Oppositional voting in the Council of the EU between 2010 and 2019: Evidence for differentiated politicisation

This article presents a new and previously unchartered dataset on roll call votes for all 28 member states in the Council of the EU between 2010 and 2019 and studies the effects of politicisation on governments' oppositional voting in the different policy areas. We contribute to the literature with two main findings. First, our study provides strong evidence for bottom-up politicisation, where Euroscepticism and the left-right positions of national political parties strongly affect governments' voting in the Council. Second, we provide new evidence for a form of differentiated politicisation where ideological standpoints of political parties in government and opposition have different effects on oppositional voting in the various policy areas.


Introduction
The European Union (EU) has been challenged and transformed by multiple crises (a financial, economic, Eurozone, and asylum crisis) (Bickerton et al., 2015;Lefkofridi and Schmitter, 2015;Schmitter, 2012;Vollaard, 2018). However, no study to date has systematically explored the extent to which these crises affected member states' voting in the Council of the EU. Nonetheless, we can expect significant changes. First, member states have been asymmetrically exposed to economic stress, which has increased economic tensions and disparities (Genovese and Schneider, 2020;Scharpf, 2015;Wasserfallen et al., 2019). Second, the crises have enhanced two contrasting political developments that manifest themselves in the Council. On the one hand, decision-making in the Council has long been characterised by a depoliticised consensus culture (Hayes-Renshaw and Wallace, 2006: 24, 56;Heisenberg, 2005: 67-79;Thomson, 2009: 760). On the other hand, the crises have enhanced politicisation in the member states (de Wilde, 2011;de Wilde and Lord, 2016;Zu¨rn, 2016), which places pressure on governments (Bressanelli et al., 2020) to adjust their policy position (Hagemann et al., 2017(Hagemann et al., , 2019Hobolt and Wratil, 2020;Wratil, 2018;Wratil and Hobolt, 2019).
However, existent research provides no guidance on whether governments' voting behaviour has changed and how potential changes in the post-Lisbon era can be explained. This article fills this gap in the research by simultaneously estimating the impact of economic-structural factors and facets of domestic politicisation on governments' voting. We present a new and previously unchartered dataset on roll call votes for all 28 member states in the Council between 2010 and 2019 to model governments' oppositional voting (defined as votes against and abstentions) in different policy areas. While a few studies explicitly examine how governments' voting differs depending on the policy (Bailer et al., 2015), variations across a wide range of policy areas, especially in the post-Lisbon area, are understudied (Hosli et al., 2011(Hosli et al., : 1267. We make two key contributions to the literature on politicisation and EU policy-making. First, when controlling for structural economic factors relevant to each policy area in the Council, we find strong evidence of a 'bottom-up politicisation' (Bressanelli et al., 2020). Independent of variations in key economic factors, our analysis reveals that the Eurosceptic and left-right positions of national political parties substantially affect governments' voting behaviour in the Council. Second, we provide new evidence for differentiated politicisation. We find that political parties' ideological standpoints in government and opposition have different effects on oppositional voting in the various policy areas. While governments' Eurosceptic and left-right position primarily seem to increase oppositional voting in policy areas with wider financial implications (agriculture and fisheries, economic and financial affairs, and the internal market), a more Eurosceptic parliamentary opposition increases the probability that governments will take a stance against Brussels only in the area of justice and home affairs.
We suggest that these trends of differentiated politicisation may result in increased differentiated integration (Schimmelfennig et al., 2015). In sum, our findings have important implications for the literature on decision-making in the Council. We reveal previously neglected facets and implications of a bottom-up politicisation driven by governments or the parliamentary opposition in the EU member states.

The Council of the EU in times of crises and politicisation
Numerous studies have focused on the pre-Lisbon period and, despite the requirement of qualified majority voting, ascertained a prevailing consensus culture in the Council (H€ age, 2011;Hayes-Renshaw and Wallace, 2006: 24, 56;Heisenberg, 2005: 67-79;Naurin and Wallace, 2010;Thomson, 2009: 760). As pivotal drivers for governments' position-taking and voting, scholars identified geographical (Hosli et al., 2011;Mattila, 2004;Mattila and Lane, 2001;Thomson et al., 2012), ideological (Hagemann, 2007;Hagemann and Hoyland, 2008;K€ onig and Luig, 2012;Mattila, 2009), and economic factors (Bailer et al., 2015;Schneider et al., 2007;Zimmer et al., 2005). These studies' general assumption is that governments act largely insulated from domestic party politics or electoral pressure. However, as these studies focus on the pre-Lisbon period, they do not account for the impact of the multiple crises (financial, economic, Eurozone, or asylum crisis) on EU policy-making (Bickerton et al., 2015;Falkner, 2017;Wasserfallen et al., 2019) or the increased tendencies of opposition towards the EU in the member states (Hobolt and de Vries, 2016).
In contrast, our study is situated in the post-Lisbon era, in which European integration has become increasingly politicised (Michailidou et al., 2014) and decisive for policy outcomes (Hobolt and de Vries, 2016;Hooghe and Marks, 2009;Hurrelmann et al., 2013;Hutter and Grande, 2014;Kriesi et al., 2008). Politicisation, in this sense and the context of this study, is defined as 'the demand for or the act of transporting an issue into the field of politics' (de Wilde and Zu¨rn, 2012: 139) 'making previously unpolitical matters political' (Zu¨rn, 2016: 167). Based on previous literature (de Wilde, 2011;Hutter and Grande, 2014), we conceptualise politicisation as a multi-faceted process where an issue (European integration) grows in saliency while actors' positions on this issue become increasingly polarised. In this article, we suggest that these facets of politicisation have become more relevant in explaining governments' voting behaviour in the Council in the post-Lisbon era.

Oppositional voting as a response to Euroscepticism
We suggest that one crucial facet of politicisation is the degree of Euroscepticism among parties and the public in a given country. As voting in the Council reflects governments' responsiveness to their domestic constituencies (Hagemann et al., 2017(Hagemann et al., , 2019Hobolt and Wratil, 2020;Roos, 2019;Wratil, 2018;Wratil and Hobolt, 2019), we argue that Euroscepticism, either directly expressed by public opinion or channelled through political parties, is also key in understanding governments' voting in the Council. Given that the mass-elite linkage on European integration is not always clear -either party positions form public opinion or parties change their positions to satisfy public opinion (Steenbergen et al., 2007) -we propose three hypotheses on how Eurosceptic party positions and opinions affect governments' voting in the Council. First, if government parties are becoming increasingly Eurosceptic -and studies indicate that they are (Winzen, 2020)we argue that this increases the likelihood of oppositional voting.
H1a: More Eurosceptic positions among government parties are associated with more oppositional voting in the Council.
Second, governments' voting behaviour in the Council is also affected by the degree to which the parliamentary opposition is Eurosceptic. Moreover, it is reasonable to expect a greater impact in this direction in the wake of the crises. Eurosceptic parties have not only made significant electoral progress in most member states during the last decade (de Vries and Hobolt, 2020) but many national parliaments have also increased their efforts to oversee governments at the EU level (Genovese and Schneider, 2020;Rauh and de Wilde, 2018;Winzen, 2012). If Eurosceptic opposition parties increasingly oversee the affairs of national governments in Brussels, then it follows that governments face greater incentives of strategically responding to them and even accommodating their position (Hagemann et al., 2019).
H1b: More Eurosceptic positions among national political parties in opposition put pressure on governments and make them more likely to oppose in the Council.
Third, voter perceptions of the EU are reflected in the popular support for European integration, which often refers to the general approval or disapproval of EU membership and unification (Hooghe and Marks, 2005). For example, Hagemann et al. (2017) demonstrate that national governments are more likely to oppose legislative proposals in the Council that extend the EU's authority level and scope when their domestic electorate is sceptical towards the EU. As it is more difficult to shift policies to domestic preferences at the EU level, governments use oppositional voting as a form of 'anticipatory representation' to communicate or send a signal to their electorate (Hagemann et al., 2017: 853-854).
H1c: Governments react to a more Eurosceptic public opinion by more frequently opposing in the Council.

Enhanced left-right party competition
Apart from an increase in saliency, politicisation further involves the increased polarisation of actors on European integration. Therefore, we suggest that this also enhances the importance of left-right positions when explaining voting in the Council (de Vries, 2010;Hooghe and Marks, 2009). The increase in polarisation, subsequently, leads to greater competition between centre-left parties that assume traditional left positions (pro-welfare, negative to economic liberalisation) and centre-right parties that take traditional right-wing positions (welfare-sceptical, positive to market-liberalism) (Abou-Chadi and Wagner, 2020; Hobolt and Rodon, 2020).
While Mattila (2004) found that governments' composition had limited explanatory power for voting in the Council and Hosli et al. (2011) revealed mixed results between old and new member states, Hagemann (2007) provided strong evidence for a generally positive relationship between governments' political orientation and voting. Refining this argument, Hosli (2007) concluded that the governments' distance from the ideological centre informs their propensity to cast oppositional votes. Building on these arguments, we maintain that in light of increased politicisation, the more extreme positions parties in government adopt on the left-right scale, the higher the probability that they will oppose decisions in the Council. While previous studies have focused exclusively on the left-right positions of governments, we follow the logic of H1b and hypothesise that more extreme positions in the parliamentary opposition also have the potential to pressure governments to adjust their position.
H2a: The more extreme the government parties' positions on the left-right dimension, the higher the probability of oppositional voting in the Council.
H2b: The more extreme the positions of opposition parties on the left-right dimension, the higher the pressure on governments to oppose in the Council.

Party system fragmentation
We identify the fragmentation of the national party system that challenges consensus at the EU level as another indicator for increased polarisation and, thus, a facet of politicisation. Mainstream parties that traditionally favour the EU political system, which once upheld the permissive consensus (Hooghe and Marks, 2009), were increasingly contested by 'challenger parties' who entered party systems throughout Europe (de Vries and Hobolt, 2020: 15-60). While challenger parties -often populist right-wing or left-populist parties (Usherwood and Startin, 2013) -mobilised new issues that contributed to altering the character of debates on the EU, the overall process of party system fragmentation weakens mainstream parties while the growing number of new parties creates new veto points. As a result, party system fragmentation should weaken governments and undermine the elite consensus on the EU that was previously needed for the smooth functioning of decision-making in the EU (Bartolini, 2005;Hooghe and Marks, 2009). H3: Party system fragmentation undermines the elite consensus on the EU and increases the likelihood that governments oppose in the Council.

Differentiated politicisation
We argue that the effect of politicisation as a multi-faceted process on oppositional voting in the Council should be most pronounced in policy areas that were especially affected by the decade of crises -that is, economic and financial matters and immigration. First, financial and economic policies have traditionally been among the most controversial policy areas in the EU (Wallace and Reh, 2020: 80). Given that the Euro and financial crises generally increased economic stress in the member states (Genovese and Schneider, 2020;Scharpf, 2015;Wasserfallen et al., 2019), controversies over financial and economic policy, and especially over the EU's role in these policies, became more salient and polarised, thus politicised, in the member states (Schimmelfennig, 2014).
Building on previous studies, we argue that the national politicisation of economic matters is especially likely to be transferred to the EU level (Hooghe and Marks, 2009). While governments, as suggested by Schimmelfennig (2014: 336), 'may have been willing to protect EU-level policy [. . .] under the immediate impact of the crisis', we suggest that the long-term consequences of the crises, which remain in indebted countries, may lead to a further decline of EU support where voters increasingly turn to Eurosceptic parties. This implies that governments' formation without such parties becomes complicated, and governments are perhaps less willing to protect the EU in such matters (Schimmelfennig, 2014: 336). This, in turn, should enhance governments' oppositional voting in the Council when dealing with economic and financial issues.
Second, the refugee crisis of 2015 has enhanced the already heated national controversies over immigration (Green-Pedersen and Otjes, 2019). Additionally, the EU's policies on immigration have strengthened parties with anti-immigrant and Eurosceptic positions (Vasilopoulou, 2018) and researchers identified a solid connection between these anti-immigrant positions and an increase in Euroscepticism (Stockemer et al., 2020). As the refugee crisis had stark consequences in all member states (Trauner, 2016), we argue that the national conflict over immigration should manifest itself in a greater probability that governments will oppose policies that concern the area of justice and home affairs in the Council (Roos, 2019).
H4: The effects of politicisation (described under H1-H3) on oppositional voting in the Council are more pronounced in policy areas challenged by recent crises. Specifically, politicisation most strongly affects Council voting in economic and financial affairs, the internal market, and justice and home affairs.
In contrast, we expect that governments' voting in all other policy areas in the Council is primarily explained by economic factors (see Bailer et al., 2015) rather than politicisation.

Data and methods
The data in our study comprise each government's individual position on all adopted legislation from January 2010 to December 2019 in the Council. This totals 29,400 individual votes on 1050 legislative acts. We use a binary approach for our dependent variable where governments are either coded as being in favour of a legislative act or opposition (defined as votes against or abstentions).
The data on all member states' voting behaviour in this period were gathered from the public register of the voting results published on the Council's website. 1 We collected the voting results on all legislative acts (regulations, directives, decisions, positions, and others 2 ) and categorised them by policy area, type of act, action by the Council, the voting rule, the legislative procedure, and the voting results for each member state. As scholars have criticised the instance of missing values in the Council's public register for voting data (e.g. Roos, 2019), we compared each vote in the public register and the retrospective voting results with the voting documented in the monthly summaries of Council's acts, which are also officially available at the Council's website. 3 Our analysis comprises all legislative cases that were adopted by qualified majority voting and unanimity. However, as votes against are not possible in the case of unanimity when adopting a case, we also include a model that comprises only qualified majority voting. The period from 2010 until 2019 was chosen since it provides the opportunity to cover one decade after the Lisbon Treaty came into effect and the United Kingdom (UK)'s membership period before leaving the EU.
Our study also has some clear limitations. Proposals that failed to gather the needed majority for approval are not included as they are not available on the Council's website. Moreover, member states sometimes choose not to record a public dissent due to national conflicts, despite officially contesting the proposal. Additionally, our study does not include country statements. While opposition or disagreement can also be expressed by governments in the form of country statements, without necessarily voting against or abstaining (Arregui and Thomson, 2014;Hagemann, 2007), we choose to focus on governments' voting behaviour for three reasons. First, we perceive oppositional voting as the strongest official form of dissent. Second, country statements involve all types of government statements and often do not signify an opposition per se. For example, countries also publish statements when they perceive that the act is not comprehensive enough in a specific area or they disapprove of minor administrative issues, such as translation questions. Lastly, the country statements are published more frequently on certain topics (e.g. fisheries) whereas voting follows a more similar pattern in all policy areas.
The Council publishes 21 different policy fields in its public register, which we classified into the different official Council formations. 4 Our analysis focuses on the six most recurrent policy areas or Council formations: agriculture and fisheries; economic and financial affairs; transport, telecommunications and energy; environment; internal market; and justice and home affairs. These areas account for 88% of all cases. The remaining 12% of votes are summarised under 'others'.
We analysed votes in the Council in mixed-effect logistic regression models. In all models, the dependent variable was whether or not a vote was in favour of an initiative. 5 Since votes were nested within legislative acts and countries, the model includes random effects for both sources of dependencies within the data.
To test hypotheses 1 to 3, the model includes fixed effects for variables of politicisation in each member state during the time of voting in the Council: the party system fragmentation within the parliament, Euroscepticism separately for national governments and parliamentary oppositions, and a left-right wing positional score (RILE), also separately for government and opposition. RILE is also included as a square term in the regression model to test for non-linear effects. Therefore, while the linear term tests whether it matters where parties in government/opposition positioned themselves on a left-right spectrum, the square term specifically tests H2 -whether it mattered how extreme a government/opposition was on this spectrum in either direction. The variable party system fragmentation is based on the Comparative Political Data Set (CPDS), 6 specifically, Rae's index (field 'rae_leg' in CPDS), with high values implying high party system fragmentation. Since the most recent data available ended in 2018, the 2018 values were copied to 2019.
RILE and Euroscepticism are based on the Manifesto Project Dataset (CMP) version 2020a. 7 While the RILE was directly available in the CMP per party, this was not the case for Euroscepticism. We, thus, followed the approach through which the RILE was calculated in the CMP and calculated Euroscepticism as a balance measure between positive and negative statements regarding the EU: Euroscepticism ¼ negative statements À positive statements negative statements þ positive statements Finally, to compute the RILE and Euroscepticism per government and opposition in each country and year, we used data from the Parliaments and Governments Database (ParlGov) 8 to compute, during the time of the vote, the average weighted mean RILE/Euroscepticism of all parties in the government/ opposition, weighted by their proportion of seats in the government/opposition. More formally, for example, in the case of the RILE of governments: As in Hobolt and de Vries (2016), we operationalised Euroscepticism within a country's population as the proportion of survey participants that answered the country's membership in the EU is a 'good thing' in the most recent edition of Eurobarometer.

Control variables
As research repeatedly has shown that macroeconomic factors strongly affect EUdecision-making (Bailer et al., 2015;Wasserfallen et al., 2019), our models also include sector-specific macroeconomic variables taken from Eurostat. Since economic variables tend to be correlated, we only include variables that are most theoretically relevant for each policy area in our study. To avoid multicollinearity in our regressions, we also exclude pairs of variables that show a correlation above the 0.5-level. 9 Since all fixed effects in the regression models are standardised, the effect sizes of the estimates of the macroeconomic variables in our models can be directly compared with the politicisation variables. Therefore, our macroeconomic variables also serve as baselines in terms of effect size.
First, for agriculture and fisheries, we included the percentage of the gross domestic product (GDP) in the country that is generated by these sectors. We, thus, account for the relative importance of economic interests in these sectors. Second, for the policy areas economic and financial affairs and the internal market, we control for the national economy's general strength by estimating the impact of GDP per capita, the annual growth rate, and the debt-to-GDP ratio. Third, for the areas transport, telecommunications and energy, and environment, we control for resource productivity, which measures how much energy is consumed in the production and transportation of goods relative to the GDP. Moreover, in the area of environment, we control for the per capita emissions in CO 2 equivalents. Fifth, we control for the unemployment rate and the per capita number of new immigrants for justice and home affairs. We also include a dummy to account for the periods before and after the 2015 refugee crisis. Finally, following the argument that countries are more likely to vote in favour of an initiative put to a vote in the Council if they hold the presidency (Hosli, 2007;Mattila, 2004;van Gruisen et al., 2019), all models include a dummy measuring whether a country held the presidency in the Council at the time of the vote.

Results
In this section, we first present an analysis of patterns in oppositional voting between 2010 and 2019. We then analyse how politicisation at the national level explains governments' oppositional voting in the Council. The comparison between policy areas in Table 1 shows that at least one country voted against or abstained from voting in one-third to one-half of all legislative acts. Compared to studies on the pre-Lisbon era that report approximately 16% contested decisions (Bailer et al., 2015), we find with 36% contestation a stark increase in member states' contestation after Lisbon (see also Wallace and Reh, 2020: 79-81).
Most contestation occurred in the policy area of environment and others (see details for this category in the Online appendix), where approximately one-half of all legislative acts were contested. This is followed by the internal market; transport, telecommunications and energy; agriculture and fisheries; and, lastly, justice and home affairs. We observe the lowest level of contestation (28%) in economic and financial affairs. As pre-Lisbon studies found oppositional voting mainly occurring in agriculture, internal market, and transport (Hayes-Renshaw and Wallace, 2006: 56;Van Aken, 2012: 35), our results indicate that contestation in the EU institutions is no longer path-dependent as previously argued (Hayes-Renshaw et al., 2006: 184). We argue that this finding warrants future scholarly attention.
Moreover, when considering differences between member states, we find that the level of opposition in all policy areas except environment is by far the greatest in the UK. In contrast, oppositional voting on environmental issues primarily stemmed from a cluster of Eastern European countries (Bulgaria, Poland, and Hungary). Furthermore, we see that Germany frequently opposes in the internal market and the area of justice and home affairs (together with Austria) and Austria frequently opposes in the category 'others', whereas the Netherlands more frequently oppose policies in the area of agriculture and fisheries. In contrast, we find the lowest level of oppositional voting in France, Cyprus, and Lithuania (Table 1).
We, further, see that in most cases, one single member state opposed the legislative act. However, to visualise the simultaneous occurrence of oppositional votes among countries, Figure 1 presents the results of a cluster analysis; the forms surrounding countries indicate the co-occurrence of oppositional voting whereas the size of the form visualises the frequency of oppositional votes (details in the Online appendix). To examine differences over time, we split all legislative acts based on the year of the vote into two separate groups of equal size (2010-2014 and 2015-2019). The finding of this comparison is straightforward. First, we see a stable cluster of countries that most frequently opposed in the Council between 2010 and 2019, including the UK, Ireland, and Denmark. Second, while Germany's oppositional voting was relatively independent of other countries before 2015, it tended to align with the Netherlands, Belgium, and Austria from 2015 onwards. Lastly, and most importantly, after the 2015 refugee crisis, a new cluster of oppositional votes developed among the Visegra´d Group (the Czech Republic, Hungary, Poland and Slovakia). 10 Returning to the oppositional votes by countries in the different policy areas over time, Figure 2 shows no clear time-trend. This is also confirmed when considering the different policy areas separately (see the Online appendix). However, we find considerable variation across years with two peaks in the overall level of oppositional voting occurring in 2011/2012 and 2016. The peak in 2011/2012 is primarily traced back to the area of economic and financial affairs, where decisive legislation has been adopted and thus, increasingly opposed, in the aftermath of the economic crisis. In contrast, the low value of oppositional voting in 2016 and 2017 occurs with a sharp decrease in the number of legislative acts in economic and financial affairs and environment. Therefore, to prevent policy area specificities from confounding the overall analysis effects, all regression models were estimated separately for the different policy areas. The Online appendix shows our politicisation variables' time-trends, which indicate a clear increase in governments' and national population's Euroscepticism (see also Winzen, 2020). Interestingly, we see no such time-trend for the Euroscepticism of oppositions and other variables, indicating partial independence of these variables. The Online appendix also includes a correlation matrix of our politicisation variables, further indicating a relatively high degree of independence.
Turning now to our hypotheses, Table 2 presents the estimates of logistic regression models with the vote of a member state as the dependent variable (oppositional voting ¼ 1). All models include random effects per country and legislative act. Summarised, Table 2 shows that different facets of politicisation affect the oppositional voting by member states in the Council in various policy areas. Furthermore, we see in Table 1 that between different policy areas, oppositional voting is primarily driven by characteristics of the government or the opposition.
Based on the estimates in Table 2, we see that H1a, which expects more Eurosceptic governments to oppose more often in the Council, is only supported in the policy areas of agriculture and fisheries, economic and financial affairs, and internal market. Thus, an increase in governments' Eurosceptic positions only raises the probability of governments' oppositional voting in policy areas affected by the economic crisis. This strongly indicates a tendency in which economic policies in the EU are increasingly politicised (de Wilde and Zu¨rn, 2012), and affected by domestic politicisation (Schimmelfennig, 2014), which supports H4. The fact that more Eurosceptic government parties oppose more often in the area of agriculture and fisheries supports the argument that EU actors deviate more in this area from the classical political party position as they face increased electoral incentives to 'vote nationally' (Hix and Marsh, 2007;van der Eijk and Franklin, 1996).    Note: *p<0.1, **p<0.05, ***p<0.01 All variables are standardised, and all models include random-effects per country and legislative act. Values in brackets indicate standard errors.
Based on these findings, we suggest that governments' Euroscepticism plays a pivotal role in explaining oppositional voting in the Council. Since all variables in the models are standardised, we see that Euroscepticism is as important as the most relevant macroeconomic variables in explaining governments' voting in the Council -given that the absolute size of individual estimates is on the same order of magnitude.
By contrast, for the area of justice and home affairs, we find that oppositional voting is only affected by the parliamentary opposition's Euroscepticism, which supports H1b assuming that Eurosceptic positions among opposition parties enhance governments' probability of oppositional voting in the Council. The findings in Table 2 indicate that higher degrees of Euroscepticism among the parliamentary opposition, which is often coupled with an anti-immigrant agenda, places pressure on governments to adjust their policy positions in the Council in the specific area of justice and home affairs. One may argue that governments in the wake of the refugee crisis, and as a result of increased politicisation of immigration policies, cannot resist the pressure from the opposition in this specific area. Moreover, this finding can be traced to the fact that immigration is crucial to 'steal the mainly right-wing opposition's thunder' and prevent vote losses in the next election (see also Meijers, 2017). It is also worth noting that voting in the Council in this policy area seems completely dissociated from the number of new immigrants that a country had during the time of voting or the national economy's growth trajectory, speaking further in favour of the increased politicisation of this area.
Similarly, we see that the Eurosceptic opposition also affects oppositional voting in the policy area environment, which supports H1b. This indicates that increased Eurosceptic positions of national opposition parties complicate a consensus in the Council on environmental issues. However, oppositional voting in this area is primarily predicted by the economy's resource productivity, which had by far the largest effect size of all variables in all our models.
Surprisingly, and in contrast to previous studies (e.g. Hagemann et al., 2017), we reject H1c, which states that governments react to a more Eurosceptic public opinion by more frequently opposing in the Council. The insignificance of a Eurosceptic population provides evidence that Euroscepticism expressed through national parties sufficiently explains voting in the Council. Additional variables on the general public support on European integration do not add any explanatory power.
Thus, we find clear evidence that governments are primarily driven by their own (Eurosceptic) position in agriculture and economic policies. Instead, governments seem to react to pressure from the parliamentary opposition in justice and home affairs. This strongly supports H4, arguing that the effects of politicisation are specifically pronounced in policy areas that were most strongly affected by the economic and refugee crises.
We observe a similar pattern regarding the stance of parties on the left-right spectrum (RILE). On the one hand, we note that governments' left-right position primarily predicts oppositional voting in economic and financial affairs and the internal market. On the other hand, oppositional voting in agriculture and fisheries; transport, telecommunications and energy; and others are primarily predicted by the left-right position of the parliamentary opposition.
Interestingly, these results contradict H2a and H2b suggesting that more extreme positions on the left-right dimension increase oppositional voting probability. Since Table 2 includes coefficients for RILE and RILE-squared, RILE as a variable needs to be interpreted as a polynomial. Figure 3 shows the form of these polynomials per policy area in which any part of the polynomial is significant (i.e. the marginal effect of RILE). In contrast to our hypotheses, we see that oppositional voting in economic and financial affairs and the internal market is, in fact, more likely when governments are located at the relative centre of the left-right spectrum. However, it is worth noting that most Eurosceptic radical right parties take centre positions on the left-right dimension in these areas (Rovny and Polk, 2020).
This indicates that oppositional voting in the Council is mainly driven by practical and non-ideological concerns or parties' positions on other issue dimensions. The results show the same tendency when considering the left-right positions of the parliamentary opposition. Oppositional voting in agriculture and fisheries; transport, telecommunications and energy; and others are most likely when governments are faced with opposition parties located in the political centre. Thus, government parties are least likely to take an oppositional stance in the Council if opposition parties exert pressure from only one end of the left-right spectrum. Thus, we conclude that extreme positions on the left-right dimension have become less important in governments' likelihood to oppose EU policy-making, which   Table 2, showing country-level variance not explained by any of the other predictors.
Note: The 95%-confidence intervals were bootstrapped to avoid parametric assumptions.
contradicts pre-Lisbon studies identifying a higher likelihood of right-wing parties to oppose (Hagemann, 2007;Mattila, 2004). It is important to note that the effect of Euroscepticism was estimated independently of party positions on the left-right spectrum.
Surprisingly, we find no support for H3, which states that party system fragmentation undermines the elite consensus on the EU and enhances the likelihood that governments oppose in the Council. A possible explanation for this lack of evidence may be that party system fragmentation is often associated with an increasing polarisation across the left-right spectrum and questions concerning the EU. Furthermore, the fact that we study fragmentation in isolation of these factors leaves little variance that fragmentation could explain. Finally, it is worth noting that with the EU having only 27-28 members during the 10 years we studied and governments spanning several years, the effective sample size on which our statistics are based is relatively small. Even though this is a common problem in country-level comparisons within the EU, and our model specifications account for these dependencies within the data, this nuance needs to be kept in mind when interpreting individual estimates.
In all, our analysis provides strong support for H4 and our argument that politicisation as a multi-faceted process in the national arena affects oppositional voting in the Council differently across policy areas. This form of 'differentiated politicisation' that we identified among the policy areas may also increase differentiated integration among the member states (Schimmelfennig et al., 2015). To indicate the amount of variance in voting behaviour among countries that is not explained by our variables of politicisation and macroeconomic controls, Figure 4 shows estimated random intercepts per country. We see that the unexplained variance is highest in the policy areas of economic and financial affairs and justice and home affairs. In other words, these are the areas where further country-specific characteristics are likely to affect voting behaviour, which warrants further scholarly attention.

Conclusion
This article examines how domestic politicisation, conceptualised as a multifaceted process that involves an increase in saliency and polarisation, is uploaded to the EU level in the form of oppositional voting by member states' governments in the Council. We combine a new dataset on roll call votes for all 28 member states in the Council between 2010 and 2019 with national-level data on governments' and oppositions' Euroscepticism and left-right positions and the national fragmentation of the party system. We contribute to the literature on EU decisionmaking and politicisation by finding strong evidence of 'bottom-up' politicisation (Bressanelli et al., 2020) where domestic national party politics, either directly driven by governments or indirectly channelled through oppositions -depending on the policy area -substantially affect voting in the Council. This contradicts the view that the Council is insulated and distant from domestic political developments (Schimmelfennig and Winzen, 2020: 83-86).
More precisely, while governments' Euroscepticism generally increased in the post-Lisbon era, the effects of Euroscepticism on oppositional voting are only apparent in agriculture and fisheries, economic and financial affairs, and the internal market. This suggests that in the aftermath of the economic crisis, parties in governments increasingly combined Euroscepticism with economic agendas. In contrast, in justice and home affairs and environment, the only variable of politicisation explaining governments' oppositional voting was the opposition's Euroscepticism. This suggests that Eurosceptic opposition parties exert pressure on governments in these areas, who subsequently adjust their positions in the Council. In the context of the refugee crisis, and due to increased relevance of environmental topics, one may also interpret this as evidence for governments' strategic voting behaviour to prevent future vote losses to opposition parties (Meijers, 2017).
Concerning the impact of parties' position on the left-right dimension on Council voting, our findings contradict the previous literature (Hosli, 2007), and suggest that voting is primarily driven by government and opposition parties with a centrist position on the left-right dimension. This may be traced back to the fact that the main opposers of the EU, radical right parties, tend to blur their left-right position and take a centrist position on economic issues (Rovny and Polk, 2020). Similar to our variable Euroscepticism, we find that governments' position on the left-right dimension only predicts voting in economic and financial affairs and the internal market, thus, again in areas affected by the economic crisis. In contrast, opposition parties' left-right positioning predicts voting in agriculture and fisheries; transport, telecommunications and energy; and our category other. We, therefore, find again evidence that governments adjust their positions in the Council as a response to the ideological positions of the national opposition. Interestingly, in the areas of economic and financial affairs and internal market, governments are the main driver for position-taking at the EU level whereas all other policy areas (with the exemption of agriculture and fisheries which reveals mixed results) are left to the national opposition and therefore, indirectly impact governments' voting. This speaks for a pronounced interest of governments to be the main promoter of economic policies at the EU level and supports the argument that the EU creates the incentive for agreements on and the realisation of economic undertakings at the EU level while simultaneously evading the political costs at the national level (Bartolini, 2005: 246-247).
Taken together, we find strong evidence for differentiated politicisation which, in turn, may increase differentiated integration -that is, an integration of different policy areas at different rates or an integration tendency of only some member states in certain policy areas (Schimmelfennig et al., 2015). Therefore, the future of the EU is also dependent on the question of how the different facets and dynamics of politicisation manifest not only in the Council but also in other core EU institutions. While the uploading effects may be stronger in intergovernmental institutions, such as the Council, we suggest that they are also relevant for the second legislator, the European Parliament. Future work, therefore, should examine the effects of national politicisation on decision-making in the European Parliament. In all, further research should address the question of how EU politicisation -the increase in saliency and polarisation over the EU -transforms domestic political conflicts and how these conflicts are integrated into EU policy-making. This especially applies in times of crises such as the current Corona crisis and future crises, which have the potential to enhance the competition over salient topics among domestic political parties and, thus, may alter EU decision-making substantially.