A step back to look ahead: mapping coalitions on data flows and platform regulation in the Council of the EU (2016-2019)

: As new legislative battles arise on the EU digital policy front, this article draws on recent EU negotiations related to the making of the Digital Single Market (DSM) to document the main lines of division between member states in this field, as well as coalition patterns in the Council of the EU. This research draws on a novel data set on the policy preferences and salience of member states in key controversial negotiations related to the regulation of data flows and internet platforms. It provides relevant analytical tools to approach future EU digital policy-making processes by investigating recent coalition patterns in the Council and beyond.


Section I: Introduction
The objective of this article is to inform our understanding of upcoming policy developments with novel data on the policy-making process of recent EU Digital Single Market (DSM) legislative files. This study draws on a new data set collected as part of a multi-year research programme on the power of member states in negotiations of the Council of the EU (or "Council"). This novel data set includes information on the initial policy preferences and issue salience of all member states and EU institutions on the main controversial issues of the following legislative negotiations: the regulation on the free flow of non-personal data, the European electronic communication code directive, and the directive on copyright in the DSM.
During these negotiations, member states have discussed extensively the opportunity to (de-)regulate data flows, and introduce new (legal and financial) obligations on internet platforms. These policy controversies will remain at the centre of the EU's political agenda for years to come, with the launch of negotiations on the Data Governance Act (DGA), Digital Services Act (DSA) and Digital Markets Act (DMA).
These new legislative processes need to be seen as a continuation of previous EU negotiations, and would thus gain at being understood in light of the coalition patterns previously mobilised by member states in the Council. Often considered as the most powerful institution of the EU legislative system, the Council is however known for the opacity of its policy-making processes, which has greatly limited academic attempts to uncover its inner-workings (Naurin & Wallace, 2008). This research explores this black box, building on a public data set (Arregui & Perarnaud, 2021) based on 145 interviews conducted in Brussels with Council negotiators and EU officials between 2016 and 2020.
By highlighting the main controversies and coalition patterns between member states on the regulation of data flows and internet platforms as part of three negotiations, this research provides relevant analytical tools to approach future EU digital policy-making processes. It underlines in particular how the ability of certain member states to form and maintain coalitions may determine decision outcomes.
Given its success in the adoption process of the free flow of data regulation, the "digital like-minded group" (or the D9+ group) could indeed be activated in the course of the next negotiations on the DGA, DSA and DMA. This paper also argues that the capacity of large member states, such as Germany and France, to formulate their policy preferences early in the process could be a key determinant of their bargaining success. Moreover, while the Council is expected to remain strongly divided regarding the regulation of data flows and internet platforms, this paper indicates why the European Parliament (EP) could have a significant role in these discussions. It also signals that member states are not equal in their capacity to engage with Members of the European Parliament (MEPs), thus suggesting that the ones with more structured channels of engagement with the EP may be more likely to be successful in upcoming negotiations.
The following section offers a brief literature review on EU negotiations and Council policy-making processes in relation to the DSM. Then, the methodological approach and data set are presented. The empirical section is divided in two parts.
The first maps the constellation of preferences and issue salience of member states in the three legislative files, and the second uncovers the main coalition patterns. The findings are then discussed in light of their implications for upcoming EU negotiations.

Section II: Unpacking EU digital negotiations
In recent years, EU digital policies have attracted a growing attention from scholars, partly due to the acceleration of EU legislative activities on key internet-related issues, such as data protection and cybersecurity. This trend reflects a broader pattern of states' increasing engagement with internet governance and policymaking to exercise power in and through cyberspace (Deibert & Crete-Nishihata, 2012;DeNardis, 2014;Harcourt et al., 2020;Radu et al., 2021).
The literature has acknowledged the increasingly active role of the EU in internet policies, as illustrated by recent regulatory and policy initiatives in the fields of data governance (Borgogno & Colangelo, 2019), privacy (Ochs et al., 2016;Bennett & Raab, 2020;Laurer & Seidl, 2020), copyright (Meyer, 2017;Schroff & Street, 2018) and cybersecurity (Christou, 2019). Recent research has also investigated the nature and determinants of a 'European approach' in regulating large internet companies and safeguarding competition (Radu & Chenou, 2015), in balancing competing policy objectives such as national security and data protection (Dimitrova & Brkan, 2018), or in promoting its 'digital sovereignty' (Pohle & Thiel, 2020). EU decisions have direct implications for member states, companies and citizens, but also for third countries (Bradford, 2020), as illustrated by the recent reform of the EU data protection framework (Bendiek & Römer, 2018). Reflecting its ambition to increase its 'cyber power' (Cavelty, 2018) on the international stage, the EU has progressively established cyber partnerships with third countries to engage on digital issues (Renard, 2018), building on the recognised role of the EU over the past two decades in public policy aspects of internet governance (Christou & Simpson, 2006). But EU digital policy can also be seen as a field of struggle (Pohle et al., 2016), with major divides among governments. As emphasised by Timmers in the case of EU cybersecurity policies, the wide diversity of interests of member states can be challenging for EU policy-making (Timmers, 2018) and thus usher in competing political dynamics in the Council. Though the literature provides refined accounts of the discourse and role of the EU in internet governance debates, more limited are political scientists' attempts to unpack the complex political processes and controversies structuring EU digital policies, and identify the "winners and losers" of policy developments from the perspective of national governments.
Exploring this gap, this study draws on recent research on the decision-making system of the EU, grounded in rational choice institutionalist analysis. Due to the intergovernmental design of the Council, a great part of the scholarship pertaining to the Council embraces a rationalist perspective, giving to national decision-makers the lead role and assuming that member states determine their actions according to their national preferences and own calculation of utility (Naurin & Wallace, 2008). This scholarship assumes that negotiations' outcomes are shaped by strategic interactions between goal-seeking governments, with bounded rationality, operating within a set of institutional constraints (Lundgren et al., 2019).
It is common knowledge that a major part of the decisions in the Council is adopted by consensus, making the emphasis on the voting phase of the decision making process less relevant than the bargaining phase. This is the reason why recent research on member states' influence in the EU decision-making system focuses on the actual negotiation processes at play (Thomson et al., 2006;Thomson, 2011).
This scholarship is primarily driven by the Decision-making in the European Union (DEU) project, followed recently by the Economic and Monetary Union (EMU) Positions dataset (Wasserfallen et al., 2019), which led to more generalisable findings on power distribution and bargaining processes in the Council. I will argue that these analytical tools are much welcome to approach EU negotiations on DSM policies. They offer an established methodology to map the constellation of preferences on key controversial issues, and document the determinants and patterns of coalitions in the Council. The following section describes the methodological steps taken to collect the data on which this article draws its analysis upon, as well as the structure of the data set.

Section III: Data and methodology
This research draws on a new data set documenting recent EU legislative processes, and covering the initial preferences of member states and EU institutions on controversial policy issues, as well as their decision outcomes. The DEU III data set (Arregui & Perarnaud, 2021)  In this data set, information on actors' policy positions and their salience is represented spatially using 'scales' according to an established methodology (Thomson et al., 2006;Arregui & Perarnaud, 2021). During face-to-face interviews conducted in Brussels, negotiators were asked to identify the main controversies raised among member states once the Commission had introduced the legislative proposal. Subsequently, the policy experts had to locate the positions of all actors along the policy scale. The experts were also asked to estimate the level of salience that actors attached to each controversial issue. Every estimation provided had to be justified through evidence and substantive arguments. A number of validity and reliability tests on the DEU III data set (for instance by systematically comparing ex-perts´ judgments and documents) have corroborated previous analysis of validity and reliability of the DEU I data set made by Thomson et al. (2006).
This methodological approach has its own weaknesses, already identified in the literature (Princen, 2012). In relation to the study of actors' influence, the first limitation is the clear focus of the DEU data set on the negotiation phase of the EU policy cycle. As the literature on agenda-setting (Princen, 2009) suggests, national governments and other stakeholders can invest significant resources into the preparatory steps of legislative negotiations, dynamics that are not thoroughly addressed in the data set. Similarly, this data set does not investigate the implementation of EU legislations and the actual compliance of member states with decision outcomes. Yet, we know that governments can repeat the same influence efforts observed during negotiations once legislative decisions are actually adopted (Blom-Hansen, 2014). Though not immune from traditional biases related to ex-pert surveys and spatial models of politics, this data set allows us to analyse and compare the structure of the constellation of preferences in the Council across the range of policy controversies under study. In a prior analysis of this data set, Perarnaud (forthcoming) shows that bargaining success is unevenly distributed among member states when looking at DSM negotiations, and suggests that these asymmetries relate in part to variations in the resources and coordination mechanisms that can be mobilised by national governments in Brussels. Instead, this research follows a policy-oriented approach and focuses on the political controversies that shaped these negotiations, in view to inform our analysis of upcoming legislative processes.
The following section presents the main policy controversies under study. The analysis focuses in particular on three overarching issues: the regulation of data flows, the introduction of new legal obligations for internet platforms, and new financial obligations for internet platforms. Then, the coalition patterns observed in these negotiations are described, laying an emphasis on the 'digital like-minded group' in the Council.

Section IV: Analysis
The three EU negotiations illustrate the extent to which member states were divided regarding the regulation of data flows, the introduction of new legal obligations for internet platforms, as well as new financial obligations for internet platforms.

Data flows
The adoption process of the regulation on the free flow of non-personal data in the EU was characterised by sharp divisions in the Council. This is well illustrated by numerous protracted influence efforts led by several member states, even prior to the publication of the legislative proposal by the European Commission in September 2017.
Before the actual launch of its proposal, certain member states had repeatedly France and Germany were initially opposed to this initiative. After the legislative proposal was internally drafted by the services of the European Commission, France was reportedly concerned about the scope of the draft proposal and pressured the Commission to delay its official publication, and successfully so between 2016 and 2017 2 . Indeed, the Regulatory Scrutiny Board (RSB), the body with the power to issue opinions to the college of commissioners on draft legislations, adopted two consecutive negative opinions on the Commission draft proposal in 2016 and 2017. One of the main arguments given to justify these two negative opinions was that the draft proposal also covered personal data, and thus overlapped with the newly adopted General Data Protection Regulation (GDPR).
After the negotiations officially started, the main controversial issue opposing member states consisted in the scope of the derogations which could interfere with the principle of free flow of non-personal data. While some advocated for only very limited derogations to the principle of free flow of data, others favoured extensive derogations for different purposes (security, culture, public archives). Proponents of the principle of the free flow of data wanted to maintain the scope of the regulation as broad as possible, as envisioned initially by the European Commission. Estonia, Denmark, Ireland, Czech Republic, Poland, the United Kingdom, Netherlands and Sweden were the most active member states in this group. These member states coordinated their influence efforts at the Brussels-level as part of the "digital like-minded group", a coalition group that will be further detailed in the following sections. Though sharing a common goal, these member states were driven by different motives. For instance, for Estonia, one of the main driving forces behind this reform, the free flow of data was both an economic and a political priority. Due to its very digitalised economy and society, the Estonian government saw an economic and political interest in supporting more data transfers within the EU, while increasing its resilience against potential third countries' attacks over its data infrastructures 3 . Interest groups representing the tech industry in Brussels were also advocating for the suppression of data localisation obstacles (DIGITALEUROPE, 2017). The free flow of data in the EU was considered as a very positive change for most internet platforms, as it could lead towards the formal establishment of the free flow of (non-personal) data between the EU and third countries (such as the United States).  only for a limited set of provisions.
Member states were strongly divided on the extension of telecom rules to OTT service providers. For instance, France and Spain had a maximalist approach and wanted to introduce wide requirements for OTTs. Germany and Poland supported a similar stance, but with less salience. Spain even proposed that these services should be covered by the telecom general authorisation regime in each member state, like any other telecom providers 5 . On the other side of the spectrum, several member states (Sweden, Finland, Denmark, Czech Republic, Netherlands, Luxembourg, Ireland, United Kingdom and Belgium) claimed that regulating number-independent services was unjustified and could hamper innovation in the EU. In its proposal, the European Commission had intended to make a step in the direction of France and Spain, without alienating other member states initially opposed to this inclusion 6 .
The adoption process of the copyright directive, and in particular of its article 17 (formerly known as 'article 13'), generated even more divides between member states (Bridy, 2020;Dusollier, 2020), and is also relevant to map the constellation of member states' preferences in relation to the regulation of internet platforms.
Member states were opposed regarding the obligations and rules that internet platforms should follow to protect rights holders' content. They had different views on how to address right holders' challenges to prevent copyright infringements online, and on the type of obligations and liabilities that should be placed on internet platforms. Several member states (including Denmark, Finland, the Netherlands, Czech Republic, Estonia, Luxembourg, Sweden, the UK) argued for a light-touch regulation towards internet platforms, while agreeing with a requirement for platforms to introduce a redress and complaint mechanism for users.

New financial obligations for internet platforms
The adoption process of the copyright directive also led to heated discussions on the introduction of new financial obligations for internet platforms, in particular with the introduction of a "neighbouring right" that would allow press publishers to receive financial compensation when their work is used by internet platforms such as Google News (Papadopoulou & Moustaka, 2020). A large number of member states were opposed to the introduction of a fully-fledged neighbouring right for press publishers. This group was led by Finland, the Netherlands and Poland, which perceived this change as detrimental to their digital sector, and in particular to news-related start-ups. Also, by regulating the practice of hyperlinks, these member states were concerned that it would restrict users' access to information and impact the core functioning of the internet. Other member states, led by France and Germany, urged in favour of the introduction of a neighbouring right for press publishers established in the EU. The high salience of this issue for France and Germany relates to the significant interests of their national press publishers and news agencies (including Bertelsmann media group in Germany) to be compensated for the use of their works 9 . Germany had already introduced in its national legislation a similar provision 10 . Other member states such as Portugal, Italy and Spain were also supporting this introduction, though investing less political capital, due to the more limited benefits this new right would bring to their publishers' industry.
These cases illustrate the overall distribution of preferences and issue salience in recent EU negotiations on data flows and the regulation of internet platforms. The next section will look at actual patterns of coordination between member states with similar policy preferences, in view to appreciate their ability to form and maintain coalitions around common interests.

Section V: Member states' coalition patterns on DSM files in the Council
As it is commonly understood, and as the previous section clearly illustrates, member states tend to be significantly divided regarding the regulation of data flows and internet platforms at the EU level. However, we also know that all member states are not equally equipped to advance their policy preferences at the EU level (Panke, 2012). All national negotiators do not benefit from the same level of 'network capital' with their counterparts (Naurin & Lindahl, 2010), neither do they possess the same administrative resources to formulate and defend their positions (Kassim et al., 2001). These asymmetries are partly determined by a set of domestic factors, such as political stability, financial means, administrative legacies, and bargaining style. In a forthcoming article, I argue that they have a direct, though not systematic, effect on the bargaining success of member states (Perarnaud, forthcoming). Qualified majority voting rules in the Council indeed incentivise member states to find coalition partners to either build political momentum around common policy preferences or constitute a blocking minority to challenge competing dynamics. But coalition dynamics require a particular set of resources and capabilities that not all member states can mobilise to the same extent. Coalitions can thus be considered as key to analyse previous and future EU policy-making processes, and this section will present coalition patterns observed as part of the three DSM legislative files under study.

The digital like-minded group
The digital like-minded group is both an intriguing and little-known coalition at Member states part of the digital like-minded group share a liberal approach to internal market and digital issues, and have in common to be digitally "ambitious", but not necessarily digitally advanced. For instance, Bulgaria belongs to the digital like-minded group, despite being one of the least digitalised member states in the EU (according to the Digital Economy and Society Index). This coalition can mobilise both during and prior to negotiations. For instance, before the adoption of the Commission proposal on the free flow of data, a number of joint letters from heads of states were forwarded by this group (Fioretti, 2016). These efforts to liaise with EU leaders are also exemplified by a number of more recent letters, for instance on the 'Digital Single Market mid-term review' in 2017 13 or in the preparation of EU leaders' meetings in 2019 14 . As part of legislative negotiations, the digital like-minded group can meet regularly to discuss text compromises and help align member states' influence efforts. This strategy proved successful in the context of the negotiations on the free flow of data, in which the high level of coordination between members of the 'digital like-minded group' allowed for the formulation of efficient strategies to contain the initial concerns forcefully expressed by France, and dismantle a blocking minority (Perarnaud, forthcoming). The format of the digital like-minded group is only used by member states when they share a rather homogeneous position on a particular subject matter. For instance, members of the digital like-minded group were significantly divided in the context of the copyright directive, and thus could not leverage this format.  (Alm et al., 2016). This development illustrates the indirect lobbying channels that corporate actors can use to secure influence in relation to the DSM, in addition to their more traditional repertoire of actions at the EU level (Laurer & Seidl, 2020;Christou & Rashid, 2021).
Since October 2017, "digitally advanced" member states meet at ministerial level as part of a larger group, known as the D9+, which also includes Czech Republic and Poland (The Digital Hub, 2018). The D9 initiative was indeed not perceived positively by all member states within and outside this group. Some negotiators argued that these nine member states did not hold sufficient voting power to influence legislative negotiations, whereas certain member states, such as Poland and Czech Republic, still wanted to cooperate with like-minded member states but were found excluded. These considerations were the main drivers behind the creation of the D9+ group. Yet, the overlap between these formats was identified as a challenge by two interviewees 16 .

The Franco-German alliance
As opposed to the "digital front-runners", France and Germany did not appear to coordinate their influence efforts as part of the three negotiations under study.
This can be partly explained by differences in their national preferences with respect to the regulation of internet platforms. Though they generally share common concerns, Germany and France had different, and sometimes, divergent interests to defend. Also, the repeated delays in the formulation process of the German posi- Länder (Sepos, 2005), especially as the German ministry of justice and the ministry of economy's positions can be difficult to reconcile on digital matters. The Franco-German alliance on DSM policies was thus only visible at high-level on a limited number of instances 17 , but rarely at the level of negotiators.
Though the research fieldwork did not allow for conducting interviews with all the negotiators involved in these processes, it should be noted that there was no evidence of other Council coalitions mobilised by member states. The literature on coalitions in the Council indicates that national representatives generally coordinate their position and strategy with their counterparts on an ad hoc basis depending on the issues at stake (Ruse, 2013), as illustrated here by the cases of Spain and Italy. Both could be considered pivotal in the shaping process of these decisions. Spain's progressive shift in the negotiations regarding the regulation on the free flow of data is for instance essential to understand the outcomes of these dynamics, and thus shows how member states not pertaining to coalition groups can still leverage influence in EU policy-making processes.

Section VI: Conclusion
These negotiation processes highlight the competing interests of EU member states and institutions with regard to key controversial aspects of the EU's DSM, and the coalitions they mobilise to amplify their message in Brussels.
These descriptive findings can be of great use for understanding the next political sequence initiated by the European Commission over the regulation of data flows and internet platforms. They provide new insights on the structure and salience of member states' preferences on these controversial issues, and the mechanisms at their disposal to gain influence.
While the observation of coalition patterns was specifically focused on the Council, it should be noted that member states can also approach other EU institutions to advance their preferences. Previous studies (Panke, 2012;Bressanelli & Chelotti, 2017) show that not all Brussels-based national negotiators regularly engage with MEPs in order to channel national interests and 'tame' the different political posi-17. See for instance the Franco-German joint statement released in the context of a high-level ministerial meeting on October 2015: https://www.economie.gouv.fr/files/files/PDF/Declaration_confer-ence_numerique_finale_FR.pdf tions voiced inside the EP. These variations partly relate to differences in the size of national delegations in the EP, but also to member states' negotiation style and administrative resources (Perarnaud, forthcoming). Interestingly, the member states with the more structured channels of communications with EU institutions are the ones that have expressed the highest salience on issues related to data flows and internet platforms regulation. As shown by previous studies (Panke, 2012), France, the Netherlands, Sweden, Finland and Czech Republic are indeed among the member states with the most powerful connections with the EP and the European Commission. Given the existing divides among member states in the Council, the European Parliament could have a greater say in the upcoming negotiations, thus possibly giving more leverage to the member states with structured mechanisms to liaise with MEPs.
The negotiations on the DSA, DMA and DGA legislative dossiers will generate significant debates and controversies at the EU level. Many detailed proposals and ideas are being voiced to shape the EU's approach on myriad of policy issues related to the regulation of data flows, competition and content moderation (Graef & van Berlo, 2020;Gillepsie et al., 2020). Policy proposals presented in the context of this new phase should take into account the informal power balance between member states in the Council, and existing asymmetries in their capabilities to defend national positions in Brussels. As it appears that strong coordination mechanisms between the most digitally advanced countries of the EU have granted them significant influence over large member states in recent years (Perarnaud, forthcoming), future research on these processes should carefully study coordination processes and coalition patterns, as well as their implications. The recent joint statement released by the D9+ group (D9+ group, 2019) ahead of the DSA negotiations is a good indicator of the relevance of studying preferences' allocation structure and coalitions as part of this new political sequence.