Explaining the politicization of EU trade agreement negotiations over the past 30 years

Abstract

Although the Transatlantic Trade and Investment Partnership with the USA and the Comprehensive Economic and Trade Agreement with Canada have elicited considerable domestic contestation in Europe, several other agreements have been negotiated into public and media indifference. What explains this difference? In this article, I put forward a number of arguments on the structural causes of the politicization of European Union (EU) trade policy over the past 30 years and test them against a newly collected dataset covering 19 preferential trade agreements. The qualitative comparative analysis suggests that the politicization of EU trade negotiations is determined by the co-occurrence of several, well-defined conditions. More specifically, it tells us that: (1) the Lisbon Treaty's reform of EU trade policymaking is the main driver of politicization, (2) the level of public support for the EU is of particular relevance when it comes to ‘deep and comprehensive’ agreements that touch on sensitive domestic issues, and that (3) high adjustment costs expected from trade liberalization can lead to the politicization of trade negotiations.

Introduction

In September 2016, thousands of people marched through Brussels demanding the European Union (EU) to drop the planned Comprehensive Economic and Trade Agreement (CETA) with Canada and the Transatlantic Trade and Investment Partnership (TTIP) with the USA. Shortly after their launch, the transatlantic trade negotiations became highly contentious in Europe, sparking massive domestic contestation by large segments of civil society (Hübner et al., 2017; Eliasson and Garcia-Duran Huet, 2018a; Duina, 2019). These episodes have led many observers to depict EU trade policy as an increasingly politicized policy domain (Laursen and Roederer-Rynning, 2017; Young, 2017a; Costa, 2018). Yet a closer examination reveals that the degree of politicization of EU trade negotiations varies considerably both over time and across contemporary deals (Leblond and Viju-Miljusevic, 2019; Meunier and Czesana, 2019; Young, 2019). While TTIP and CETA have generated significant domestic political turmoil, most of the EU's recently negotiated agreements have indeed taken place in the indifference of European citizens and the media.

What explains these differences?¹ As politicization has significant implications for the EU's ability to successfully conduct trade agreements, tackling this question is of utmost importance (Dür et al., 2019). To this end, I put forward a number of arguments on the causes of the politicization of EU trade policy over the past 30 years, grounded in the institutional, regulatory, and economic environment of trade negotiations. In particular, while extant literature has primarily devoted its attention to agent-centered explanations, investigating the strategies employed by political and societal actors to (de)mobilize European public opinion, this work looks at the ‘enabling’ conditions that create opportunities for politicization in the first place. The arguments advanced are tested against a newly collected dataset covering 19 bilateral and interregional trade agreements with 29 countries, negotiated or under negotiation by the EU since the early 1990s.² Based on large-scale media analysis, I create an original measure of the degree of politicization of preferential trade agreements (PTAs), which I use as the outcome variable in a qualitative comparative analysis (QCA). The analysis suggests that the politicization of trade negotiations is reached through multiple causal paths given by the co-occurrence of several conditions. Specifically, the findings indicate that (1) the Lisbon Treaty's reform of EU trade policymaking is a major driver of politicization, (2) the level of public support for the EU is especially relevant when it comes to ‘deep and comprehensive’ agreements touching on sensitive domestic issues, and (3) trade negotiations with economically significant partners can lead to politicization if the adjustment costs are expected to be substantial.

The remainder of this article is structured as follows. First, I review the existing literature on the politicization of EU trade policy and its causes. Next, I elaborate on the explanatory conditions investigated in this study and provide a theoretical justification for their inclusion in the analysis. I then introduce the methods and data used and conclude with a discussion of the findings and their implications.

The politicization of EU trade policy

In the past, political conflicts around trade negotiations were confined to clashes between groups representing opposing economic interests, while the broader public was rarely involved (Dür, 2008). Yet this ‘permissive consensus’ on trade liberalization began to falter in the first decade of the 21st century (Meunier and Czesana, 2019). The negotiations of a series of new PTAs generated a significant level of political turmoil and mobilized a wide range of non-economic actors, leading many scholars to speak of the politicization of EU trade policy.

Politicization in general terms refers to a situation in which a policy domain, previously subject to elite-driven policymaking, becomes the subject of public debate and controversy (Zimmermann, 2019). Although different understandings exist of what exactly this process entails, there is growing agreement on its three-dimensional nature (De Wilde et al., 2016). The first dimension is salience, which refers to the emphasis attached to a certain issue in the public sphere. The second is actor expansion, which gauges the number and variety of societal actors involved in the debate. The last is polarization, which captures the intensity of political conflict over the issue. Based on this definition, politicization qualifies as a stand-alone phenomenon that differs from other related concepts in the political conflict literature, most notably contestation (see Hackenesch et al., 2021). Although the latter may see the participation of a limited number of actors, such as in the case of intergovernmental bargaining and technocratic debates taking place behind closed doors, the former involves making the issue at hand a matter of broad public resonance. Indeed, while contestation can certainly lead to politicization, as long as opposition remains confined to a restricted set of stakeholders it is not enough to speak of a politicized issue.

Scholars researching this phenomenon have devoted their attention primarily to the ‘black box’ of politicization and looked at the strategies used to mobilize European public opinion against or in favor of trade negotiations (Chan and Crawford, 2017; Bouza and Oleart, 2018; Meunier and Roederer-Rynning, 2020). Dür and Mateo (2014), for instance, explore the underlying reasons for the success of the anti-ACTA (Anti-Counterfeiting Trade Agreement) campaign. They show how interest groups managed to increase the public salience of the issue, and how the subsequent public opposition made decision makers opt against ratifying the agreement. Gheyle and De Ville (2019), Eliasson and Garcia-Duran (2018a), and Buonanno (2017) analyze how an initially small group of European civil society organizations (CSOs), by engaging in a considerable amount of external lobbying, managed to expand the scope of conflict and turn the TTIP negotiation into a publically and politically salient topic. Along the same line, Gheyle (2020) investigates coalition formation dynamics in some EU member states that experienced different degrees of TTIP contentiousness, trying to determine which factors explain the success of domestic coalition formation initiatives and the resulting politicization. Siles-Brügge and Strange (2020) and Siles-Brügge (2018) examine narratives, myths, and framing strategies employed by opposition campaigners to construct transnational networks of activists and bring CETA, TTIP, and the General Agreement on Trade in Services negotiations to public attention. In turn, Eliasson and Garcia-Duran (2018b) and Garcia-Duran et al. (2020) focus on the other side of the public debate, looking at the rhetorical strategies employed by political elites to counter the arguments of PTA opponents.

These studies, despite variations in detail, share a strong preference for agent-centered explanations. They conceive of politicization as the product of deliberate actions on the part of organized actors to advance their preferred policy goals by influencing public opinion. Accordingly, and with few exceptions, they came with a major limitation: they tend to focus almost exclusively on cases where politicization has taken place – that is, TTIP, CETA, and ACTA. As Leblond and Viju-Miljusevic (2019) point out, however, ‘it is not sufficient to analyze the contested nature of EU trade politics in particular trade agreement negotiations’ (p. 1841). Drawing inferences by selecting cases on the dependent variable may lead to overlooking some determinants of politicization and overestimating others. Furthermore, focusing on such a limited set of cases limits the generalizability of the findings (Geddes, 2007). By contrast, to determine the reasons for the variation in the level of politicization, it is necessary to analyze the occurrences of both the presence and absence of this phenomenon, so that we can identify which factors correlate with the different outcomes. The work of De Bièvre and Poletti (2020) represents a first attempt to address this issue by examining how combinations of several structural factors may lead to politicization. Yet they do not empirically test this proposition. Here, I take up and expand on their arguments through a comparative analysis of those ‘enabling’ conditions that, by shaping the context in which societal and political actors unfold their agency, create (or deny) opportunities for the politicization of trade negotiations in the first place.

Five causes of politicization

In the section below, I introduce the explanatory conditions and provide a theoretical justification for their inclusion in the analysis. In line with previous studies, I acknowledge that hardly a single factor suffices in itself to produce politicization (Meunier and Czesana, 2019; De Bièvre and Poletti, 2020). Rather, I argue that they unfold their full explanatory power when in conjunction with others, thereby acting as reinforcing rather than rival explanations.

Regulatory depth

The first condition focuses on the content of trade agreements. In particular, it associates the increased contentiousness of trade negotiations with the recent regulatory turn in the EU trade agenda (De Ville and Siles-Brügge, 2016; Young, 2016, 2017a; Duina, 2019).

Since the mid-2000s, the EU has been negotiating a new generation of ‘deep and comprehensive’ PTAs which, compared to older ones, cover a wider range of issues and aim not just at eliminating at-the-border trade barriers but rather at integrating markets. As these new agreements reach far into the domestic sphere, their negotiation is likely to result in a clash between the parties' regulatory models, with one side eventually prevailing (Büthe and Mattli, 2011). The consequent prospect of regulatory downgrading may be seen as a threat by citizens and civic interest groups based in the EU, which usually have higher standards than their counterparts. Also, deep trade integration is increasingly perceived as an impediment to member states' capacity to regulate domestic sensitive issues – for example, environmental protection, public health, and safety – fueling worries about the legitimacy of the EU trade policymaking (Buonanno, 2017; Siles-Brügge, 2018). Concerns about the loss of domestic policy autonomy are indeed a persistent issue in many anti-globalization campaigns in Europe. Under this new reality, latent public support for trade liberalization can no longer be assumed; conversely, citizen groups and CSOs are expected to mobilize more and more (Young, 2017b).

This argument seems well suited to explain the increase in the degree of politicization of EU trade policy over the last few decades. Its logic is compelling: the deeper and more comprehensive the commitments in the new generation of PTAs have become, the more politicization has occurred. Moreover, as among recently negotiated agreements some involve deeper regulatory commitments than others, the same logic may also account for variations between contemporary trade negotiations. Nonetheless, this explanation clashes with some empirical evidence. For one, trade deals containing regulatory provisions as deep and comprehensive as those of TTIP and CETA – for example, EU–Japan agreement (Suzuki, 2017) – went largely unnoticed. This suggests that although deep trade agreements are more likely to become politicized, depth alone is not enough (De Bièvre and Poletti, 2020).

Relative economic size

The second condition concerns the economic importance of the EU negotiating partners. The literature has long acknowledged economic size as a major factor influencing trade negotiations (Baier and Bergstrand, 2004; Bergstrand et al., 2016), stressing how the allocation of economic power affects the EU's chances of securing its ideal outcome (Young and Peterson, 2014). The size of the single market usually grants the Union considerable leverage over its trade partners (Damro, 2012) and allows it to impose its regulatory standards (Conceiçao-Heldt, 2014). When dealing with countries of comparable economic importance, however, the EU is forced to be more flexible than in negotiations with countries of lesser economic clout (Akhtar and Jones, 2014). In this respect, it tends to adjust its expectations and negotiating strategy to the bargaining power of its counterparts. While in cases of favorable power asymmetry, the EU pursues regulatory coordination through convergence toward its own standards, when dealing with partners of similar economic size, it usually follows an approach based on mutual recognition (Young, 2015). This means that the higher the economic power of the counterpart, the higher the risk that the EU will make concessions on sensitive domestic regulatory issues. This in turn incentivizes CSOs and citizen groups to mobilize, thereby leading to politicization.

This explanation is surely consistent with the path observed during the TTIP negotiations. Opposition to the agreement was largely driven by fears that EU standards would be lowered by having to adapt to American ones. Most of the anti-TTIP groups perceived the USA as a negotiating partner more powerful than the EU and thus able to impose its preferences and demands (Eliasson and Garcia-Duran, 2016). By the same logic, one would expect to observe similar patterns of politicization with other major trading partners. Yet the negotiation of the Comprehensive Agreement on Investment with China, the only country other than the USA to rival the EU in economic size, elicited little or no public mobilization. The same goes for the negotiations with Japan, another leading player in the global trading system. Empirical evidence therefore suggests that partners' economic size should not be seen as a stand-alone reason for politicization. Rather, a wiser approach would be to consider its effect in conjunction with other factors, such as the scope and depth of agreements (see Young, 2016).

Public attitudes

The third condition concerns public attitudes toward the EU. Recent studies have shown that international politics is an unfamiliar topic for the general public (Dellmuth, 2016) and that only a few citizens have in-depth knowledge of international trade issues (Rho and Tomz, 2017). This causes people to rely on informative shortcuts to form opinions on these matters (Bechtel et al., 2014). For instance, research has found that political cueing and framing by political elites are major drivers in citizens' orientations toward PTAs (Hicks et al., 2014). Heuristic processing has been observed in EU-negotiated trade agreements as well (Dür and Schlipphak, 2021). In this regard, Steiner (2018) advanced the idea of a ‘treaty partner heuristic,’ arguing that the pattern of support (or contestation) for TTIP was largely shaped by individuals' views of the two treaty partners. Scholars have indeed documented a strong association between public predispositions toward the potential contracting parties and opposition to specific trade deals (Jungherr et al., 2018). Data on EU citizens' attitudes toward third countries, though, are limited to a few cases and cover only a few years, making them unsuitable for medium-scale comparative analysis. I therefore focus on the other type of treaty partner heuristics: public perceptions of the EU.

On the European side, PTAs are initiated and negotiated by the EU institutions. Because the public largely perceives them as an EU initiative, those who distrust the Union are usually skeptical of trade deals as well (Steiner, 2018). A skeptical public, for its part, represents a fertile breeding ground for interest groups seeking to mobilize opposition against the agreements. In a scenario of diffuse disaffection and distrust of EU institutions, political entrepreneurs are therefore expected to be more effective in their role as agents of politicization. This is in line with the argument of Zürn et al. (2012) that the exercise of political authority by legitimacy-deficient international institutions is at risk of encountering high levels of opposition from civil society.

From a cursory glance, this explanation seems to offer a plausible reason for the recent politicization of EU trade policy. CETA and TTIP, the trade negotiations that aroused the fiercest public opposition, were launched in the aftermath of the 2008 financial crisis when the decline in public trust in the EU was most pronounced (Armingeon and Ceka, 2014). Still, by itself, this explanation cannot account for the whole picture. First, it is inconsistent with the fact that many of the PTAs negotiated during that period did not undergo the same degree of politicization as those with Canada or the USA. Moreover, trends in public attitudes toward the EU hardly coincide with the politicization of its trade policy. Although the first substantial downturn in citizens' support occurred as early as the second half of the 1990s, the trade agreements negotiated at that time prompted little to no response from the European public.

Import competition

The fourth condition addresses the political implications of the adjustment costs arising from trade agreements. Recent years have seen growing public concern over the distributional effects of trade liberalization on certain sectors of society. These concerns, in turn, are regarded as a major driver of the resurgence of economic nationalism in Europe and elsewhere (Colantone and Stanig, 2018; Young, 2019). Significantly, such dynamics concern specific trade events as well. The prospect of increased import competition may lead potential ‘losers’ to oppose certain agreements. Then, to the extent that these groups successfully engage in actions to instigate public opposition against such agreements, politicization may ensue (De Bièvre and Poletti, 2020). There is ample evidence that by using external lobbying strategies, interest groups can affect both the salience of trade issues to the public (Dür and Mateo, 2014) and the position people take on such issues (Siles-Brügge, 2018).

In this regard, the magnitude of adjustment costs depends primarily on the type of change in countries' trade patterns (Brülhart and Elliott, 2002). If the trade expansion occurs intra-industry, then the adjustment process can be expected to be cheaper compared to a mainly inter-industry expansion. This is because labor can more easily mobilize to the same industry than to a different industry. Extensive literature in fact suggests that trade liberalization among countries with sufficiently similar factor endowments is associated with relatively lower adjustment costs (Hamilton and Kniest, 1991; Menon and Dixon, 1997). Following that logic, it is safe to assume that the more economically similar the partners are, the less politicized the trade negotiations are likely to be.

Still, looking only at countries' economic structures, one can hardly find a pattern. The EU's most politicized PTAs include both developed countries such as the USA and Canada and lower-cost countries such as Mercosur members. The same is true for non-politicized trade agreements, negotiated both with advanced economies such as Israel, Australia, and Singapore and with less developed Mediterranean and South American countries.

Institutional setup

The last explanatory condition regards the enhanced role of the European Parliament (EP) in EU trade policymaking following the Treaty of Lisbon (ToL). The Common Commercial Policy was traditionally considered a domain reservé to executive actors, with the Council issuing the mandate and the Commission carrying out the negotiations on behalf of the member states (Woolcock, 2000). The Treaty partially reallocated decision-making power in favor of the Parliament, making this institution a de facto veto player. According to the ToL rules, the EP must give its consent to all trade agreements negotiated by the EU and must be fully informed at all stages of negotiations. As for EU trade legislation and its implementation, the EP now stands on an equal footing with the Council under the ordinary legislative procedure (Van den Putte et al., 2014). In addition, since the ToL entry into force, the EP has been able to widely expand its informal role, exerting its power and scrutiny well beyond the provisions as laid out in the treaties (Ripoll Servent, 2014; Héritier et al., 2019).

The emergence of the EP as a key player has considerably increased the potential for contestation of trade negotiations. As the only democratically elected body, the EP is the most direct path for the public to influence EU trade policymaking. As such, it has become ‘an open forum for debate and lobbying on economic topics, with civil society non-governmental organizations (NGOs) being particularly active’ (Woolcock, 2017: 169). Its empowerment enabled many new actors to gain access to the decision-making process and provided them with a new venue to make their voices heard (Zimmermann, 2019). This encouraged the emergence of the kind of civil society structures that make politicization more likely (Costa, 2018). A strong EP also provides diffuse interest groups with resources – access to expertise, development of synergies, and economies of scale – that can mitigate their collective action problems and reduce mobilization costs (Roederer-Rynning, 2017). Moreover, this institution does not merely act as an enabler of social mobilization; it contributes proactively to politicization by using public pressure as leverage to assert itself against other actors, namely the Council (Greenwood and Roederer-Rynning, 2019). Overall, the EP is assumed to act as the fulcrum of societal groups and organized interests' mobilization, making a major contribution to the politicization of EU trade policy in general, and trade negotiations in particular (Roederer-Rynning, 2017).

The ToL argument is certainly consistent with the fact that most EU trade negotiations displaying a high degree of politicization took place in the post-Lisbon era. However, although the institutional changes introduced by the Treaty may explain variation over time, they alone cannot account for differences across contemporaneous negotiations: the ToL made all PTAs more susceptible to politicization. The fact that not all post-Lisbon trade negotiations were characterized by high levels of politicization suggests that institutional opportunities must be considered in conjunction with other conditions.

Research design and data collection

Case selection

The dataset consists of 19 trade agreements negotiated or being negotiated by the EU over a period ranging from the early 1990s to the present (2023). The selection of cases was primarily driven by data availability, in particular as regards the negotiating directives used to assess the ‘regulatory depth’ condition. Directives are not systematically made public by the Commission. Because of the sensitive nature of the documents, the EU authorities will normally restrict public access to them even after the formal conclusion of negotiations. For this reason, it was only possible to obtain 19 of them. This admittedly resulted in a limited number of trade agreements being analyzed.

Although this could potentially result in selection bias, there is no reason to believe that the distribution of missing data reflects a specific categorization or criterion that could distort the results and threaten their validity. The cases show a high degree of variation in both the outcome and the conditions investigated. For instance, they cover several generations of PTAs concluded over 30 years of EU trade policy, with agreements negotiated both during the so-called ‘permissive consensus' period and afterward. Also, the sample comprises trading partners that differ in terms of both geographical location and economic power, for example, advanced economies such as Japan and the USA and developing countries in Southeast Asia and Central America. The same is true for missing cases. Undisclosed directives include both pre- and post-Lisbon agreements, as well as major economies like China and South Korea and developing countries from different continents, such as Mexico, Kazakhstan, and eastern European countries.

Equally important is that the analysis includes all trade negotiations deemed by the extant literature as the most politicized, and thus the most significant for this work. So, despite the limited number of cases analyzed, the composition of the sample permits the comparison of multiple configurations of conditions and the identification of difference makers between the two groups of agreements, namely those displaying the outcome and those that do not.

The outcome: politicization

I estimated the degree of politicization of the trade negotiations by analyzing the respective public debates as unfolded in the European media. As the time covered extends over 30 years, to hold the analytical context constant and keep the research efforts feasible, I focused on five major member states whose membership in the EU dates prior to the early 1990s: Italy, France, Spain, Germany, and the UK. The media analysis relies on one quality newspaper per country, namely La Stampa (Italy), Le Figarò (France), El Paìs (Spain), Süddeutsche Zeitung (Germany), and Financial Times (UK). To account for possible variations across countries, national outlets were further supplemented with two major European media, EurActiv and Agence Europe. I selected publications with relatively high levels of readership and which are likely to feature both EU actors and trade-related news in their coverage. Also, quality newspapers are generally regarded as opinion leaders who may influence other media content (Gattermann and Vasilopoulou, 2015).

The relevant media coverage was collected manually through keyword searches in the LexisNexis and Factiva news archives. This resulted in a sample of nearly 3000 articles. Next, I identified all relevant statements made by the stakeholders involved in the public debate and coded them according to their relevance to the empirical cases and the positions adopted with respect to the agreements. The technique I used is largely based on the method of political claim analysis (Koopmans and Statham, 1999), in which the unit of study consists of an actor making a particular claim rather than a whole article or core sentence (see De Bruycker and Beyers, 2015; Andrione-Moylan et al., 2021).

The elements constituting politicization have been operationalized as follows (see De Bruycker, 2019). Salience was measured based on the number of articles that covered each trade negotiation in the selected media outlets. Actor expansion was operationalized as the number of non-governmental actors – that is, non-governmental organizations (NGOs), social movements, business organizations, citizen action groups – participating in the public debate. Polarization was measured through an ordinal dispersion index computed from the public statements of political elites, which indicates whether the actors debating a given agreement share a similar position or whether views conflict. In a final step, I combined the three measures in Hutter and Grande's (2014) politicization index:

$$politicization = salience\;\times \;( actor \;expansion + polarization) .$$

The conditions

Regulatory depth

Existing studies mostly assess the depth and scope of PTAs based on the texts and annexes of the agreements (see for instance Dür et al., 2014; Hofmann et al., 2019). Adopting the same approach here could be problematic, as the content of an agreement – what policy areas it covers – is endogenous to the level of politicization. Pressure coming from public opinion and interest groups can indeed affect the outcome of the negotiation process. The case of CETA is emblematic. Although the parties had already closed the negotiations, the mounting public opposition forced them to re-draft the provisions on investor–state dispute settlement (ISDS) and replace the traditional mechanism with a new investment court system (Hübner et al., 2017).

To avoid endogeneity issues, I created an original measure of the ‘expected' depth of PTAs by coding the negotiating directives (see Table A4 in the Annex). Directives are issued by the Council before the start of negotiations and set out what the Commission is to achieve in the deal. Certainly, they do not convey the same amount of information as in the text of the agreement. My purpose, however, is not to assess the scope and depth of the provisions adopted but rather to obtain a valid and reliable measure of the regulatory commitments expected during the negotiation of the agreement.

One potential shortcoming of this approach is that directives are seldom made public before negotiations are concluded. Yet despite the secrecy surrounding them, they are still accessible to a wide range of actors, including Council representatives, Commission officials, and members of European and national parliaments. On several occasions, part of their content – or the full mandate, as in the case of TTIP and Mercosur – has filtered into the media and reached the public.³

Relative economic size

In international negotiations, the allocation of bargaining power between the parties hinges on several factors (Drahos, 2003). In this paper, the focus is on the economic dimensions of the actors involved. The literature largely regards the economic size of a country as one of the most relevant sources of power in trade negotiations (Elsig, 2016; Soo, 2017). Countries with large domestic markets that others want access to can make credible threats and promises. Moreover, as I am interested in assessing the ‘perceived’ balance of power between the EU and its counterparts, it is sensible to focus on those factors that are easy for outsiders to observe – contrary to resources that depend on individual negotiators, such as strategies and skills, which are harder to grasp (Bailer, 2010). Because the notion of relative economic power is inherently comparative, that is, it compares the strength of one negotiator with that of others, it is measured as the ratio of the gross domestic product (GDP) of the negotiating partners to that of the EU. The data used to construct the variable are extracted and elaborated from the World Bank database. Given the limited variation in GDP figures over the period analyzed, the data concerns only the first year of each PTA's negotiations.

Public attitudes

Public support for the EU is derived from the Standard Eurobarometer country membership indicator. Operationalizing this variable using one-dimensional indicators is hardly optimal since previous studies have revealed that support for the EU is of a multi-dimensional nature (Braun and Tausendpfund, 2014). To my knowledge, however, it is the only available source of data for the period under review. Data on attitudes toward EU membership have in fact been consistently available in the Eurobarometer survey since the 1970s.

I recorded the variable into a binary in which 0 indicates membership is a ‘bad thing’ or ‘neutral’ and 1 indicates membership is a ‘good thing.’ ‘Do not know’ and ‘refusal’ were excluded from the analysis. Finally, I computed the value of the ‘EU attitudes’ condition for each PTA as the average value for the years covered by the respective negotiations, for the member states considered in the analysis. To account for differences between sample composition and actual population, as well as differences in size between European countries, raw data were adjusted by using the Eurobarometer post-stratification weighting procedure.

Import competition

(Dis)similarity in partners' economic structures is appraised through a trade complementary index (TCI) that measures the extent to which the export profile of one country matches the import profile of another country (for instance, Nanda, 2014). It is calculated as the sum of the absolute value of the difference between the import category shares and the export shares of the concerned countries, divided by two and converted to percentage form (see the Annex). The index ranges between 0 and 100, with zero indicating no overlap and 100 indicating a perfect match in the trade pattern. Here, the order has been reversed for better calibration. To calculate the TCI, UN Comtrade official statistics disaggregated at the two-digit level were used.

Institutional setup

Variations in the institutional framework are captured through a dichotomous variable, which codes with 1 those PTAs negotiated after the ToL came into force (December 2009), and with 0 those negotiated before it.⁴ I acknowledge that relying on a dichotomous measure entails potential limitations, especially when it comes to capturing the evolving role of the EP in EU trade policymaking. Scholars have indeed found that the Parliament's assumption of powers and competencies on trade matters occurred gradually, through an incremental process (see Héritier et al., 2015, 2019). Nonetheless, I consider the ToL to be a major watershed in the EP's control over EU trade policy and worth including in the analysis.

Calibration

QCA is a set-theoretic method designed to test for causal complexity by comparing cases according to their membership in sets. Central in operating the analysis is the calibration of sets, that is, establishing whether and to what extent individual cases belong to the outcome and conditions. The outcome variable – politicized EU trade negotiations – is calibrated using the direct method of calibration on the politicization index I computed. This method uses a software-based routine to transform numerical raw data into fuzzy sets – decimal fuzzy scores ranging from 0 to 1 – to reflect very fine-grained differences in degree. It involves a prior definition of three ‘empirical anchors’ that guide this transformation: a cross-over point (0.5), which allows for qualitative distinction in the case of membership; a full set membership (1); and a full non-membership (0). The anchors are chosen based on both conceptual knowledge (i.e. presence of all three constitutive dimensions of politicization) and substantive knowledge of empirical cases. A logistic function is then employed to fit the raw data in between these three thresholds. In the present case, the set membership anchors are established at levels 0.7 (fully out), 27.3 (fully in), and 12.5 (cross-over) of the politicization index. Based on calibration, TTIP, CETA, and agreements with Mercosur, Japan, and the UK are rated as politicized (>12.5).

The conditions variables are formulated in terms of crisp sets. The set membership thresholds are defined based on existing literature and substantive knowledge of the empirical cases. For the regulatory depth condition, cases with a value of 8 or lower on the depth index are calibrated to 0 (shallow PTAs) and above to 1 (deep PTAs). For the crisp set on relative economic size, cases with a coefficient of 0.095 or higher are calibrated as 1 (large economic size) while the others are calibrated as 0 (small economic size). The institutional context condition was already binary and did not need to be recalibrated. I simply attribute each PTA negotiated after the adoption of the ToL a set membership score of 1, and all other agreements a set membership score of 0. The point of indifference for the set import competition is in the middle of the TCI scale, at 50%, which means that cases crossing this threshold are in the set (1) and the others are out (0). Finally, since there were no conceptually sound justifications for defining qualitative thresholds on public attitudes toward the EU, I identified notable ‘gaps’ in the empirical distribution and used them as qualitative anchors (see Kuehn et al., 2017). Accordingly, only cases with a percentage of respondents in favor of EU membership equal to or above 56.5 are coded as 1 (positive attitude), while all others are coded as 0 (negative attitude) (see Figure A2 in the Annex). Table A5 in the Annex compares the calibrated data with the raw data (Table 1).

Table 1. Calibration of anchor points for the conditions and outcome

Results

In this section, I present the results of the QCA model. The analytical procedure consists of two different steps: first I check for the conditions that are necessary for the outcome and then perform the analysis of sufficiency.

Analysis of necessary conditions

The results of the necessity analysis are shown in Table 2. Only one condition, namely the institutional setup, meets the required threshold of consistency with a value of 0.92. This suggests that the institutional reform introduced by the ToL in 2009 was central in determining the politicization of EU trade agreement negotiations. Whenever the outcome occurs the institutional factor is also present. However, its levels of coverage and relevance of necessity (RoN) are below the corresponding threshold level, scoring 0.40 and 0.43 respectively. This indicates that the necessity relationship is of limited empirical relevance. The source of such trivialness may lie in the skewed distribution of the sample of cases analyzed, both in terms of outcome and condition of interest. Out of 19 PTAs covered by the QCA, only five are calibrated as politicized (26%) and only six were negotiated in the pre-Lisbon era (31%). This resulted in a large difference in size between the condition and outcome sets, with the former coming close to being a constant. This still indicates perfect set-theoretic consistency but, as in the current scenario, coverage will be low.

Table 2. Analysis of necessity for the occurrence of the outcome

Analysis of sufficient conditions

I now turn to identifying the (combinations of) conditions that are sufficient to explain the outcome. The first step in the analysis of sufficiency consists of producing a truth table, which summarizes all logically possible combinations among the conditions specified, also called types. Each row identifies a specific configuration of conditions and the cases that are covered by that configuration, with the number of rows matching the overall number of possible combinations. Based on empirical information, each row is either considered sufficient or not sufficient for the outcome. Table 3 reproduces an abbreviated truth table, where all types for which there are no empirical cases have been removed.

Table 3. Truth table for politicization

The next step consists of summarizing the information by logically minimizing the truth table through a Boolean approach. I present here the intermediate solution, while the other solutions derived from the analysis – conservative and parsimonious – are reported in the Annex. Table 4 shows the results of the minimization process. They indicated that the outcome is reached through two solution paths, given as: DEP*SIZ*INST*~ATD + SIZ*INST*IMP ⇒ POL. The first path indicates that the combination of high expected regulatory depth, large economic size, reformed institutional setup, and poor public attitudes toward the EU is sufficient for a high degree of politicization of EU trade agreement negotiations. This combination of conditions is found in Japan, Canada, and the USA PTAs. The second path suggests that the combination of reformed institutional context, large economic size, and the prospect of an increase in import competition also suffice for having highly politicized trade negotiations in the case of Mercosur. When either of these two combinations of conditions occurs, trade agreements negotiated by the EU are characterized by a high level of politicization. The solution consistency is 0.897 showing a good model fit. The overall coverage is 0.641, meaning that the results do not cover all the fuzzy-set membership scores in the outcome leaving some trade agreements unexplained. The raw and unique coverage values reveal that the first path covers the most cases and is therefore of greater empirical significance than the second path.

Table 4. Intermediate solution of sufficient conditions

Overall, the results show that the degree of public support for the EU is especially relevant when it comes to ‘deep and comprehensive’ agreements that touch on sensitive domestic issues, and trading partners holding high relative economic power. With ‘heavyweights’ such as the USA, Canada, and Japan, the presence of a skeptical public opinion with little trust in EU institutions has amplified European citizens' fears of regulatory downgrading, facilitating general mobilization against trade negotiations. In the case of Mercosur, a shallower agreement centered on the abolition of at-the-border trade barriers, the availability of institutional access to EU decision making, combined with an economically significant negotiating partner with a trade structure little compatible with EU's, proved sufficient to bring about politicization, even in the absence of general distrust toward the Union. The rationale behind the public mobilization was in fact primarily related to the scale of potential loss resulting from trade liberalization. Such a perspective aroused strong opposition from European farmers, who feared competition from lower-cost South American countries. Eventually, through the actions of interest groups engaged in raising public awareness, opposition also spread to various civil society actors. Particularly important in this regard was the participation of NGOs fearing the environmental consequences of the agreement, including the ramping up of deforestation in the Amazon region due to greater European demand for Mercosur agricultural products.

On the other hand, PTAs such as the association agreements (AAs) between the EU and its Mediterranean partners – namely Algeria, Egypt, Israel, Jordan, Lebanon, and Morocco – show little or no politicization. Since these agreements, rather limited in scope, were negotiated in the pre-Lisbon era with countries of limited economic influence at a time of broad public support for the EU, the potential for public contestation and mobilization was rather low. The PTAs with Central America, Singapore, New Zealand, and Chile also received little attention from European citizens. More recent than the AAs, the negotiations of these ‘deep and comprehensive’ agreements took place after the adoption of the ToL and the consequent reform of the EU trade policymaking. However, the limited bargaining power of the trading partners was not perceived by the public as a serious threat to European domestic regulations, limiting the opportunities for politicization.

Conclusion

The politicization of EU policymaking can have far-reaching consequences on its trade relations. A more polarized – and protectionist – European public opinion may limit the EU's ability to reap the benefits of trade liberalization and undermine its status as a major trading power. Yet our understanding of this phenomenon is rather limited. The present paper contributes to filling this gap by investigating the causes behind the politicization of EU trade policy over the past 30 years. Contrary to most existing studies, it looks not at the agents of politicization per se but at those ‘enabling’ conditions that create opportunities to mobilize public opinion in the first place. Moreover, the analysis is not limited to single cases of politicization but adopts a comparative perspective covering 19 bilateral and interregional EU trade agreements.

The QCA indicates that there is no one-fits-all path to politicization, rather it identifies two distinct patterns. The first one covers the agreements with the USA, Canada, and Japan, while the second the PTA with Mercosur. Specifically, the results show that (1) the enhanced institutional access to EU trade policymaking is a key driver of politicization, (2) the extent of public support for the Union is particularly relevant when it comes to ‘deep and comprehensive’ agreements touching on sensitive domestic issues, and that (3) whenever the agreement is expected to generate high adjustment costs, this can result in its politicization. The only instance of politicization that is not covered by the solution term is the PTA with the UK. Unlike the other agreements, it was not negotiated with a new trade partner but with a former EU member state. This peculiarity is also evident in that the public debate on the deal was characterized by a high level of media salience but a low level of polarization and expansion of actors. This case is arguably driven by different dynamics and needs a separate explanation.

Certainly, this study suffers from some limitations and needs to be approached critically. First and foremost, it addresses the politicization of trade agreements from an EU-wide perspective. Yet an equally important line of inquiry concerns the divergent levels of politicization among member states. As one example, notwithstanding the overall unprecedented level of politicization experienced by TTIP and CETA, their contentiousness in public and political debates has been far more pronounced in Western European countries than in Eastern European countries. This anecdotal evidence suggests the importance in future research of investigating how domestic-level variables influence the ability of entrepreneurs to mobilize and raise public awareness of agreements. On this account, several studies have shown how factors such as differences in public receptiveness, resource mobilization, and national political opportunity structures shape the success of political and societal actors in acting as agents of politicization in EU matters (see De Wilde and Zürn, 2012).

Nonetheless, the findings still contribute to a better understanding of the observed phenomenon. The comparative analysis shows that the presence (or absence) of a single factor is no guarantee that a trade agreement will become salient to the European public or result in a polarization of policymakers' positions. Instead, it reveals that politicization is a complex process resulting from the concurrence of several causal mechanisms. That all the conditions previously discussed serve as prerequisites for the politicization of trade negotiations helps, in this sense, to account for the non-linear pattern and limited scope of this phenomenon. That is, politicization is not a generalized trend but it requires the presence of specific combinations of conditions to occur. In line with previous work (Young, 2019), this study thus suggests that highly contentious PTAs such as TTIP and CETA are likely to be the exception rather than the rule. At the same time, the availability of greater institutional opportunities resulting from the ToL also implies that episodes of politicization in EU trade policy will occur more often than in the past.