Talking cheap, or speaking Euro? Assessing congruence between leaders’ communication and negotiators’ positioning in the Eurozone crisis*

ABSTRACT

Whether European leaders ‘talk the walk’ is a key question in EU studies. Prior research has made substantial inroads into explaining how EU and national leaders communicate about European integration, as well as what positions member states and EU institutions take in EU negotiations and why. However, we do not know to what extent leaders’ communication is congruent with what their negotiators do in Brussels, nor what explains such (in)congruence. This article presents a first dive into these questions. It formulates two theoretical arguments on why leaders’ communication and negotiation positions may diverge: accidentally, due to coordination costs, or intentionally, due to strategic and electoral calculations. Leveraging three novel databases on the Eurozone crisis, we come to two important conclusions: first, the degree of incongruence is substantial and similar to what is found at the national level. Second, our results favour a political-strategic argument for incongruence, but many important questions remain.

KEYWORDS:

• European Union
• responsiveness
• politicisation
• leadership
• communication
• cognitive mapping

Introduction

Do European leaders talk cheap or ‘talk the walk’? Since Putnam’s (e.g., Putnam, 1988) seminal article on two-level games, it is widely recognised that EU negotiations play out across multiple levels. The complexity of EU politics both implies high coordination costs for national governments and makes it difficult for citizens to assign responsibility for negotiation outcomes to their representatives at the EU level (Schneider, 2018, 2020). This occurs despite the politicisation of EU governance, which means EU leaders are increasingly constrained by public opinion and political opposition domestically (Hooghe & Marks, 2009; Van der Veer & Haverland, 2019). Given this context, in which the complexity of EU decision making yields a public accountability deficit while politicisation simultaneously creates incentives to deflect blame for unfavourable negotiation outcomes (Schimmelfennig, 2020), the positions of member state negotiators during EU-level negotiations may deviate from what political leaders communicate to their constituents.

Few studies have explored to what extent what leaders communicate reflects what negotiators do at the EU level, despite the importance of transparent leader communication to their national constituents in terms of democratic accountability. The literature on political responsiveness has examined the signalling of responsiveness by national governments to their constituencies mostly through the consistency of negotiation positions over time (e.g., Schneider, 2018, 2020; Wratil, 2018), or the public legitimation of EU policy making in terms of leader communication (Rauh et al., 2019; Traber et al., 2020). To the best of our knowledge, however, there is no research examining the congruence between government negotiation positions at the EU level and leaders’ communication about those positions to domestic audiences. There are also significant knowledge gaps regarding the specific mechanisms that link public opinion to the (domestic legitimation of) policy positions of actors at the EU level (Zhelyazkova et al., 2019) and thus there is a need to open the black box between voter preferences and policy output (De Wilde & Rauh, 2019). This article seeks to contribute to these debates by examining the question of what drives (in)congruent communication: the extent to which high political leaders’ communication (does not) reflect the positions advocated by their negotiators during EU-level negotiations.

In this article, we provide the first large-N study that explicitly compares EU leaders’ communication on EU-level negotiations with the actual positions taken by negotiators during those negotiations. This allows us to better understand to what extent and under what conditions congruence and discrepancy between action and public communication occurs with regard to European integration. In doing so, this project speaks to vital questions of democratic accountability, leadership and responsiveness in multi-level EU governance. Our aim is twofold: we first seek to establish whether and to what extent we can observe incongruence between positioning and communication, and secondly, we examine what factors drive this incongruence. We do so by juxtaposing two competing theoretical arguments that explain why position-communication incongruence may come about, which both offer alternatives to our null hypothesis that communication is faithfully congruent with position-taking.

The first theoretical argument leans into the notion that multilevel EU negotiations are highly complex and acknowledges that usually EU political leaders do not actually conduct negotiations – or may not always even determine – negotiation positions themselves. In democratic systems, the mandate of EU negotiators is set through a complex process of political bargaining at the national level (Bailer, 2010; Frieden & Walter, 2018; Moravcsik, 1998). Hence, the coordination of the negotiation process itself may produce barriers to the implementation of domestically-agreed negotiating mandates. Previous research into the fulfilment of campaign pledges, for instance, finds that on average government parties fulfil 60 per cent of their pledges partly or fully, but these scores vary quite significantly depending on the political system and country (Thomson et al., 2017). As such, it is to be expected that leaders’ communication will at times divert from negotiators’ positions at the EU level.

The second account is more closely aligned with research on leaders’ strategic responsiveness in an increasingly politicised EU context (Rauh et al., 2019; Schneider, 2018; Traber et al., 2020). It assumes that incongruence arises intentionally as the result of strategic decision-making by leaders, who may omit information or bend the truth in their public communication when they seek to pursue unpopular policies but fear the electoral consequences thereof (Hood, 2010; Traber et al., 2020). Depending on the context, leaders have more or less incentives to do so and are presented with different opportunities to communicate strategically without bearing the consequences (Davis & Ferrantino, 1996; De Vries et al., 2021).

Our empirical focus is the negotiations regarding the management of the Eurozone crisis, because this offers us a unique opportunity to examine position-communication congruence via the combination of two novel data sets: a data set including (1) national positions in 40+ negotiations on EMU reform during 2010–2015 (Wasserfallen et al., 2019) and one with (2) publicly communicated policy ideas of heads of state and government, national central bank governors and EU-level leaders (such as Commission presidents) spanning the same period (Müller & Van Esch, 2020; Van Esch, 2007; Van Esch et al., 2018). We subsequently examine the extent to which leaders’ public communication is congruent with the position of the negotiators representing their member state or institution, using a number of explanatory factors associated with the two theoretical arguments outlined above.

Theory

High political and financial leaders – as the main spokespeople of their country or institution – have always played a central role in political communication; a role that has strengthened due to modern developments like the personification and mediatisation of politics (Fotopoulos & Morganti, 2021; Hubé et al., 2015). As the EU is commonly characterised as a distant political system, the prominence of leaders’ communication in cueing national publics is especially pronounced the EU context (Antonakis & Atwater, 2002; Popper, 2013). Absent societal politicisation (De Wilde et al., 2016), domestic publics are largely unaware of what happens in the Brussel’s bubble and depend for a large part on the communication by high political and financial leaders at the national and EU level. While the EU’s poly crisis has stimulated the emergence of a pan-European public space (Zeitlin et al., 2019) and measures have been taken to increase the transparency of the EU, overall the public interest in and knowledge of the EU remains low compared to national political systems (Hillebrandt, 2021; Hix, 2015). Moreover, key moments in EU-negotiation processes, whether during crisis or routine policy-making, still take place outside of the view of the public (e.g., trilogues), further limiting citizens’ insight into EU decision making. Instead, the public relies heavily on the strategic and symbolic representations of what goes on in the political arena as communicated by national and EU leaders and as relayed by the media.

The literature on the legitimacy of the EU focusses predominantly on input and output legitimacy and the responsiveness of EU elites to the concerns of citizens (Rauh, 2019; Schneider, 2018; Van der Veer, 2022). However, open and true communication plays an important part in terms of throughput legitimacy (Schmidt, 2013). As stressed by works on transparency, the quality of citizens’ input into the democratic process depends on the amount, quality, completeness and truthfulness of information and knowledge. Moreover, leaders can only be held to account for the output they generate when citizens have enough and adequate information. Given the key role high political leaders play in informing citizens in the EU context, truthful public communication by these leaders thus plays a central role in the legitimation of EU governance. However, as stated in the introduction, two different strands of research in political science offer alternative accounts that suggest that incongruence may occur between leaders’ communication about the negotiation positions taken by ‘their’ negotiators in EU negotiations and the actual positions taken by negotiators at the negotiating table.

The first is rooted in the literature on position coordination in politico-administrative systems at both the EU and national level (Gärtner et al., 2011; Kassim, 2014; Panke, 2010). As is the case at the national level, there are good reasons why the publicly asserted positions of leaders on the EU do not always correspond to the positions that are being negotiated behind closed doors. First, while leaders may exert more influence than other political actors – especially during crises – in no political system do they determine policy by themselves. Decision making power is always shared by different actors, and the extent of such sharing depends on the political system. Heads of state and government of parliamentary or federal systems or those heading a coalition government may be expected to have less direct influence on policy than those of (semi)presidential or unitary system or those having a majority in parliament. In addition, the strength and political independence of the national civil service, may influence how much leaders’ professed preferences divert from the actual negotiating positions (Gärtner et al., 2011).

In addition, and specific to the domain of EU policy making, leaders do not determine mandates for EU negotiations by themselves, nor do they conduct most of the negotiations. The development of EU negotiation positions is a process that goes through various steps and engages multiple national and EU actors. The process differs between member states and the type and political saliency of a policy. In most EU member states, the coordination of EU policy is conducted by one lead ministry or the permanent representation (PR). To formulate the negotiation mandate, the lead ministry or PR consults and coordinates the issue at hand with other affected ministries, again to varying degree in different member states. The national parliament is also involved, but again to a different degree: ranging from receiving mere ex-post information to holding veto rights (Gärtner et al., 2011; Kassim, 2014; Ministerie van Buitenlandse Zaken, 2021; Panke, 2010). Once the instructions and the margins of manoeuvre are clear, they are sent to the PR which is in charge of the negotiations. In the ideal situation, the PR reports back to the national level during the negotiation process, so instructions may be adapted accordingly. Finally, despite studies showing mixed evidence (Connolly & Kassim, 2016), officials working in the Brussels’ bubble are often assumed to be more EU minded, leading to the hypothesis that if coordination is in the hands of the PR, it is likely that the negotiating position is more pro-European. The literature on politico-administrative coordination thus yields the following hypotheses:

H1a Speech acts by national leaders from political systems with more diffusely distributed decision-making powers will be more likely to deviate from the positions taken by their member state’s negotiators in EU level negotiations;

H1b Speech acts by national leaders from political systems with more decentralised coordination of EU negotiations will be more likely to be more Eurosceptic than the position taken by their member state’s negotiators in EU level negotiations.

The second account stems from the literatures on political communication (Rauh et al., 2019; Traber et al., 2020) and politicisation management in the EU (Bressanelli et al., 2020; Schimmelfennig, 2020), which recognise that much communication by political leaders is strategic in nature. In order to send specific signals to domestic constituencies, or to strategically influence the behaviour of other countries, leaders may omit facts, downplay problems, shift blame or blatantly lie (Frieden, 1999; Putnam, 1988).¹ Although the notion of strategic communication by political leaders is neither new nor specific to a particular political system, different political systems and contexts do offer more or less incentives and opportunities for leaders to strategically create incongruence between what happens at EU negotiations and how they communicate about those negotiations to the wider public – and whether they can get away with it.

Focussing first on drivers of incongruence as recognised by these literatures, the multi-level and fragmented nature of the EU decision making means that decision are often either taken collaboratively or presented as such (Schimmelfennig, 2020); this holds even for decisions taken by single EU institutions such as the European Council or Council of Ministers. This makes it difficult for publics to attribute credit and blame, and to hold individual leaders accountable. Moreover, while the transparency of EU decision making has increased over the years, key aspects of negotiations are still outside public view (Brandsma & Meijer, 2022). This creates incentives for vote-seeking leaders to signal responsiveness in their communication to specific constituencies even if policy-seeking incentives push leaders to compromise and ‘hidden cooperation’ at the negotiating table (Schneider, 2018). Moreover, strategic communication may become especially prevalent when politicians are faced with policy failures or crises that they fear may have electoral consequences if they are found to be responsible for them (Traber et al., 2020).

Political leaders do not have free reign in terms of what they communicate to domestic publics, however, and are constrained by the degree to which publics are aware of EU negotiations (Hooghe & Marks, 2009). EU politicisation has steadily but heterogeneously increased since the early 90s and is characterised by a growing salience of EU issues in national public spheres, increased polarisation of attitudes towards the EU, and the expansion and mobilisation of actors and audiences engaged with EU affairs (De Wilde et al., 2016). Recent research on the societal politicisation of EU governance suggests politicisation may both constrain as well as enable strategic communication by political leaders (Bressanelli et al., 2020; Schimmelfennig, 2020): as politicisation means publics are more aware of EU affairs, higher levels of politicisation may decrease opportunities for leaders to strategically create incongruence between communication and the positions actually forwarded by their negotiators.

However, national leaders are still the main transmission belt between EU-level negotiations and domestic publics, and are by far most visible in domestic public spheres. Recent research finds that these leaders’ tend to hedge against electoral losses when domestic constituencies become more Eurosceptic by becoming more negative towards the EU in their public communication to these audiences (Rauh et al., 2019). The above suggests EU politicisation may push leaders towards strategies of reactive politicisation management, including blame avoidance, by strategically distorting communication on what they are trying to achieve in EU negotiations (Bressanelli et al., 2020). Research on responsiveness to politicisation shows that elites are especially likely to signal responsiveness to Eurosceptic audiences under the specific configuration of EU politicisation in which Euroscepticism is widespread among both voters and the parties representing them (Rauh et al., 2019; Van der Veer, 2021).

These mechanisms underpinning leaders’ strategic communication from the politicisation literature are supported more generally by literature on political campaigning and blame games in the EU context, which teaches us that negative campaigning and blaming the EU is electorally beneficial and prevalent (Davis & Ferrantino, 1996; Traber et al., 2020). Positive campaigning – promising to deliver beneficial EU policies and outcomes prior to negotiations – may lead to scrutinisation once policies are implemented. Negative campaigning and shifting blame are electorally less risky (Davis & Ferrantino, 1996). Taken together, this leads us to the second set of hypotheses:

H2a Speech acts by leaders facing more polarisation of public opinion on the EU in their home country will be more likely to be supportive of a European solution than the positions taken by their member state’s negotiators in EU level negotiations;

H2b Speech acts by leaders facing a more Eurosceptic national parliament in their home country will be more likely to less supportive of a European solution than the positions taken by their member state’s negotiators in EU level negotiations;

H2c The deviation of speech acts by leaders towards more Europeanised measures will be strongest under constellations of EU politicisation in which they face both strong public and partisan Euroscepticism in their home country.

In addition, the extent to which leaders would strategically create incongruence between their public communication and negotiation positions is dependent on the repercussions such misrepresentation may have in the short term – i.e., whether voters are capable of sanctioning them (Schneider, 2018, 2020). First, incentives for misrepresentation may derive from the fact that politicians want to be re-elected: strategically downplaying support for EU integration in public communication in the face of a Eurosceptic domestic audience constitutes a strategy of ‘restrained politicisation’ and thus seems more likely for elected officials (Bressanelli et al., 2020; Schimmelfennig, 2020). Unelected leaders, like EU commissioners and the various EU presidents, face fewer incentives to undermine EU integration in the face of politicisation (Davis & Ferrantino, 1996; Schimmelfennig, 2020). Moreover, technocratic leaders (Commission or (E)CB presidents) are less concerned with electoral consequences as they do not have a stake in party political competition (Caramani, 2017). In fact, previous research shows that unelected EU leaders actually have a stronger propensity to defend the European Union in the face of stronger EU politicisation in their home country, albeit using increasingly complex language (Rauh et al., 2019).

For politicians facing elections the intentional public distortion of communication about one’s negotiation position may have serious repercussions (Davis & Ferrantino, 1996): if a politician is perceived to be untruthful, voters may revaluate their support for the politician and her party. Extant research on EU politicisation also shows that national leaders may actively politicise EU issues in domestic election campaigns in order to sway voters that can be mobilised on such issues (Bressanelli et al., 2020; Schimmelfennig, 2020).² Similarly, leaders are inclined to ‘signal responsiveness’ in EU negotiations through publicly commitments that cater to specific domestic constituencies when elections are near, and may delay compromising on those positions during negotiations until after these elections (Schneider, 2018, 2020).³ Hence, we may expect more divergence from the actual negotiating position when elections draw near, as national leaders seek to more strongly cue domestic audiences favourably on their involvement in EU affairs. This leads us to three additional hypotheses based on the literature on strategic communication:

H2d The speech acts of technocratic leaders (central bankers and EU presidents) are less likely to diverge from the actual negotiating position than those of elected leaders (heads of state or government);

H2e The divergence of speech acts from actual negotiation positions will on average be more pro-European for technocratic leaders, whereas it is more Eurosceptic for elected leaders;

H2f Speech acts by elected leaders facing elections in their home country will be more likely to be more Eurosceptic than their associated negotiation positions in EU level negotiations.

Data and methods

To test these hypotheses, we combine three datasets on: (1) national positions in 40 + negotiations on EMU reform during 2010–2015 (Wasserfallen et al., 2019), (2) cognitive maps of the publicly communicated policy ideas of national leaders during the same time (Müller & Van Esch, 2020; Van Esch, 2007; Van Esch et al., 2018), and (3) E(M)U politicisation (Van der Veer, 2021; Van der Veer & Haverland, 2018).

EMU choices data-set

The positions of the EU member states and EU institutions in negotiations concerning the Eurozone crisis are derived from the ‘EMU Positions’ dataset from the EMU Choices project (Wasserfallen et al., 2019). This dataset was constructed based on a methodology similar to the decision-making in the EU (DEU I and II) projects (Thomson et al., 2012; Thomson and Stokman, 2006) and includes data on policy preferences of EU member states and EU institutions in the formal negotiation stages between 2010 and 2015 (the ‘contested issues’), as well as the saliency thereof.

On the basis of official EU documents, quality news media, European Council conclusions and EurLex documentation, the following 10 proposals were identified: (a) the initial proposal to support Greece; (b) the EFSF; (c) the ESM; (d) the Six-Pack on fiscal and economic governance; (e) the Two-Pack on the coordination of national budgets; (f) the Fiscal Compact; (g) the Banking Union; (h) the Financial Transaction Tax; (i) Eurobonds; and (j) the ‘Five Presidents’ Report’. For these proposals, a range of contested issues were identified through secondary sources and interviews. The preferences of all member states and EU institutions were coded for all 47 issues, providing a score between 0 and 100. We recode these scores into two binary variables, Negotiator supportive and Negotiator opposed, which indicate the negotiators’ position on the issue at hand. In each case, 0 indicates a neutral position (see Appendix 1 for details). Several of our hypotheses (H2a–c) concern the question as to the extent to which the positions of leaders are more or less supportive of Europeanised measures than those of the negotiators. In order to test these hypotheses, we have omitted the two EMUChoices issues that do concern more or less Europeanisation (SPA5 and FC7).

Cognitive mapping data

The professed positions of national and EU leaders were derived from a database of 85 cognitive maps of 41 different political and financial leaders from 10 EU countries (see Appendix 1).⁴ Cognitive mapping is a method specifically designed to establish the policy ideas of political actors as reflected in spoken or written communication (Axelrod, 1976; Young & Schafer, 1998; Van Esch & Snellens, 2024). In contrast to other text-analysis techniques, the basis of a cognitive map are the causal and normative relationship alluded to in a text or speech act (Axelrod, 1976). These relations can be positive, negative or non-existent. By transforming these relations into a visual graph in which the concepts are depicted as points and the relations between these concepts as arrows, a cognitive map is created. The maps used in this research were derived via hand coding from 292 public speeches⁵ made by the different leaders. To code the texts a coding manual was created following the pre-existing and formalised coding rules for cognitive mapping that have been developed over the years (Bonham & Shapiro, 1986; Van Esch & Joosen, 2015; Wrightson, 1976). These coding rules have reached a state of precision such that intercoder reliability is fully compatible with the accepted standards of good quantitative work in the social sciences (Axelrod, 1976, p. 84).⁶

By analysing the paths in a map it may be established whether a certain idea is supported by a leader. For instance, assuming that ‘solving the crisis’ is considered a positive goal, we can derive from the CM of the Dutch Prime Minister Rutte in Figure 1 that he supports fiscal discipline whereas he does not support wider yield spreads as the former is positively related to ‘solving the crisis’ and the latter negatively. A CM thereby reveals leaders’ positions and whether they are positive or negative. Finally, a CM also provides an indication of the strength of a particular position by assuming that ideas that are mentioned more frequently and in connection to multiple other ideas are more salient to someone’s belief system. The number of connections between ideas is directly visible in a map while how many times an actor refers to an idea is reflected indicated by the weight of a relation. By combining this information, we can derive from Figure 1 that the concept ‘market trust’ has a saliency of three while ‘fiscal compact’ has a saliency of one, showing that the former is more salient for Rutte than the latter.

Figure 1. Excerpt from the CM of the Dutch PM Mark Rutte.

In order to establish leaders’ preferences regarding the contested issues identified in the EMU Choices project, all concepts featured in the maps of the leaders were classified using the EMU Choices codebook (see Appendix 3). To do this, all the issues in the EMU-positions data set were transformed into binary choices.⁷ With regard to the first EMU Choices issue for ‘fiscal compact’, for instance, the three possible positions available in the EMU Choices database (rejection of fiscal compact, allow for opt-outs, all MS in fiscal compact) were reverted to two, making a distinction possible between CM-concepts indicating support for the fiscal compact (regardless of opt-outs) and a rejection thereof.⁸ For instance, as Figure 1 shows that Rutte values the concept ‘fiscal compact’ positively, this thus corresponds to a positive stand to the implementation of treaty in the EMU Choices database.

If multiple concepts in a CM could be connected to a EMU Choices issue, the evaluation for all the concepts pertaining to that issue in a CM are summed up. To make the scores comparable across CMs they are listed as a percentage of the aggregate saliency of all concepts in the map. This leads to an overall saliency score of leaders’ positions on that issue, whereby positive scores indicates support for the issue, a negative score indicates opposition and a score of zero indicates an ambiguous position (Van Esch & Snellens, 2024), moreover the more extreme the score (more positive or negative), the stronger a leader’s position is assumed to be. The saliency scores of leaders’ positions thus combine information about the direction of their stance (positive, negative, ambiguous) and the strength of their position. Following this procedure, the positions of all the leaders on each of the issues in the EMU choice data-set were calculated. If for a particular contested issue from the EMU Choices data no corresponding concepts were found in any of the cognitive maps (such as for the issue Eurobonds in Figure 1), the issue was omitted from the analysis.

Explanatory variables

The measures required to examine our explanatory hypotheses are taken from various sources, most notably a dataset on politicisation in the context of EMU (Van der Veer, 2021; Van der Veer & Haverland, 2018). This dataset provides various politicisation-related indicators such as the polarisation of public opinion on the EU, the strength of Eurosceptic parties in the home country of a leader (mobilisation), and the presence of elections in a given period. It also includes a number of EMU-specific control variables, such as macroeconomic covariates and the economic left-right position of the government the leader represents.

We extend this dataset with measures the degree of control leaders themselves have over negotiation mandates and process. These include the degree of political constraints on the executive, and the level of centralisation in political-administrative relations in the coordination of EU matters in the leader’s home member state (Gärtner et al., 2011; Henisz, 2002).⁹

The final dataset uses leaders-period-issue triads as the unit of observation. Our main issues were derived by collapsing the EMU Choices subcategories into their parent categories (e.g., ESM1 and ESM2 become ESM). Our five periods split the Eurozone crisis into episodes, and the cut-offs are chosen as they represent critical junctures in the crises that may have instigated changes in positions.¹⁰ This does, however, imply that our periods are not always of equal length, nor match the periodisation of our predictor variables. We resolved this issue by taking lagged averages of certain predictors in a way that substantively made sense. For example, for the fourth period, which runs from July 2012 to December 2014, we use annual averages for 2012 and 2013.¹¹ As we are the first to analyse communication-position (in)congruence in EU negotiations, we begin with a descriptive analysis of this (in)congruence to get a sense of the scope of the phenomenon. We subsequently turn to explanation using statistical modelling.

Analysis

Descriptive analysis

The first question raised in this paper concerned the simple but important question: To what extent do high political and financial leaders communicate faithfully about the actual positions taken during the EU negotiations in Brussels? In order to answer this question, we transformed the saliency scores indicating leaders’ communicated positions into categorical scores (omitting the information on the strength of the position), where positive scores indicate support for the proposed policy and a negative score indicates opposition. Neutral scores remained as they were.¹² Table 1 provides an overview of the positions of the leaders and negotiators on the issues in the EMU choice dataset as well as the congruence between them.

Table 1. Positions of national and EU leaders and negotiators (N = 686).

Leaders	Negotiators
	Opposed	Neutral	Supportive
Opposed	134	40	132
Neutral	7	0	14
Supportive	106	23	230

Table 2 provides more details on our results, now split by national and EU leaders. We find that the position of the negotiators and EU leaders are congruent for 49 per cent of our observed issues. In the case of national leaders, the congruence is 54.7 per cent. Moreover, where incongruence is observed, communication by national leaders is roughly as often more supportive of further integration as it is more oppositional compared to the positions taken by their negotiators, whereas EU level leaders’ communication is generally more supportive than the negotiators. This finding seems congruent with prior research indicating communication by EU leaders tends to be more positive (Rauh et al., 2019).¹³

Table 2. Distance between negotiators’ and leaders’ positions.

Score	Meaning	National	Nat. %	EU	EU %
−2	Leaders negative, negotiators positive	88	18.0	18	9.1
−1	Leaders more negative than negotiators	26	5.3	4	2.0
0	Overlap	267	54.7	97	49.0
1	Leaders positive more positive than negotiators	30	6.1	24	12.1
2	Leaders positive, negotiators negative	77	15.8	55	27.8
Total		488	100	198	100

Figure 2 shows the congruence between leaders’ and negotiators’ positions per issue that was discussed during the negotiations. While for EU leaders the congruence between their communication and negotiators’ positions was the highest on the question of whether to install the European Stability Mechanism (ESM), this was the issue where national leaders’ communication diverged most from the negotiators.¹⁴ The second panel of the figure shows the different types of incongruence encountered, indicated by negative values if leaders’ communication was more oppositional than negotiator’s positions, and positive values in case they were more supportive. A value of +2 (or – 2) indicates communication was supportive of (or opposed to) a proposal whereas the actual negotiation position was opposed (supportive). The figure indicates that EU leaders were less supportive of the ESM or providing it with a broad budget and liberal amount of funds than negotiators. For national leaders congruence was highest on the issue of the Eurobonds (EB), with incongruence spread equally across cases where leaders communication was more supportive or more opposed. For many of the other issues we observe around 50 per cent congruence between national and EU leaders and the negotiators. EU leaders’ communication is generally more in favour or Europeanised measures than negotiation positions, while national leaders tend to be more oppositional, especially on the issue of the assistance to Greece and the Six-Pack and SGP.

Figure 2. (In)congruence between positions of leaders and negotiators per issue area. (ESM: European Stability Mechanism; FC: Fiscal Compact; G: Greek Assistance; PR: President Report; SPA: Six-Pack of the Stability and Growth Pact; FTT: Financial Transaction Tax).¹⁸

Split into different periods, our data indicates that for national leaders, the outbreak of the Eurozone crisis (period 0 to period 1) at first increases incongruence between the positions of leaders and negotiators (see Figure 3). Throughout the crisis, however, the level of congruence increased beyond the level observed before the crisis. This may indicate that due to its unprecedented nature, the outbreak of the Eurozone crisis initially upset standard operation procedures, but that the saliency of the crisis soon yielded stronger national coordination. A similar trend is visible for EU leaders until the election of the Greek Syriza-led government (period 4).¹⁵ ‘Supportive’ (+1/+2) and ‘oppositional’ (−1/−2) incongruence between national leaders’ communication and negotiators’ positions is divided equally in each of the periods, except for period 5. In this six-month period between the election of the Greek Syriza government and the Greek bail-out referendum, the few dissenting leaders all opposed more Europeanised measures while their negotiators advocated for them. For the EU leaders, the majority of dissenting voices were again more in favour of European measures, also with the exception of period 5.

Figure 3. (In)congruence between positions of leaders and negotiators per period.

Finally, when splitting out the data between the different member states (Figure 4), the high level of congruence between Danish leaders and negotiators stands out. With congruence standing at more than 75 per cent, this finding aligns with the long term assessment in the literature that Denmark has an excellent internal procedure for coordinating European policy-making (Ministerie van Buitenlandse Zaken, 2021). The only other country where the congruence exceeds 50 per cent is the Netherlands. All other member states and the ECB presidents show 50 per cent or near 50 per cent congruence between the communication of leaders and negotiators’ positions. The European Commission has the lowest congruency score: this incongruence is strongly driven by its leaders communicating more supportively about negotiation proposals than the position-taking of its negotiators. This finding is congruent with the Commission’s role as the ‘engine of integration’ and can be read as its attempt to pre-emptively soften opposition during negotiations. It is also congruent with research indicating EU executives tend to communicate most positively about integration (Rauh et al., 2019). Looking again at the type of incongruence between national leaders and the negotiators, incongruence seems at face value to align quite well with levels of Euroscepsis of governments in the different member states. In Eurosceptic countries like the UK, the Netherlands and Hungary, leaders are far less supportive of more Europeanised measures than their negotiators. While communication by leaders from typically pro-European member states like Spain and Germany is more supportive of further Europeanisation of the management of the Eurozone, compared to negotiators’ positions.

Figure 4. (In)congruence between positions of leaders and negotiators per member state.

Explanatory analysis

This raises the question as to what explains the (in)congruence between the leaders’ communication and the negotiators’ positions. To answer this question, we run a series of multilevel models in which we account for the cross-classification of observations across issues and the polities our leaders represent (e.g., France, Germany or the EU as a whole) by the inclusion of a random intercept for these nesting categories (Schmidt-Catran & Fairbrother, 2016).¹⁶ We centre predictors on their grand mean and standardise them by two standard deviations, as recommended by the statistical literature (Gelman & Hill, 2006): this implies a one unit change in a predictor corresponds to a change from one SD below the mean to one SD above it. As the congruence measure used in our descriptive analysis is rather crude, for the explanatory analysis we opt to revert to the original measure for the positions of the leaders’ that indicate both the nature of the positions (opposition, support, neutral) as well as the strength of the position (see page 12). The conducted analysis predicts the saliency of leaders’ position using our various covariates, while controlling for their negotiators’ position. As not all of our covariates are available for leaders representing the EU as a whole, we use three specifications: model A1 includes both national and EU leaders but excludes Centralisation, CBI and constraints (N = 685); model A2 is identical to A1 but excludes EU leaders (N = 487); model A3 adds the remaining predictors to model A2. The results of these models can be found in Table 3.¹⁷ We repeat this analysis using a second outcome variable (position strength) that includes the absolute values of the saliency of leaders’ position (thus omitting the direction of their position) to gauge under what conditions leaders communicate in a stronger fashion (regardless of whether they are positive or negative, see Models B1-3 in Appendix 1).

Table 3. Results of multilevel regression models predicting saliency of leaders’ positions.

	A1		A2		A3
Coeffcient	Estimates	P-Value	Estimates	P-Value	Estimates	P-Value
Intercept	0.45 (−2.25–3.16)	0.742	0.49 (−0.32–1.29)	0.234	0.40 (−0.43–1.23)	0.341
Negotiator: Opposed	0.57 (−0.39–1.54)	0.245	−0.06 (−0.91–0.79)	0.887	−0.10 (−0.96–0.75)	0.812
Negotiator: Supportive	0.88 (−0.03–1.78)	0.057	0.87 (0.06–1.69)	0.036*	0.82 (0.00–1.65)	0.049*
Technocrat	−0.24 (−0.87–0.39)	0.453	−0.15 (−0.56–0.27)	0.490	−0.15 (−0.57–0.28)	0.500
Political power	7.64 (5.22–10.06)	<0.001***	0.50 (−0.27–1.26)	0.205	0.64 (−0.22–1.50)	0.145
Govt. left-right position	−0.46 (−1.08–0.15)	0.141	−0.05 (−0.44–0.34)	0.804	0.00 (−0.40–0.41)	0.991
Polarisation (public)	−2.97 (−4.56 – – 1.38)	<0.001***	0.12 (−0.55–0.80)	0.716	0.29 (−0.52–1.10)	0.483
Mobilisation (parliament)	5.56 (2.49–8.63)	<0.001***	0.49 (−0.43–1.40)	0.297	0.15 (−0.95–1.25)	0.791
Election in period	−0.54 (−1.15–0.07)	0.084	−0.24 (−0.62–0.14)	0.212	−0.29 (−0.67–0.10)	0.144
Govt. deficit	−0.09 (−1.07–0.88)	0.851	−0.43 (−0.90–0.05)	0.077	−0.68 (−1.24 – – 0.13)	0.016**
Govt. debt	4.62 (1.89–7.36)	0.001	−0.04 (−0.77–0.68)	0.910	−0.33 (−1.54–0.88)	0.592
Unemployment rate	−0.64 (−2.51–1.24)	0.505	0.26 (−0.37–0.90)	0.415	0.64 (−0.18–1.46)	0.128
Polarisation * Mobilisation	−2.03 (−4.22–0.16)	0.069	−0.38 (−1.33–0.57)	0.431	−0.26 (−1.27–0.76)	0.619
Centralisation					−0.98 (−2.33–0.38)	0.157
CBI					−0.79 (−1.71–0.13)	0.094
Constraints					−0.12 (−1.14–0.91)	0.822
Random effects
σ^2	10.30		4.75		4.74
τ00	17.63 MS		0.18 MS		0.26 MS
	0.96 emuc_cat2		0.73 emuc_cat2		0.72 emuc_cat2
ICC	0.64		0.16		0.17
N	11 MS		10 MS		10 MS
	6 emuc_cat2		6 emuc_cat2		6 emuc_cat2
Observations	685		487		487
Marginal R^2/Conditional R^2	0.278/0.743		0.048/0.202		0.058/0.219

note: ***p < 0.001; **p < 0.01; *p < 0.05.

The first striking finding concerns the congruence between leaders’ communication and negotiators’ positions: across the board, we find a small but significant positive relationship between the saliency of leaders’ positions and negotiators’ positions being supportive – indicating congruence between positive communication and supportive positions. We do not find a significant negative relationship between sentiment and negotiators’ opposition; as Table 1 indicates, in only 54.3 per cent of cases when negotiators are opposed do leaders also communicate opposition.

Turning to our main hypotheses, we find no evidence that any of our predictors related to institutional constraints on leaders’ ability to ensure congruence between communication and positions matter. Model A3 indicates Centralisation, CBI and Constraints do not significantly predict leaders’ positions when controlling for negotiators positions; these results hold when including interaction terms between negotiators’ positions and these covariates, as well as when predicting Position strength (Model B3). We thus find no evidence to support hypotheses 1a, 1b or 1c.

Our politicisation-related predictors do produce strong effects in model A1: a one unit increase in polarisation, corresponding to a change from a relatively consensual public opinion on the EU in the leaders’ home country to a relatively polarised one, yields a 2.97 decrease in the saliency score of leaders’ position (P = 0.000). A similar change in the degree of mobilisation of Eurosceptic sentiment, however, produces an increase in the saliency of leaders’ position of almost twice as strong (5.56, P = 0.000). However, the near-significant (P = 0.069) interaction effect between polarisation and mobilisation indicates that the strength of the positive effect of mobilisation decreases substantially as polarisation increases (Figure 5). These effects disappear when EU leaders are removed from the sample (A2), but further analysis provides no evidence that these effects are conditional on whether leaders represent member states or the EU, or the position taken by negotiators. Hence, while we find support for H1a and H1c, we cannot establish with certainty whether these effects are driven by EU leaders or the increased sample size. These findings hold also in our analysis of position strength (see Appendix).

Figure 5. Conditional effect of polarisation and mobilisation.

We furthermore find no evidence that technocratic leaders differ from elected leaders when predicting saliency of position or position strength. However, model A1 suggests leaders are slightly less likely to communicate supportively when facing elections in their home country (P = 0.084); this effect becomes significant in models predicting position strength that exclude EU leaders (B2 and B3). Again, however, further inclusion of interaction effects between technocrat, election and negotiators’ positions yield no results. Hence, these results yield credibility to H2f, but not to H2d or H2e.

Finally, in terms of our control variables, two things stand out. First, we consistently find that leaders from more politically powerful member states tend to communicate more strongly about European integration (B1-3). Model A1 suggests this is in part driven by supportive communication by these leaders, suggesting leaders from more powerful countries feel a stronger need to defend European integration during crises. Second, in the full sample, we find that leaders from more indebted countries tend to communicate more supportively (A1) and strongly (B1) about integration during the Eurozone Crisis. In the smaller sample of national leaders, the unemployment rate has a similar effect (cf. Traber et al., 2020). These effects may be explained by the fact that governments of creditor states were, on average, far more supportive of joint European solutions to the crisis (Iversen et al., 2016).

Conclusions

The aim of this article is to study to what extent the communicated positions of European and national leaders’ reflect the positions forwarded by their countries’ or institutions’ negotiators during the Eurozone crisis. Whereas little is known about this, the question is vital from a democratic legitimacy perspective: leaders’ public communication about the EU level is a key information resource for EU citizens. Both from a political-strategic and public administrative perspective, there are several reasons, however, to suspect that the communication by leaders may be incongruent with negotiation positions.

Combining three existing databases, we find congruence between communication and positioning in roughly half of the observations. More precisely, in the case of EU leaders we find full alignment in 48.2 per cent of the cases, while the professed positions of the national leaders corresponds to what happens in EU level negotiations in 53.6 per cent of the time. From the perspective of democratic accountability, this may seem like a less than ideal outcome. However, if we take the literature on national campaign pledges as a benchmark, this is only seven to twelve per cent lower than the average fulfilment of 60 per cent of campaign promises by national governments. Taking into account the complexity of EU multilevel governance, this may actually be a rather decent score. In addition to this general pattern, we find that in case of incongruence, EU level leaders are in general more in favour further Europeanisation than the negotiators while national leaders are equally more positive as negative. This aligns with recent findings in the literature suggesting that elected leaders are more inclined to listen to Eurosceptic domestic voices while EU leaders will defend further Europeanisation in the face of increasing Euroscepsis (Rauh et al., 2019).

Both the level and direction of incongruence between leaders’ communication and negotiators’ positions differs across time, member state and the issues featured in the negotiations. Interestingly, our findings show that throughout the crisis the agreement between leaders and negotiators increases, in particular in the case of national leaders. Moreover, when we compare the congruence between leaders professed positions and the EU level negotiators per country, Denmark stands out for its remarkably large correspondence in positions. As Denmark is known for its strong coordination of EU policy making, both findings suggest support for the public administration perspective. When reviewing the nature of the incongruence between negotiators and leaders per county, we find that in Eurosceptic countries like the UK, the Netherlands and Hungary, leaders are far less supportive of pro-European measures than their negotiators. This in turn would lend support for the politisation hypothesis.

Our explanatory statistical analyses, however, provide no evidence of the validity of public-administrative explanation. Instead, they provide some evidence in favour of the alternative argument that suggests leaders’ communication is driven by electoral and strategic considerations: leaders facing more Eurosceptic parties in their home country are significantly more likely to advocate less Europeanised measures than the positions of the negotiators. On the other hand, we unexpectedly find weak support for the claim that leaders facing an election in their respective member state are more likely to support more integrative measures in their communication than the negotiators. Yet here, too, strong conclusions about congruence cannot be drawn. None of the other politicisation-related variables, nor the interaction effect, appear to offer much predictive capacity.

Our findings are largely congruent with a strategic and politicisation-driven account of leaders’ communication and correspond with recent advances in the literature on strategic communication and the management of politicisation-induced ‘bottom-up pressures’ by EU leaders (Bressanelli et al., 2020; Rauh et al., 2019; Traber et al., 2020): politicisation-induced pressures push leaders to communicate strategically to domestic audiences, yet their responses depend on whether such pressures are perceived to enable or constrain certain courses of action. They also speak to research on position-taking in the Council, which suggests the ability and willingness of voters to sanction leaders’ drives leaders’ signalling of responsiveness to these audiences (Schneider, 2018, 2020). Our major contribution to these literatures is their linkage: our unique data offers the ability to compare position-taking with communication, which advances our understanding of both.

Our results remain subject to some limitations. Foremost, we observe (in)congruence between communication and positioning in a rather unique context: our observations all concern Economic and Monetary Union, and stem from a period of severe crisis. Hence, our findings pertain directly to politics in the area of ‘core state powers’ (Genschel & Jachtenfuchs 2018), and during crisis. Both the policy area and occurrence of crisis are factors that greatly increase the stakes for leaders. The congruence between leaders’ communication and positions of negotiators may therefore be greater and more determined by public-administrative factors in other issue-area’s or everyday policy-making. Secondly, the causality between societal politicisation and elite behaviour is inevitably complex and endogenous: politicisation shapes leaders’ communication and position-taking, which in turn cues politicisation. This implies that such complexity must be matched by careful interpretations based on adequate research design. Our careful temporal matching of leaders’ communication to negotiators’ positions and use of cross-classified multilevel models controlling for temporal clustering are standard practice in the politicisation literature (see e.g., Van der Veer & Haverland, 2018), and the congruence between our findings and other contributions to that literature suggest leaders indeed respond to politicisation-induced pressures (Bressanelli et al., 2020; Rauh et al., 2019).

Our findings should also be read as an urgent call for more research into the communicative practices of leaders in the EU context, while linking these clearly to actual behaviour and decision-making (Van Esch & Snellens, 2024). We need to know to what extent these findings hold when including leaders from other member states, and in other issue areas, including non-crisis decision making. In that sense, the Eurozone crisis and subsequent refugee crisis may still be unique events in that they saw both an unprecedented need for cooperation between EU elites and an unprecedented public opposition to this cooperation among the elites themselves and the EU public (Van der Veer & Haverland, 2018); the EU’s more determined handling of the COVID pandemic and the war in Ukraine appears to have at least appeased a larger part of the EU citizenry. This further underlines the need for more research into when and why leaders talk the talk, but do not walk the walk.

Acknowledgement

We gratefully thank the reviewers for their very helpful comments. We acknowledge financial support from the European Union's Horizon 2020 research and innovation programme under grant agreement No 649484 for the data-collection for the majority of the cognitive maps, as well as the EU Politics, Administration and Law (EUROPAL) hotspot of the Institute for Management Research of Radboud University Nijmegen. The cognitive maps of the Greek leaders were used with the permission of Annelou Veen (Veen, 2017).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by EUROPAL and European Union Horizon2020 research and innovation program [grant number 649484].

Notes on contributors

Femke van Esch

Femke van Esch is a professor of European Governance and Leadership of the European Union at the Utrecht University School of Governance

Reinout van der Veer

Reinout van der Veer is an assistant professor of international relations at the Department of Political Science at the Radboud University Nijmegen

Notes

1 Note that our argument here primarily concerns communication about negotiation positions to general publics. Wasserfallen et al. (2019, p. 12) argue that the information-rich environment of EU negotiations, combined with the strong institutionalised setting and repeated character of negotiations, actually dissuade leaders to strategically misrepresent their ‘true’ negotiation positions vis-à-vis other negotiating parties.

2 Existing studies indicate that both the type of political system and ideological position of a leader may influence their propensity to deviate from reality in their public speeches: Winner take all systems induce strategic communication and right wing citizens are more acceptant of politicians lying (De Keersmaecker & Roets, 2019). Unfortunately, our data does not have sufficient variation to test these expectations.

3 It must be noted that Schneider’s (2018; 2020) work on public commitments as signals of responsiveness actually suggests leaders are less likely to deviate from initial negotiation positions when elections are near; this reluctance towards compromise constitutes the ‘public commitment’ to responsiveness. However, the EMU choices data does not account for shifts in negotiation positions over time. Schneider does not examine how these leaders communicate about those positions to domestic audiences. Our argument here is that such communication is likely to deviate from actual positioning when elections are near, as such communicative signaling is relatively low in cost for leaders, while publics have few alternative sources of information regarding actual positioning.

4 See Appendix 1 for how the cognitive maps used for this study were compiled.

5 See Appendix 2 for the speeches that were used to derive the cognitive maps from as well as the list of leaders included.

6 For the intercoder reliability for the different coding steps in this project, see Appendix 1.

7 For a detailed description of this, see Appendix 1.

8 Support for the position was coded as 1, opposition as −1.

9 See Appendix 1 for more detail on the operationalisation and the descriptive statistics for of these variables concerning the subsample of national leaders prior to centring and standardisation.

10 Period 1 runs from after the announcement of the excessive Greek deficits in October 2009 to the first EU measures taken in March 2010. The second period runs until Draghi’s ‘whatever it takes’-speech. The third until the election of the Greek Syriza government. The fourth period ends with the Greek bail-out referendum.

11 Although the cognitive mapping data is longitudinal and communication of leaders in period 0 has per definition occurred prior to negotiations, the EMU choices data does not include temporal indicators for negotiation positions. We can therefore not fully determine the sequencing of leaders’ communication and the position of negotiators for all issues in our dataset. However, given that the political salience of issues tends to be highest during negotiations and that leaders’ communication can both be intended as legitimation of negotiation positions for domestic audiences as well as position signaling to negotiation partners, we believe it is credible that most of our observed communication stems from after the establishment of initial negotiation positions as captured by the EMU Choices data.

12 For a similar procedure see (Lehner & Wasserfallen, 2019).

13 As indicated in Appendix 1, the intercoder reliability of four to five issues should be considered unreliable by common standards. To check the robustness of our findings, we reran the analysis omitting these issues. Comparison to the original analysis reveals that for both the descriptive and explanatory analysis the results hardly change and none of the key theoretical factors are affected (see Appendix 1, pages 16–17). This signals that our original results are robust.

14 For the Financial Transaction Tax, we only had one observation, so we omitted it from any further analysis.

15 This may be due to the fact that in this period we only have data on four cases

16 We do not use leaders themselves as a nesting variable because this results in over-specification of our models. Similarly, we do not use periods due to the high overlap between period and issues in our dataset.

17 No modelling assumptions were violated by these models; for details, see Appendix 1.

18 For all of the figures below, the bars indicate the %, the absolute cases per category sometimes vary quite significantly (see appendix).