EU politicization and policy initiatives of the European Commission: the case of consumer policy

ABSTRACT European integration is increasingly contested in public. What are the policy consequences of this EU politicization? This article argues that politicization challenges the hitherto often technocratic mode of policy preparation in the European Commission. Increased public attention and contestation render the diffuse public a more relevant stakeholder for Europe’s central agenda-setter because future competence transfers to Brussels are more likely to be scrutinized in the public realm. This incentivizes Commission actors to generate widely dispersed regulatory benefits through its policy initiatives, particularly where an initiative covers publicly salient issues. Applying this expectation to 17 European consumer policy initiatives suggests that the Commission orients its policy proposals towards wide-spread consumer interest during periods of high EU politicization and issue salience. However, the mechanism is constrained by internal turf conflicts and anticipated Council preferences. These findings highlight that politicization entails both chances and risks for further, policy-driven integration in Europe.


Introduction
Analyses of European integration increasingly revolve around the idea that supranational 'decision making has shifted from an insulated elite to mass politics' (Hooghe and Marks 2009: 13). The surge of Eurosceptic parties in the latest European Parliament (EP) elections or the Brexit referendum provide particularly glaring examples for this claim. Contrasting the 'permissive consensus' (Lindberg and Scheingold 1970), we observe an 'increase in the polarization of opinions, interests or values and the extent to which they are publicly advanced towards the process of policy formulation within the European Union' (De Wilde 2011: 560).
There is, however, only scant empirical knowledge on the actual policy consequences of EU politicization. Recent research debates whether politicization constrained responses to the Euro and Schengen crises (Börzel and Risse 2017;Schimmelfennig 2014). Other recent evidence suggests that supranational elites communicate public interests more strongly in the face of politicization (De Bruycker 2017). But does politicization also affect the mid-range policy choices along the 'community method' that have shaped the course of European integration thus far?
The European Commission is particularly relevant for this question. It embodies the tensions of insulated, technocratic policy-making on the one hand, and publically accountable governance on the other (Christiansen 1997;Wille 2013). Its monopoly of initiative in the regulation of European markets grants significant agenda-setting power over the contents of EU lawa power the Commission has often used in an entrepreneurial manner, adapting treaty interpretations to changing context conditions (Cram 1997).
This article argues that EU politicization incentivizes Europe's central agenda-setter to be more responsive to public interests. A Commission interested in the retention of its competences should aim at widely dispersed regulatory benefits when the EU is heavily debated in public. But this strategy risks alienating traditional stakeholders, most notably member state governments and cross-nationally operating businesses. Hence, the Commission will tilt its proposals towards diffuse public interests mainly in cases where the public is likely to note the respective policy choices. This is true if general EU politicization combines with a high public salience of the specific regulatory issues at stake.
The article provides a plausibility probe of this argument in European consumer policy. Harmonizing European markets in this area, the Commission faces a choice between serving narrowly concentrated producer interests or widely dispersed consumer interests. A summary of 17 case studies on consumer policies between 1999 and 2008 is consistent with the expectation that the Commission's positioning on this continuum is sensitive to generalized EU politicization and specific issue salience. However, such enhanced responsiveness is constrained by the aggregated preferences of the EU member states and the heightened conflict potential within the Commission itself.
Supranational policy-making in the public spotlight

The politicization of European integration
Supranational integration has always involved politicalthat is, collectively bindingdecisions, but the politicization concept stresses the degree to which these decisions are also collectively debated. Zürn et al. (2012: 71) understand politicization as 'growing public awareness of international institutions and increased public mobilisation of competing political preferences regarding institutions' policies or procedures'. Politicization is thus more than just a synonym for declining support for supranational governance. It contains both the resistance against specific international institutions and the formulation of demands for more or other international policies. The concept furthermore reaches beyond institutional disagreement and captures in how far decisions are pulled into the public spotlight. In the context of European integration, Statham and Trenz (2012: 3) understand politicization as debates and controversies on supranational issues in the public sphere. Hutter et al. (2016) also emphasize that politicization finds it expression in decidedly public, mainly mass-mediated debates. In his encompassing conceptualization, De Wilde (2011) defines EU politicization as an increased public involvement of societal actors such as political parties, mass media, social movements in the process of European integration, and the degree to which this resonates among the wider European citizenry.
Despite disagreement on the root causes of this EU politicization, there is a clear convergence on its empirical components (De Wilde et al. 2016;Rauh 2016: Chapter 2). The first one is salience, meaning the degree to which European integration is visible and important to the broader citizenry. The second component is polarization, meaning the degree to which public opinions on European integration diverge. The third component is mobilization, meaning the extent to which public debates expand to include actors beyond supra-and international executives.
This threefold conceptualization, allows us to assess the long-term, aggregate politicization potential of European integration. To hold context conditions largely constant, while allowing a sufficient time frame and keeping research efforts feasible, I concentrate on the EU-6 countries here. Public salience of the EU is proxied by the monthly share of articles referring to the EU in the headline in one opinion-leading newspaper per country. To study the polarization of public opinion, the seminal Eurobarometer item on support for a country's EU membership taps into the cumulative assessments of European integration in the wider public (Lubbers and Scheepers 2005). I focus on the bi-annual EU-6 averages of the item's variance and kurtosis as indicators of opinion polarization (Down and Wilson 2008). Mobilization on European integration, finally, is captured by the monthly counts of publically staged protest events addressing the EU or its policies (Uba and Uggla 2011). 1 Standardizing and combining these indicators into an additive index provides reliable comparisons of EU politicization over time. As indicated by the OLS time trend in Figure 1, the politicization potential of European integration in the wider public has consistently increased since the early 1990s mainly driven by an almost continuously differentiating public opinion on EU membership. But EU politicization is also subject to significant shortterm swings driven by local peaks in media salience and protests around major integration events such as EP elections, enlargement rounds, and especially treaty revisions.
From the perspective of Brussels' hallways, these data suggest that public attention and contestation have become an increasingly important context condition of supranational policy-making while corresponding pressures also fluctuate in the short run. So how, if at all, can we expect the European Commission to respond to the thus shaped politicization of European integration?
Commission policy on and off the public's radar One of the most distinguished features of the European Commission is its monopoly in initiating supranational legislation, which provides it with significant agenda-setting powers (Tsebelis and Garrett 2000). Like broad strands of the literature, this article assumes that the Commission exploits these powers Figure 1. Aggregate index of EU politicization (salience, polarization, and mobilization) in the EU-6 countries.
Notes: The grey line presents monthly index values, the black line their six-month moving average. The straight line provides a monthly OLS time trend and its 95% confidence interval. For details, see Rauh (2016: Chapter 2). also to ensure the expansion of its own supranational competences (Franchino 2007). 2 This requires that the Commission can build on a sufficient stock of output legitimacy, that is, the added value its activities produce for major stakeholders (Majone 2002;Moravcsik 2002). Member state governments are most important in this regard. They have delegated competences to the Commission in order to overcome short-term political pressures that hamper cross-national cooperation (Moravcsik 1998). Substantially, this delegation has focussed on the stimulation of economic growth through the creation of a common market that removes, overrules, or harmonizes national policies hampering cross-border trade (Scharpf 1999). This has rendered cross-nationally operating producers a second important stakeholder group for the Commission (Coen 1998).
In the context of EU politicization, however, these traditional sources of output legitimacy are insufficient at best. First, politicization means that the supranational level becomes a much more direct addressee of public evaluation (Zürn 2006). The more alert the public becomes to supranational political authority, the more rational it is for a competence-seeking bureaucracy to care about the public acceptability of its policies. Otherwise, the Commission jeopardizes the further transfer of competences to the European level. This may work indirectly where increasing public contestation lets individual national governments adopt more critical positions toward European integration (Hooghe and Marks 2009), but it may also work through a much more direct routeas the 2005 referenda in France and the Netherlands, the surge of Eurosceptic parties in the 2014 EP elections, or the Brexit referendum in 2016 have forcefully demonstrated. In either way, the politicization of European integration increases the weight of the wider public in the utility function of a competence-seeking Commission.
Second, the mere pursuit of economic liberalization is hardly suitable to demonstrate an immediate added value to the public. While gains from open markets take time to materialize and often can only be demonstrated by counterfactual arguments, the immediate political costs of liberalization for vested societal interests are instantaneously visible. Moreover, while European elites tend to care mainly about competitiveness, the wider public often prefers market-flanking policies instead (Dehousse and Monceau 2009;Hooghe 2003). In result, a competence-seeking Commission has incentives to visibly serve such public preferences.
Enhanced responsiveness comes, however, at the cost of undermining the immunity to short-term political pressures and extant policy solutions that the Commission's traditional stakeholders value. For a competence-seeking bureaucracy trying to generate output legitimacy vis-à-vis all of its stakeholders, this creates trade-offs.
Policy responses to EU politicization should accordingly depend on the Commission's anticipation whether a particular initiative has a chance to actually influence the public's overall evaluation of the EU. The public, however, hardly follows each and every proceeding on the supranational agenda. Rather, public attention is selective, particularly on issues exceeding the domestic domain (Oppermann and Viehrig 2011). Some Commission initiatives fly safely below the public radar, while others may touch upon specific topics that the public currently cares about.
General EU politicization will thus especially matter if the Commission receives signals that the legislative initiative in question addresses specific issues that the diffuse public currently considers relevant. In this view, the public salience of the specific issues to be regulatedthat is, the degree to which these issues are easily understandable and visible among the broader public at the time of drafting (Epstein and Segal 2000) moderates the link between general EU politicization and the Commission's policy output. Figure 2 summarizes the model. Politicization renders the European public a more relevant stakeholder for further European integration, incentivizing the Commission to adapt its policy choices to widely shared public interests (a). On the level of individual policy initiatives, this link is moderated by the contemporaneous public attention to the specific issues the respective Commission proposal covers (b). 3 In publically salient domains, the Commission weighs the risk of alienating traditional stakeholders or contradicting extant policy against the risk that its policy position is immediately perceived and understood by the wider public (c). If the latter risk is sufficiently high, the public's evaluation of the particular initiative may feed back into the overall political evaluation of European integration and, by implication, the Commission's competences (d).
This argument contains endogeneity in two respects. First, general EU politicization can be understood as a cumulative response to the sum of individual supranational decisions (De Wilde 2011: 565-66;Weßels 2007). In fact, the proposed argument rests on this assumption. From the Brussels perspective, public attention to an issue area increases the likelihood that the Commission's final policy choice will actually be perceived by the wider public and taken into account in the general evaluation of the supranational polityno matter whether the public has perceived the issue as an originally European one in the first place. For an individual legislative initiative, however, the level of the general politicization of European integration is a given context condition that is analytically distinct from the specific contemporaneous salience of the issues the initiative is going to cover.
Secondly, endogeneity may occur on the level of individual initiatives because the Commission can manipulate public salience during policy formulation. Individual Directorates-General (DGs) may try to push an issue up on the public agenda by publishing studies, by circulating consultation documents, or by approaching public media directly. This creates a short circuit in the model, but it does not contradict the expectation that issue-specific public salience is a decisive variable moderating the effects of general EU politicization if the Commission adapts its position after the salience of the initiative in question has been successfully increased. The possibility of such short circuits highlights that an evaluation of the model has to take policy formulation process into account.

Consumer policy
To evaluate the model, we also have to specify which policy choices actually serve widely shared public interests. To render this possible, this article focusses on European consumer policy, defined as regulatory measures that aim at protecting the end user of products or services against risks and disadvantages in economic life (Weatherill 2005). The diffuse nature of consumer interests lets us initially expect little political clout vis-à-vis producer interests. Consumer policy often serves as a prime example for collective action problems, which can only be overcome by entrepreneurial regulators (Pollack 1997).
Two ideal types of consumer policy are conceivable (Cseres 2005: 320). From a purely economic perspective, consumer policies remedy market failures that preclude rational consumers from reaping the full choice offered in a perfect market. Information-seeking and risk bearing rest primarily with the consumers (caveat emptor principle). Regulatory interventions under such a laissez-faire model would solely focus on allocative efficiency and boil down to competition policy or information rules. In contrast, an interventionist model of consumer policy emphasizes the re-allocation of consumer rights (Janning 2004;Micklitz and Weatherill 1993). This model rests on the assumption that the consumer is structurally disadvantaged in markets obscured by product differentiation and the multiplicity of packaging, advertising, and distribution. Respective regulatory interventions establish producer responsibility for economic and physical consumer risks (caveat venditor).
National consumer policy regimes vary in between these ideal types, which creates trade barriers justifying supranational action in this area. However, harmonizing specific consumer policy issues confronts the Commission with a choice on the level of protection in the proposed European law. The political spoils the Commission can distribute with this choice vary strongly over the ideal-type consumer policy models.
Supranational policies tending towards the laissez-faire model increase market access, pleasing especially producers interested in cross-border trade. This is fully consistent with the traditional market-making mandate of the Commission (Micklitz et al. 2004). Respective policies alienate the traditional stakeholders the least. They do not necessarily hurt consumers, but they also do not allow the communication of immediately visible benefits to the diffuse public. Yet, the Commission also has some room to tilt its proposals towards the interventionist model. Since Amsterdam, the treaties entail a clear, albeit legally weak basis for a 'high level of consumer protection' (Micklitz et al. 2004). In contrast to focussing on market access only when harmonizing European consumer law, proposing more interventionist policies concentrates costs of market risks on a narrow set of producers while the Commission can claim to have protected each and every European consumer.
The more the diffuse public is considered as an important stakeholder, the more attractive is this distribution of narrowly concentrated costs and widely spread political benefits for a competence-seeking Commission. The key choice variable in consumer policy, thus, is the distribution of rights and obligations among producers and consumers. The theoretical model above thus leads to the following hypothesis: The higher the levels of general EU politicisation and public salience of the regulated issues are, the more Commission initiatives will tend towards an interventionist model of consumer policy.
For three reasons, this has to be understood as a probabilistic prediction. First, politicization and issue salience can only be meaningfully understood as continuous variables (Hoeglinger 2016). Second, the dependent variable as well must be perceived as a continuum. There is no politically unbiased absolute measure of consumer interventionism (Janning 2004) and the Commission does not operate in a regulatory vacuum. It rather faces a status-quo that is either given by an existing supranational law or by national consumer policies. Third, there is a myriad of additional factors impinging on the Commission's policy choices. Most importantly, it is well established that the Commission strives to propose policies that are politically feasible in the EU's interinstitutional decision-making process. Thus, especially the preferences of (qualified) Council majorities will effectively limit the range of available options without, however, determining the Commission proposal in full (Tsebelis and Garrett 2000). On specific consumer policy issues producers themselves may lobby for consumer policy interventionism in order to get protection against extra-European competition (Trumbull 2006). And finally, any policy choice has to be coordinated among the Commission's different DGs. Each of these might have their own competence-seeking motives, varying stakeholder networks, and Commissioners with different national and partisan backgrounds (Hartlapp et al. 2014).

Research design
Against these challenges, the empirical analysis aims at a 'plausibility probe' (Eckstein 1975). It intends to show that proposed argument warrants further attention by providing a guided comparison of a medium-N set of detailed qualitative analyses that allow for alternative explanations, within-case variation and explorative insights. Empirically, I draw from a project on position formation inside the European Commission (Hartlapp et al. 2014).
The unit of analysis is an individual drafting process that leads to a formal Commission proposal in consumer policy. The policy area is a circumscribable EU competence since Maastricht (Weatherill 2005), but only the Prodi Commission established a fully fledged DG with a respective legislative mandate (DG SANCO, Guigner 2004). Accordingly, the investigation period starts in 1999 and ends with the onset of the broader project in 2008, covering two Commission terms in full.
I first identified all 247 proposals for binding secondary EU law that the Commission itself flagged as consumer policy. 4 Given the ambition to trace individual drafting processes in detail, this had to be broken down further. Constructing a perfectly guided sample, however, was not achievable since key variables of the modelmost notably consumer interventionism and issue saliencewere not readily available ex ante. Thus I rather aimed at capturing relevant variation in the sample by three steps. First, I scattered the selected initiatives across the investigation period to maximize the chance to find variation in EU politicization and issue salience. Second, I tried to avoid biases towards or against consumer interventionism by including cases drafted in different Commission DGs and subject to varying internal coordination logics (oral and written procedures). Finally, to capture the Commission's most relevant consumer policy choices, I selected the cases with the broadest legal scope from the resulting strata, that is, preferring general purpose instruments over product-or service-specific regulations. This results in the 17 policy initiatives summarized in Table 1. While this is a purposive sample aimed at learning rather than generalization, note that it contains all essential legislative initiatives on contractual consumer rights, product safety, and food safety submitted by the Prodi and Barroso I Commissions.
To capture the dependent variableconsumer interventionism proposed by the Commissioneach legislative proposal was divided into a set of key provisions, meaning an array of articles that distribute logically linked rights among consumers and producers. 5 For each provision, I then compare extant policy options, stakeholder demands, and the choices in the final legislative text to the respective regulatory status-quo. The dependent variable captures whether the Commission proposal undermines, exceeds, or simply reinforces the regulatory distribution of rights and risks among producers and consumers. This requires a detailed and rather technical perspective at times but provides replicable qualitative judgements (Rauh 2016: Appendix E).
Regarding the major independent variable, I resort to the EU politicization index developed above. Cross-case comparisons focus on the index' mean levels while within-case comparisons take its fluctuation during the drafting period into account. The second independent variable, public salience of specific issues, is proxied by media prominence during the drafting periods (Epstein and Segal 2000). An automated content analysis retrieved the terms most frequently used by and across the different actors that issued position papers on the Commission proposals in food safety, product safety, and contractual consumer rights, respectively. These term lists were used as inputs for a LexisNexis newspaper search, resulting in a monthly indicator for their media prominence. These three broad salience indicators were complemented by a case specific, qualitative analysis of newspaper articles retrieved by searches along specific key words used in the recitals of the respective Commission proposal (Rauh 2016: Appendix C3).
Beyond comparisons of the model's input and output variables, the analysis asses validity of the proposed argument also by systematically analysing within-case variation while taking explorative insights and major alternative explanations into account. Since access to internal Commission data is restricted, the corresponding case histories draw strongly on interviews with involved Commission officials (Rauh 2016: 57). Interviewees were identified along the officials' formal positions during the respective drafting processes and reputational information gained during initial research. Information was cross-validated by trying to talk to all involved Commission DGs and hierarchy levels. A total of 41 officials could be interviewed for the cases at hand. The greatest cluster (20 officials) lies in DG SANCO, which signed responsible for the majority of proposals and was associated to all others. Lower Commission echelons could be more easily accessed (24 desk officers), but the underrepresentation of higher hierarchies is also due to the fact that they tend to sign responsible for several cases in the sample (10 heads of unit, 3 Directors, and 4 Cabinet members).
Finally, this was complemented with a broad range of secondary information from internal process documentation, discussion papers, consultation documents, press releases, and impact assessments, as well as interest group's position papers and coverage by Agence Europe and major European newspapers. Pulling these sources together, systematically structured case histories were drawn up for each of the 17 drafting processes (Rauh 2016: Chapters 4-6), focussing on the temporal dynamics in the model's main variables as well as the role of internal conflicts and anticipation of member state preferences. The remainder of the article summarizes the key insights of these analyses.

Findings: serving widely shared interests under constraints
Comparing policy initiatives Table 1 condenses the case studies. Seven initiatives favoured more consumer rights as compared to the regulatory status-quo. In five cases, the respective initiative rather emphasized producer freedom while four proposals merely reinstated the status-quo. Departmental divisions in the Commission do not explain this pattern. For example, the proposed consumer interventionism varies heavily across the proposals drafted by the then DG for Health and Consumer Protection (SANCO).
How plausible is it, however, that the combination of EU politicization and issue salience pushes the Commission towards consumer-friendly regulation? Clearly, a purposively constructed medium-N sample hardly allows immediate generalization. But already the aggregate view on the 17 major policy initiatives highlight that the argument cannot be readily refuted. Focussing on the mean levels of politicization and salience during the respective drafting period unveils seven cases that appear immediately consistent with the theoretical model. Comparatively high values on both independent variables are associated with a re-distribution of rights favourable to consumers in the initiatives on universal service, air passenger rights, consumer credit, and food information. Where high politicization combined with only low public salience of the specific issues, the Commission initiatives on sales promotions, pyrotechnic articles, and food supplements exhibited further liberalization of the affected laws instead.
Other initiatives defy clear-cut judgments, especially where we observe only intermediate levels on the dependent variable. In the consumer rights case, for example, high levels of EU politicization and issue salience occurred during drafting, but the final proposal took an interventionist position on only some, particularly salient key provisions covering online trade. Other provisions reinforced or undermined the status-quo, for example, with regard to guarantee periods. Similarly mixed interventionism abounds for the toy safety, the 'new approach', and the food additives proposals.
More importantly, there are four drafting processes for which we only have intermediate politicization values. The table's classification assesses whether the mean of the politicization index in the drafting period differs significantly from the average politicization levels observed before. This sets the bar very high and points to a conceptual threshold problem. EU politicization has increased, but we do not know which level is sufficient to create the theorized Commission incentives and whether this level has been reached before the investigation period already. Future research could tackle this by testing the argument in earlier periods of integration, by normalizing the politicization of European integration against other political issues (Hoeglinger 2016), or by comparing policy responses to politicization across different international organizations (Rauh and Bödeker 2016). For the present project, we have to focus on the within-case variation discussed below.
The comparative perspective yields more consistency on issue salience. Of the eight cases that indicate interventionist deviations from the status-quo, six were drafted under comparatively high media attention to the issues covered by the respective proposal. The four cases with intermediate interventionism are also roughly in line. Two of thesethe 'new approach' revision and the food additives proposalmerely re-codified the regulatory status-quo. The two othersconsumer rights and toy safetycontained some liberalizing but also some rather interventionist provisions on specific issues that were on the public agenda at the time of drafting. Regarding online terms and conditions as well as chemicals in toys, the Commission was responsive, although these specific issues did not tip the aggregate judgement on these proposals.
In sum, the comparative summary yields seven cases fully consistent with the theoretical model, six cases where it can neither be confirmed nor falsified, and two cases in which the observed consumer interventionism was not driven by EU politicization and issue salience. While these findings cannot be readily generalized, the patterns encourage further analysis of the argument.

Adapting positions during drafting
The Commission's choices on consumer interventionism during the drafting process provide additional information to evaluate the theoretical argument's plausibility. Panel three in Table 1 provides a rough indication during which stage politicization and issue salience reached their mode, and notes whether this coincided with swings towards more interventionist or liberalizing positions.
This within-case perspective initially reinforces the role of issue salience. For example, in toy safety, DG ENTR foresaw rather liberal rules during more than five years of drafting but suddenly adopted more interventionist provisions on safety requirements, on chemical regulation, and on labelling once huge recalls of Chinese toys hit the European market in 2007. The soaring public salience of the issuenot the least driven by poisoned Barbie dolls in the European marketput DG ENTR under pressure in the final stages of drafting because 'people wanted to have immediate results' (COM25:234). The late inclusion of a social review clause into the universal service proposal, when media attention to widespread internet access rose in late 1999, serves as a similar example.
More importantly, the reconstruction of the drafting processes reveals cases during which issue-specific media prominence remained stable and the Commission adapted its position along swings in the EU politicization index. The 2001 air passenger rights proposal provides an example. Commission officials were aware that bad customer treatment by airline companies 'was all the time in the media' (COM33:136), but DG TREN only committed to a formal initiative when the general EU politicization started to rise again during the preparation of the Nice negotiations. In this context, the drafting officials argued that the EU-induced air transport liberalization had been about 'advantages for the industry' while it was now time to 'refocus transport policy on the demands and needs of its citizens' (COM33:140). Against outright industry opposition, the officials publically proposed to quadruple consumer compensations for delayed and cancelled flights only when EU politicization had almost reached its local 2001 peak.
The same peak was exploited by DG SANCO in drafting the proposal on consumer credit. In contrast to what was internally agreed with DG MARKT before, SANCO only then publically committed to very interventionist positions in this proposal. This concerned the inclusion of mortgages and especially a principle that made creditors responsible for correctly assessing the financial situation of debtors. Knowing that this was 'a very political file […] with a big public appeal' (COM119:171), the officials tried 'to create a very comprehensive, very exhaustive consumer credit regulation which would be burdensome for industry ' (COM89:46).
Also during that summer, SANCO's initial preparations for the unfair commercial practices proposal entailed a generalized producer duty for fair trading. However, this interventionist position was scaled down to a neatly circumscribed set of prohibited practices during the sub-standard levels of politicization in the years 2002 and 2003. This change of the Commission's policy position occurred despite constant media prominence of unfair commercial practices in cross-border online trade.
Likewise, the high politicization phases in 2004 and 2005, and the later decline were mirrored in Commission positions. The proposed 'new approach' overhaul provides evidence. Major enhancements of market surveillance such as the centralization of the notification system were contemplated by DG ENTR during 2004. This was later cut back to a mere re-codification of the existing rules during the final stages in 2006 and 2007. The other 'neutral' case in the sample showed a parallel decline in Commission interventionism. When formulating the policy on food additives, DG SANCO considered the inclusion of processing aids and the time-limited authorization of additives during the high politicization phase of 2005, but dropped these ideas against producer opposition when EU politicization reverted during the final stages of drafting. Lastly, even in the finally rather interventionist 2008 Commission proposal on food information we see concessions to producers on alcoholic beverages and origin labelling in parallel to declining EU politicization in the last half-year of position formation. Again, this happened despite rising public salience as captured in frequent media reporting on food regulation.
All these examples concern only parts of the policy proposals, and they switched the judgement on the final outcome only for some. But these within-case adaptations of policy positions provide tangible evidence consistent with the theoretical claims. Not only was the Commission sensitive to varying salience of the specific issues in question, but where salience remained stable, its policy positions also mirrored major upsurges and declines in general EU politicization.

Procedural observations and alternative explanations
The case histories reveal additional detail on how EU politicization and issue salience affect the intricacies of policy drafting inside the Commission, but also indicate limits of the theorized effect.
More interventionist choices by the Commission usually met very explicit opposition of affected industry groups. Only in three product safety cases, certain industries supported a more interventionist stance as a means to protect them against external competition, most notably from China. Lobbying by consumer groupsespecially through their European umbrella organization, BEUCcould be observed across all but one proposal in the sample. They consistently pushed for more interventionist rules. But only in cases subject to high EU politicization and issue salience, the Commission's policy choices indicate neat congruence to the demands that consumer groups voiced in their position papers. This also holds within cases: shifts towards more interventionist policy positions occurred most often by explicitly taking those consumer group demands on board that had been declined earlier. Public interest groups provide the Commission with readily available templates for which public attractiveness can be assumed. This finding bodes well with research on public salience as a relevant lobbying resource (De Bruycker 2016;Dür and Mateo 2014;Klüver 2011).
The case studies furthermore suggest that salient initiatives in a politicized context involve the Commission's political hierarchies to a greater extent. For example, either rising politicization levels as in the consumer credit case or increasing issue salience as in the toy safety case made desk officials explicitly seek the political backing of their cabinets and Commissioners. The administrative level, however, has significant agenda-setting powers in this regard. As one official put it: 'the services, we have a larger view, which allows us to choose at every moment, depending on the political momentum, the kind of proposal we would like to boost at that time' (COM33:164). In the bus and maritime passenger cases, an outgoing Commissioner opposed more interventionism, but the service level merely paused until the political leadership changed.
Commissioner nationality or partisanship hardly explain consumer policy choicesin the present sample interventionism varies more within than across Commissioners' office terms. The comparatively liberal proposals on unfair commercial practices or the food fortification initiative but also the highly interventionist proposals on consumer credit or health claims, for example, were adopted under Commissioner Byrne. In 10 cases, the political leadership of the lead DG changed during drafting, but no changes in consumer interventionism could be detected in response. Future research studying policy effects of politicization should then not only focus on the Commission's political leaders but also on its administrative echelons.
The findings indeed suggest that public attention creates opportunities for competence-seeking bureaucrats. In a number of instances, Commission DGs actively tried to increase public salience in order to garner support for their regulatory plans. Glaring evidence abounded in the air passenger rights case. In a context of repeated media reporting on overbooking practices of airline companies, DG TREN publically pushed airlines into a voluntary agreement only to override it with an even stronger regulation later on. Similarly, DG SANCO used widely circulated consultations, the publication of studies, and a range of soft-law measures on obesity in Europe to prepare the ground for its encompassing and comparatively interventionist food information proposal.
In this vein, politicization and issue salience can be both a driver and an asset for turf conflicts inside the Commission. The case studies highlight that more interventionist policy plans are internally often opposed by DGs that defend their traditionally more liberal approach. For example, DG SANCO's interventionist positions in the consumer credit case met the fierce opposition of DG MARKT. But once politicization peaked in 2001 against a constantly high salience of consumer over-indebtedness in the press, DG SANCO waged to publically commit to internally highly contested provisions, for example, on the responsible lending principle, and managed to assert most of them until the proposal was formally adopted. DG SANCO had also been unsuccessfully pressing DG ENTR on a clear reference to its own, more interventionist general product safety directive in 'new approach' instruments. Only when a range of toy recalls increased public salience during late 2007 when ENTR drafted a regulation on toy safety along this model, SANCO's long-standing demands suddenly became assertive. Here, DG SANCO strategically used the publication of its own data on toy recalls as well as several public speeches of its own Commissioner to put political pressure on DG ENTR. Likewise, frequent newspaper appearances and speeches of Commissioner Kuneva promoting consumer protection by the Commission shielded the 2008 consumer rights proposal against internal concessions during the final internal negotiations with DG MARKT.
For future research on Commission responsiveness, these examples highlight two things. First, extending the competence-seeking assumption to individual Commission DGs helps to explain policy choices of the Commission as a whole. Second, EU politicization and issue salience can present an opportunity structure for some of these internal actorssupporting especially policy mandates that address diffuse public concerns more directly.
While these findings make it plausible that public politicization and issue salience affect the Commission's policy choices, the case studies also underscore that the member state governments remain the Commission's most important stakeholders. In 13 cases, drafting officials engaged strongly in uncovering the political feasibility of policy choices in the Council. Strategies included not only formal communications to the Council and consultation of individual governments, but also the involvement of national working groups in which consecutive draft proposals were negotiated. Comparing the final legal texts to national concerns voiced during drafting indicates that the latter sometimes provide upper bounds for consumer interventionism. For example, the Commission excluded products used in services from the scope of the general product safety directive or removed obligatory colourcoded nutrient information from the food information proposal in response to anticipated Council opposition. But sometimes national preferences also provided lower bounds and limited liberalization plans envisaged by the responsible DG. For instance, most of the remaining trade restrictions in the pyrotech proposal can be traced to national demands. National preferences thus trimmed the lead DG's ideal positions without determining them fully. This suggests that the political feasibility in the Council defines the leeway in which the Commission's policy reactions to the politicization of European integration can unfold.
A Commission official summarizes this succinctly: We have looked at an issue and we are trying to move forward in line with the common European interest. But then there are two considerations. These findings suggest that EU politicization can indeed increase supranational responsiveness. It creates an incentive structure in which non-governmental organizations representing diffuse societal interests can strengthen their influence on supranational policy by raising the public salience of their requests (cf. De Bruycker 2016; Dür and Mateo 2014;Klüver 2011). EU politicization is thus not only constrainingin policy fields that allow the immediate presentation of wide-spread public benefits it may also lead to more integration instead.
Enhanced output responsiveness may help in convincing European citizens that EU policies are not fundamentally biased against public preferences. This will not fully gratify the proponents of a politicized integration process whose benchmark is popular democracy (Hix 2006). Nevertheless, the Commission's engagement in demonstrating immediate public benefits creates a valuable precondition for enhancing input legitimacy in the future. The more citizens become aware that Europe can do something for them in the short term as well, the more EP elections would be about the European policies that citizens actually demandcarrying them beyond being mere contests on whether voters support the EU in principle.
Yet, as feared by critics of more politicized decision-making (Majone 2002) the price for such responsiveness is less efficiency. Public salience can lead the Commission to challenge deeply entrenched regulatory solutions, resulting in partly contradictory law and pronounced internal turf conflicts. The case studies furthermore underline that the political feasibility of Commission initiatives in the Council remains a decisive constraint. An overall assessment of the policy consequences of politicization has to take its effects on Council decisions into account. A reduced bargaining space or biases towards member states with stronger domestic dissensus are plausible expectations in this regard (Hooghe and Marks 2009;Wratil 2018).
Caution is also warranted when generalizing these insights to other policy domains. On the one hand, the Commission's activity in social policy is likely to remain regulatory in nature (Cram 1997). On the other hand, an initiative's salience mediates only whether the Commission considers the public's reception in a politicized climate, while the question of how it responds is subject to the distribution of societal interests on the issue in question. For the latter, consumer policy is an easy case. Narrowly concentrated and widely dispersed interests can be identified and cut across European societies. In other recently salient issue areas such as social spending, migration, or monetary policy, the public is split along class lines, along identity considerations, and even along national borders (Hutter et al. 2016;Scharpf 2010). In such contexts, the logics presented here may drive a Commission that is sensitive to EU politicization to blame avoidance strategies, letting it shy away from decisions on salient but much more controversial issues in which no unequivocal societal winners can be identified.
In this view, EU politicization is both a chance and a risk for further European integration. To disentangle positive and negative effects, future large-N research could extend the initial empirical insights to longer time periods, different policy areas, and other institutions in the supranational decisionmaking process. Capturing interventionism by automated text scaling of policy choices against interest groups or public demands provides a promising tool in this regard (e.g., Klüver 2011). The plausibility probe summarized here highlights that we need more research on how European institutions can channel widespread public contestation into acceptable policy choices. Notes 1. A more detailed discussion of these indicators is provided in Rauh (2016: Chapter 2). Data are available at www.christian-rauh.eu/data-and-resources (accessed 5 March 2018). 2. Given varying attitudes in the Commission (Kassim et al. 2013), this is a simplifying assumption. However, Commission officials tend to assess their organizational environment along rational calculations (Bauer 2012) and realize that the politicization of European integration challenges supranational competences (Bes 2017). The model developed here also works with the more constrained assumption that Commission officials on average hold a preference to at least retain the regulatory powers of the Commission. 3. This implies an interaction effect of general EU politicization and the salience of specific issues. Both together account for policy choices geared towards serving wide-spread interests. In a hypothetical scenario in which supranational authority is not publically contested at all, salience of specific issues should not have a direct effect on a Commission without direct electoral accountability. Conversely, where the contemporaneous public would not care about a specific issue at all, the Commission should not deviate from the interests of its primary stakeholders. 4. Identification resorted to the EUR-Lex directory codes 'general consumer policies' (15.20.10), 'consumer information, education and representation ' (15.20.20), 'protection of consumer health and safety ' (15.20.30), and 'protection of economic interests' (15.20.40). The full universe of cases is part of the replication package.
5. For example, the 2008 proposal on consumer rights comprises six key provisions on scope and harmonization approach, pre-contractual information obligations for traders, trader obligations for off-premise and distance contracts, trader obligations after the sales contract, commercial guarantees and general rules on contract terms.

Disclosure statement
No potential conflict of interest was reported by the authors.

Notes on contributor
Dr Christian Rauh is a senior researcher in the Global Governance department of the WZB Berlin Social Science Center. More information at www.christian-rauh.eu