implementation and enforcement. Unfortunately, that does not seem to be the case as the proposal suggests that the Commission will devote 80 FTE officials to perform the numerous (and complex) tasks contained in the proposal. That is clearly insufficient. To give a sense of proportion, OFCOM, the UK’s communications regulator, had 901 employees as of 31 March 2019, whereas the remit is limited to one country. 80 FTE officials is an absurdly low number, which should be reconsidered.
Note that as far as resources are concerned, what matters is not only the number of officials, but also their qualifications. While the Commission ranks contain many lawyers and economists, in this case engineers, IT experts and data scientists will be required. This approach is also being pursued by some national competition authorities and national regulators. The Commission should follow suit.
If the Commission is not able to deploy sufficient resources due to budgetary constraints, then I only see two alternatives. One would be to draw resources from the Member States, which would for instance second officials to the Commission. It is not clear that Member States would have incentives to do so considering they have little say in the implementation and enforcement process, their role being limited to participating in the Digital Markets Advisory Committee. The other approach would be to involve Member States in the enforcement process. In the competition law field, national authorities do not hesitate to tackle complex cases involving large tech platforms. They represent an alternative venue to complainants that are told by the Commission, as it regularly happens, that although their complaint is very interesting, they have limited resources and must focus on their priorities.
I should finally emphasize that key to the success of the DMA will be the speed with which it can deal with harmful behaviour. Indeed, one of the key rationales for having a DMA is that competition law proceedings are often too slow to deliver timely relief. This is another reason why the Commission needs to be properly resourced. But I think procedural innovations will also be needed to accelerate processes. The Commission should look at the best practices used by other industry regulators.
The CERRE assessment paper also suggests that the “body” in charge of the implementation and enforcement of the DMA should share the characteristics that the Commission requires from national regulators in the areas of telecoms and energy for instance, and thus be fully independent from “political power”. While in an ideal world that should be the case, it may be too much asking from the Commission. The numerous calls that were made in the past for creating an independent competition agency that would be sheltered from the Commission never led to fruition, and in an area as sensitive as digital markets, the Commission will want to keep some control of the implementation and enforcement processes.
Finally, the CERRE paper suggests that the enforcement process should be cooperative rather than adversarial as is the case in competition law. While cooperation with gatekeepers in the implementation and enforcement process is a not necessarily a bad idea (and the proposal already provides for a “regulatory dialogue” with respect to the implementation of the Article 6 obligations) one should be aware of the risk of capture, especially pronounced if implementation is in the hands of a small number of officials (80 officials means a handful of directors and heads of units), and obfuscation where gatekeepers would use the cooperative approach to delay implementation.
In sum, the Commission DMA is a good document which contains many attractive features, but it could be materially improved by making some targeted changes along the lines discussed above.