The federalization of trade politics in Switzerland, Germany and Austria

ABSTRACT Since the 1990s, sub-federal units have become increasingly active in trade politics, a domain that is usually an exclusive jurisdiction of the federal level. Conceptualizing this process of institutional change as federalization, this paper adopts a most similar case design to examine how four factors interact to generate different patterns of trade policy engagement in Switzerland, Austria and Germany. The study suggests that institutional resources and preference intensity are crucial to explain long-term variation in all three federations, while the role of party politics and social mobilization are contextual rather than causal factors.

sub-federal units' participation in trade politics appears to be an increasingly widespread phenomenon. The patterns of engagement, as well as the causes behind this trend, however, indicate significant variation. On the one hand, it is rather uncontested that the changing nature of trade policy itself is a main driving force behind the emergence of multilevel trade politics (Baccini 2019;Broschek and Goff 2020). On the other hand, the role of intermediating factors such as institutional resources, party politics or social mobilization is still unclear. The case of Wallonia, for example, suggests that a high level of social mobilization prompts sub-federal engagement in trade politics (Bollen, De Ville, and Gheyle 2020). But this factorin isolationcannot explain the ongoing participation of sub-federal units where mobilization is low like in Australia, Canada or Switzerland.
This study uses the three affluent Continental European federations Switzerland, Germany and Austria in order to examine how different factors identified in the literature interact to generate different patterns of sub-federal engagement in trade politics. 2 More specifically, it addresses three core questions pertinent to the study of multilevel trade politics: First, what factors explain differences in sub-federal engagement in the trade policy domain? Second, what does this mean for the larger debate about the rise of subfederal units in trade policy? And third, what are the potential long-term consequences of enhanced sub-federal activism in the field of trade policy?
Building on the concept of federalization (Paquet 2019), the paper argues that sub-federal participation in trade politics represents a gradual, transformative process of institutional change in federal systems. Rather than being an episodic occurrence, sub-federal engagement has been 'in the making' for quite some time. Cantons and Länder have established themselves as stakeholders in trade politics but, notwithstanding certain similarities, the three case studies also reveal important differences. Most notably, the Swiss cantons have emerged as powerful sub-federal actors early on. By contrast, the role of the German Länder is moderate-strong and those of the Austrian Länder comparatively weak, and their engagement is more contingent on political conditions.
The first section provides a framework for comparing the trajectories of sub-federal engagement in trade politics by unpacking the concept of federalization. The second section systemically compares similarities and differences of trade policy federalization in the three cases. The third section discusses the relative weight and interplay of four factors identified in the literature on federalism and multilevel governance to explain for the rise of regional and sub-federal units in European and international affairs: institutional resources, preference intensity, party politics and social mobilization. The concluding section maps different causal configurations for each case, arguing that institutional resources and preference intensity are the most important factors for explaining the robustness of sub-federal engagement, while party politics and social mobilization are more contingent factors with the potential to amplify federalization processes.

Trade policy federalization
The Swiss cantons and the German and Austrian Länder have emerged as stakeholders in trade politics, similar to other sub-federal units in North America, Europe and Australia. Building on Paquet (2019), this paper conceptualizes this process as federalization. Paquet defines federalization as a transformative shift in the governance structures and processes within a policy domain. More specifically, through this shift, the dominance of the federal level is gradually replaced with a governance structure where both governmental tiers are active.
Federalization is different from centralization or decentralization, which implies that either the federal or sub-federal level assumes new legislative or regulatory competencies. Rather, federalization results from interdependencies within federal systems, which require coordination (Benz 2020;Bolleyer 2009). Such de jure and de facto interdependencies are an inherent feature of all federal systems, regardless of whether they belong to the more dual cluster in the Anglo-American tradition or the more integrated cluster in Continental Europe (Benz 2020, 39-45).
In case of trade policy, interdependencies resulton a general levelfrom the changing scope and depth of new trade agreements: Although, for the most part, an exclusive jurisdiction of the federal level, trade policy provisions increasingly penetrate sub-federal authority, which creates a need for both tiers of government to interact (Broschek and Goff 2020). The scope and depth of economic integration began to change profoundly with the negotiation of ambitious international trade agreements or treaties like the North American Free Trade Agreement (NAFTA) in 1994, the General Agreement on Trade in Services (GATS) in 1995, the Single European Act (SEA) in 1987 or the Maastricht Treaty (1993). More recently, the proliferation of new 'mega' or 'second generation' free trade agreements like the Transatlantic Trade and Investment Partnership (TTIP), CETA or the Comprehensive and Progressive Agreement for Trans-Pacific Partnership (CPTPP) have further amplified this trend (Baccini 2019;Ravenhill 2017;Young 2016). These trade agreements no longer primarily seek to reduce 'at-the-border' measures like tariffs, but also cover trade-related issues with 'behind-theborder' effects such as regulatory cooperation, investment protection and the liberalization of services, intellectual property or government procurement. In federal systems, such provisions often impinge on sub-level jurisdictions (Baccini 2019;Broschek and Goff 2020).
The changing scope and depth of trade policy reveals the ambiguity entailed in the division of competencies. Even if the federal (or supra-national) level enjoys broad and exclusive formal powers to negotiate and conclude trade agreements, it needs to ensure that sub-federal units comply with trade policy provisions that fall within their jurisdiction. In the German case, where the second chamber is composed of Land governments, the federal level also needs to secure formal sub-federal support in the ratification process. Moreover, new trade agreements can profoundly affect regional economies or the legislative and administrative autonomy of subfederal units, like in case of public procurement (see Freudlsperger 2020). Accordingly, sub-federal units may become active in this domain because new trade agreements variously mobilize their economic, political or institutional interests (Broschek and Goff 2020). Whenever trade policy provisions directly, or sometimes even indirectly, affect sub-federal units' competencies, the federal level will find it difficult to take refuge in the argument that it has exclusive jurisdiction over trade policy. Sub-federal units, on their part, may be in a position to demonstrate the limits of the scope of federal trade competencies, carving out sub-federal policy space within the larger trade policy domain.
These broad and general trends are hardly contested. Still puzzling from a comparative perspective, however, are the different patterns of trade policy federalization. In order to determine similarities and differences more systematically, the following section introduces a set of indicators to characterize the process of institutional change. It also illustrates how these indicators apply to the cases at hand.

Patterns of trade policy federalization in Switzerland, Austria, and Germany
Similarities Three indicators show similarities among Switzerland, Austria and Germany (Table 1). All three sequences of institutional change had the same trigger (1) and have been unfolding over the same time period (2, duration). Trade policy initially entered the realm of sub-federal politics through Europeanization. Sub-federal units began engaging with trade policy in the 1990s in the context of the Maastricht Treaty and the creation of the common market, and in international trade agreement negotiations since about the early 2000s. The Swiss cantons participated during the entire negotiation process over the so-called Bilateral Agreements I, which were concluded in 1999 after a majority of the population had rejected membership to the European Community in the 1992 referendum over Switzerland's accession to the European Economic Area (EEA) (Schweizerischer Bundesrat, 1999, 6145). 3 Around the same time, Austrian and German Länder began to address certain aspects of internal market reforms. Both passed resolutions, for example, on a Council Directive on the harmonization of construction products in 1999 (Deutscher Bundesrat, 1998;Verbindungsstelle der Bundesländer, 1998). The German Bundesrat also engaged intensively with the Directive on Services in the Internal Market (Deutscher Bundesrat, 2004a, 2004b, 2004c (Ziegler 2020). Similarly, the German Länder adopted resolutions on the World Trade Organization's (WTO) effort to liberalize agriculture trade (Deutscher Bundesrat, 1999, and the Austrian Länder demonstrated a comparatively high level of engagement with GATS (Verbindungsstelle der Bundesländer, 2003). 4 Since then, the Austrian Länder continued to issue occasional resolutions, for the most part dealing with specific aspects of trade liberalization within the European Union (EU) like product-related matters but also, more recently, with trade agreements in general. Two resolutions in 2014 and 2017, for example, entail rather general demands concerning new generation trade agreements like TTIP, CETA or TiSA (Trade in Services Agreement) (Verbindungsstelle der Bundesländer, 2014a, 2017). The German Länder ratified numerous trade and association agreements, most of them without any discussion. They also passed more specific resolutions, for example on the prevention of market access for products from exploitative child labour (Deutscher Bundesrat, 2010), on trade in services in 2006(Deutscher Bundesrat, 2006 on international trade and steel (Deutscher Bundesrat, 2018a) and Brexit (Deutscher Bundesrat, 2017a, 2017b, 2018b. In addition, the Länder drafted several resolutions addressing European and international trade policy more generally, as well as individual agreements such as TTIP and CETA (Deutscher Bundesrat, 2014, 2017c. Considering this ongoing engagement in trade politics, all three federalization processes can be described as transformative (3). First, according to authors like Pierson (2004) or Falleti and Mahoney (2015), a process of institutional change is transformative when a new activity continues even in the absence of the immediate conditions that were responsible for its initialization in the first place. Hence, while Europeanization performed as the trigger of trade policy federalization, the shifting focus of sub-federal engagement from matters related to the internal market to international trade policy suggests that the process has indeed become self-reinforcing. Second, neither the European Commission nor the federal level in all three countries have tried to reverse this trajectory. To be sure, the European Commission has only reluctantly declared CETA as a 'mixed' agreement. Likewise the Commission appears to negotiate controversial investment protection agreements increasingly separately, like in case of the EU-Japan Economic Partnership Agreement (EPA), in order to avoid sub-federal opposition. But neither federal governments, nor the Commission have outright opposed the federalization of trade policy in principle.

Differences
At the same time, the three cases display variation. First, although federalization has been transformative, the dynamic (4) underpinning each of the three processes of institutional change varies. According to Falleti and Mahoney (2015, 220-222), a self-reproducing, transformative process can be further distinguished as a continuous, self-amplifying and self-eroding process. In case of a continuous process, participation of sub-federal units becomes stably reproduced, while in case of a self-amplifying process initial steps toward federalization become gradually stronger over time. By contrast, in a self-eroding process the engagement becomes weaker. While no case shows any indication of a selferoding process, Switzerland corresponds to a continuous pattern, while in Austria and Germany the logic has been self-amplifying.
Switzerland represents the most robust case of federalization. The federal government and the cantons cooperate on trade-related issues on an ongoing basis. Although the federal level still has the final decision-making authority in trade politics, the cantons can substantially shape trade policy negotiations, either through consultation or, in some cases, even as a part of the Swiss trade delegation. The Austrian and German cases are different. In both federations, the Länder began to engage rather sporadically in matters related to trade policy between the late 1990s until about 2013. With the emergence of so-called second-generation trade agreements, this pattern changed. Since 2013 (Germany) and 2014 (Austria), respectively, the Länder have become more actively involved in trade politics than before.
Second, federalization patterns vary with regards to the role of cooperation and conflict in the intergovernmental arena (5). Tatham (2016, 15-19) distinguishes three types of modes: cooperative, conflictual or noninteraction. In Switzerland, interactions among the federal level and the cantons, as well as among the cantons, are generally cooperative, and both levels see cooperation on trade related matters as mutually beneficial. For example, cantonal executives highlighted the encompassing, transparent and ongoing involvement in the GATS negotiation process in several responses by cantonal governments to written questions submitted by cantonal parliamentarians (see for example Regierungsrat des Kantons Basel-Stadt, 2006;Regierungsrat des Kantons Bern, 2004). The federal government also acknowledged the important role of the cantons early on, embracing the benefits of cooperation with the cantons in the Bilaterale I process (Schweizerischer Bundesrat, 1999, 6145).
Although intergovernmental interaction in Austria is generally characterized as cooperative (Bußjäger 2006;Karlhofer and Pallaver 2013), recent developments in trade policy suggest a deviation from this pattern. On the one hand, the federal government honoured its obligation to inform the Länder in due time about ongoing or planned trade negotiations. The Länder, for their part, used their formal participation rights in European and international affairs to provide input and raise concerns. On the other hand, conflict arose as the positions between both levels of government diverged and the Länder unanimously opposed the ratification of new trade agreements like TTIP, TiSA or CETA. Despite this strong sub-federal opposition, the federal government pushed through unilaterally the CETA ratification bill in 2018 (Österreichisches Parliament, 2018).
In Germany, conflict intensified not only between the Länder and the federal government, but, even more pronounced, between the federal government and certain Länder. While all Länder shared basic concerns related to new trade agreements like ISDS provisions, they were not united in opposition like their Austrian counterparts. Some Länder were in principle supportive of new trade negotiations, but others were reserved or even explicitly opposed them.
Third and, perhaps most importantly, the patterns of federalization also differ in terms of their impact (6). This indicator captures the degree to which sub-federal engagement matters for the conduct of trade politics. Federalization had a strong impact in Switzerland, and a rather low impact in Austria, at least for now. Germany ranges somewhere in between.
Early on, the Swiss cantons were able to establish themselves as subfederal units with the power to shape significantly trade policy whenever their competencies are affected. The federal level has not only the obligation to inform and consult with the cantons, similar to Austria and Germany. The cantons can also participate effectively in the formulation of the negotiation mandate and in the negotiation process itself.
The Austrian Länders' ongoing engagement has been less consequential, as the unilateral ratification of CETA has shown. Cooperation always unfolds in the 'shadow of hierarchy' (Scharpf 1997). This does not mean, however, that federalization has been without any impact. For one thing, unilateral action comes at a price. It will be difficult for future federal governments to use their power to unilaterally ratify controversial trade agreements against broad sub-federal opposition. For another, trade policy will remain on the sub-federal agenda, and the federal level must continue cooperate with the Länder in a meaningful way to avoid legitimation problems.
Finally, the German Länder are confined to information and consultation rights, similar to the Austrian Länder. Moreover, they can't participate in the formulation of the negotiation mandate, or in trade negotiations like the Swiss cantons. The Bundesrat, however, affords them with a potential veto right in the ratification phase. The federalization of trade politics in Germany thus emerged within the path dependent pattern of joint decision-making (Scharpf 1988). The anticipatory effect of a potential Länder veto at the end of the negotiation process generates pressure on the federal and supranational level to take sub-federal concerns more seriously.

Explaining variation in trade policy federalization
The following analysis adopts a most similar case design and uses a process tracing approach to establish the causal configuration responsible for generating variation in the three federalization processes (Beach and Pedersen 2013;Blatter and Blume 2008). According to Blatter and Blume (2008, 322) this requires demonstrating (1) that the interaction effects of two or more factors is critical for the outcome, and not just their (co-)existence; (2) that these factors work within a specific context and (3) that a first causal factor is necessary for the activation of the other factors.
The previous sections established the second and third condition. Section one has shown that the changing nature of trade policy, especially the growing penetration of sub-federal authority through behind-the-border effects, ultimately triggered federalization processes. Section two has demonstrated contextual similarities, most notably Europeanization. Another similarity pertains to their historical-institutional legacy: The three federations share important characteristics, especially if compared to their Anglo-Saxon counterparts: They have a common legal tradition, feature an integrated allocation of competencies where sub-federal units implement federal law, as well as a comparatively strong institutionalization of intergovernmental relations (Thorlakson 2003).
This section introduces four factors that have been identified in the literature on federalism and regionalism that appear to be of particular importance to understand how and why sub-federal emerge as actors in areas that had traditionally been exclusive domains of the federal or supranational level. It focusses on the relative causal weight each factor carries for explaining the three cases. In the final concluding discussion, the paper then ascertains dynamic interaction effects between and among those factors.
First, institutional resources are obviously crucial for sub-federal units to position themselves in trade politics (Broschek and Goff 2018;Freudlsperger 2018Freudlsperger , 2020. This entails formal participation rights or competencies that empower them in different stages of the policy cycle, ranging from access to information or the right to be consulted over the right to participate in the negotiation or ratification process. In addition, institutional resources include intergovernmental capacities to coordinate positions horizontally among sub-federal units or vertically vis-a-vis the federal level, most notably through strongly institutionalized intergovernmental bodies (Bolleyer 2009;Schnabel 2020).
Second, as Blatter et al. (2010) and Tatham (2016) have argued, preference intensity can explain the rise of regions in foreign affairs more generally. Accordingly, it is reasonable to expect increased sub-federal engagement in trade politics whenever they perceive trade policy agreements to have an impact on their regional economies, culture, or political autonomy. As Tatham suggests, the degree of sub-federal preference intensity not only explains their activity per se, but alsopotentiallythe pattern of interaction between lower level and higher-level tiers. Low sub-federal preference intensity facilitates cooperation or encourages no interaction at all. By contrast, high sub-federal preference intensity may facilitate cooperation, but only if the preferences of sub-federal and federal governments converge. If they diverge, conflict is likely to arise (Tatham 2016, 22-24).
Third, party politics can matter. The configuration of party politics in multilevel systems has been analyzed primarily along two dimensions: the degree of congruence between party systems and the degree of organizational integration of political parties that operate on different scales (Detterbeck 2012). Tatham (2016, 26) operationalizes this factor more generally by looking at the partisan composition of federal and sub-federal governments at a given point in time. Accordingly, two configurations have the potential to variously affect sub-federal engagement in trade politics: If governments on both governmental tiers are from the same political party or, in case of coalition governments, represent similar political ideologies, it is reasonable to expect cooperative mode. Vice versa, divergence in the political composition of governments is more likely to generate conflict on trade policy related matters.
Finally, the literature on trade policy, in particular, has emphasized the role of social mobilization as a key explanatory factor to understand the emergence of sub-federal engagement (Bollen, De Ville, and Gheyle 2020;Meunier and Czesana 2019). According to this perspective, global trade has increasingly been perceived and framed as a matter that affects regional collective identities, therefore mobilizing collective action through social movements and interest groups. Potential constraints on local 'self-rule' not only mobilize the institutional self-interests of sub-federal governments, but also regional communities per se (Hooghe and Marks 2016). From this theoretical angle, increased sub-federal engagement in trade policy results from a high degree of social mobilization as issue salience almost inevitably enters party politics on different scales, and therefore, sub-federal politics. By contrast, if social mobilization is low, the logical expectation is that sub-federal units remain largely inactive.

Institutional resources
Freudlsperger (2020) has shown in his comparative analysis of the United States, Canada and the EU that participation rights and a strongly institutionalized system of intergovernmental relations facilitates cooperation and, ultimately, empowers sub-federal units to participate in trade policy in a meaningful way. However, although the three federations are rather similar in this respect, several differences warrant closer examination.
Regarding competencies, sub-federal units in all three federations successfully pushed for institutional reforms in the wake of Europeanization that provided them with new constitutional participation rights. In Switzerland, a bill was agreed upon in 1996 and constitutionalized in 1999 in the wake of a major constitutional reform through Art. 54 (3) and 55 of the Swiss Constitution (Schweizerischer Bundesrat, 1994, 1997, 1999. The German Länder were able to gain major concessions from the federal government, entrenching new participation rights in European affairs through the Gesetz zur Einheitlichen Europäischen Akte in 1986 and the new Article 23 Basic Law in 1992. In addition, the Lindauer Abkommen (Agreement of Lindau) from 1957 laid out a general framework for Land participation in international affairs whenever treaties have implications for their competencies. Finally, the Austrian Länder negotiated constitutionalized rights through Art. 23 of the Federal Constitution, which took effect already in 1992, about three years before Austria formally joined the EU (Bußjäger 2006). Regarding intergovernmental relations, which exhibit a comparatively high degree of institutionalization in all three federations, certain differences need to be highlighted, too. Arguably, the German Länder stand out insofar as the system of intergovernmental relations has traditionally been very dense and differentiated, vertically as well as horizontally. Moreover, the Bundesrat itself works in tandem with the various ministerial conferences: deliberations held in the latter usually prepare the agenda for legislative decision-making in the Bundesrat, which serves as an extension of intergovernmental relations (Hegele and Behnke 2017). This allows the Länder to coordinate themselves effectively. Since 2013, many Bundesrat resolutions, for example, were prepared in different horizontal bodies. Land ministers began developing positions on trade policy in 2014 within sectoral intergovernmental bodies like the Conference of Ministers for European Affairs, the Conference of Ministers for the Environment, Conference of Ministers for Agriculture, the Conference of Ministers for Economic Affairs or the Conference of Ministers for Consumer Protection (Broschek, Bußjäger, and Schramek 2020, 223). Meeting protocols indicate different concerns related to trade agreement provisions in TTIP and CETA for areas like consumer safety, environmental policy and standards regarding infrastructures and services as well as claims regarding the procedural involvement of the Länder. At the highest level, coordination took place within the framework of the Conference of the Minister Presidents (Ministerpräsidentenkonferenz, MPK). Trade policy also entered the vertical arena of intergovernmental relations. Land representatives met on a regular basis with the federal Ministry of Economic Affairs and Energy (BMWi), through Land observers in committee deliberations in the federal parliament, the Bundestag, but also through written correspondences and telephone conferences between both levels. 5 These intergovernmental channels were all activated as needed, and fluctuated depending on issue salience and the general importance of a trade agreement. 6 While the German Länder emerged as potential veto players in trade politics because of the Bundesrat, the Austrian Bundesrat is only similar in name. The Austrian Bundesrat is not only weaker, as it has usually only a suspensive veto, but also differs in its composition. Although its main responsibility is to protect sub-federal interests, de facto it has become a vehicle of party politics as its 61 members are elected by sub-federal parliaments (Fallend 2015). This is somewhat different in Germany: Although the German Bundesrat has never been immune from party politics either, the interplay of territorial and partisan interests is more contingent on the general configuration of governments on the Land and federal level (Lehmbruch 2000). In recent years, moreover, territorial interests seem to have become more important than partisan logics in the decision-making processes of the Bundesrat. Not surprisingly, the Austrian Bundesrat played a rather insignificant role for trade policy federalization. On the one hand, the Bundesrat passed a first resolution on TTIP in July 2014 (Ősterreichischer Bundesrat 2014), echoing largely the concerns expressed by the Länder, and a second resolution in May 2016 on CETA. This resolution reaffirmed the Länders' position, explicitly mentioning, for example, a detailed resolution of the Landtag of Salzburg on CETA from 29 April 2016 and opposing the provisional application of CETA (Ősterreichischer Bundesrat, 2016). Despite these clear commitments, the Bundesrat eventually ratified CETA in June 2018 with the support of the Christian ŐVP and the right-wing FPŐ, the two political parties that had formed the federal coalition government in December 2017.
The most important intergovernmental resource for Land governments is not the Bundesrat, but the horizontal Landeshauptleutekonferenz (LHK). Established in the 1960s, the LHK serves as the main instrument for the Länder to position themselves vis-à-vis the federal level, in European affairs, but also more generally. Their general resolutions are not binding, but often have a political effect especially if supported unanimously (Bußjäger 2006;Karlhofer and Pallaver 2013). The Austrian Land governors' strength in trade politics was their capacity to collaborate closely through this peak-level horizontal body. Overall, the Länder adopted three joint resolutions and one general resolution on trade policy between 2014 and 2017 (Verbindungsstelle der Bundesländer 2014a, 2014b, 2017.
Until the early 1990s, the Swiss cantons lacked similar horizontal capacities. This changed with the creation of the Konferenz der Kantonsregierungen (KdK) in 1993. Since its creation, the KdK has emerged as the key intergovernmental body for cantonal governments to coordinate their interests and engage with the federal government, especially in European and foreign policy (Schnabel and Mueller 2017;Vatter 2018). Early on, the KdK became crucial for developing joint positions among the cantons. Through the KdK, the cantons established a point of contact for the federal level in all questions regarding foreign affairs. Less than a year after its founding, the KdK was able to reach agreement with the federal government about the instalment of a cantonal information officer within the federal Department of Foreign Affairs (EDA) and the Department of Economic Affairs (EVD). Moreover, the Bundesrat agreed to include cantonal representatives from the KdK into federal delegations that negotiate treaties with the EU whenever cantonal jurisdictions are affected (Mayer 2003;Ziegler 2020). Since the early 2000s, the KdK participated effectively in GATS (KdK 2003(KdK , 2005(KdK , 2006 and all other important international trade negotiations.
To summarize, direct formal competencies and intergovernmental capacities furnish the Swiss cantons with strong institutional resources. Under certain circumstances, the German Länder can activate the Bundesrat, which, in addition to their strong intergovernmental capacities, establishes them as potentially powerful stakeholders. The Austrian Länder, by contrast, can only capitalize on their encompassing information and consultation rights, in combination with the LHK.

Preference intensity
Preference intensity towards trade policy also displays variation among the three cases. The Swiss cantons usually formulate comparatively detailed recommendations or demands applying to each trade agreement under consideration. These resolutions are rather technical, and often discuss potential effects on their competencies, most notably in the field of services, procurement, agriculture, health or education. Since the Swiss cantons are involved in the formulation of the negotiation mandate, cantonal resolutions seek to determine the scope of concessions the federal government can commit to in the negotiation process. Occasionally, resolutions also include more political concerns, like in case of the 2013 China and Switzerland Free Trade Agreement (CSFTA) (KdK, 2010). Overall, preference intensity can be characterized as moderate to strong, depending on the trade agreement. Although the cantons sometimes seek to curb the scope of intended liberalization commitments by the federal government, both levels usually agree on a joint position. 7 This is different in Austria and Germany. In both cases preference intensity has increased over time, which is not only reflected in the frequency of traderelated resolutions since 2013, but also in the level of engagement. More than in the past, the Länder have been dealing quite intensively with trade policy developments in general, and individual trade agreements more specifically. The German Länder stand out in terms of the sheer number of trade-related resolutions: Between 2013 and February 2020, the Länder adopted 12 resolutions 8 and ratified seven trade agreements. Although the Austrian Länder only passed four resolutions between 2014 and 2017 (Verbindungsstelle der Bundesländer 2014a, 2014b, 2017, the content of these resolutions suggests high issue salience. While Austrian and German Länder acknowledge the importance of open trade relationships, they are unambiguous in expressing primarily political and institutional concerns. In Germany, the critical moment occurred already before TTIPwhich is often portrayed as a game changer (Bollen, De Ville, and Gheyle 2020;De Ville and Siles-Brügge, 2016)entered the agenda in 2013. The Länder had to ratify the EU-Central America Association Agreement in February 2013. The bill passed the first reading (Deutscher Bundesrat, 2013a), but the scope of the agreement hit Land governments by surprise. 9 Only a few weeks later, in May 2013, the Bundesrat had to ratify a second major trade agreement between the EU and Colombia and Peru (Deutscher Bundesrat, 2013b). This agreement stirred a major debate in the Bundesrat on May 3, 2013 (Deutscher Bundesrat, 2013c). Eventually, the Bundesrat passed both agreements with an accompanying resolution initiated by the SPD (Social Democratic Party)-Green Party coalition government of North Rhine-Westphalia that referred to both agreements. On 7 June 2013, the Bundesrat eventually passed the ratification bill of the EU-Central America Association Agreement in a second reading, also with an accompanying resolution that refers to both agreements (Deutscher Bundesrat, 2013d). Interestingly, both resolutions were not primarily concerned with sub-federal competencies. While the Länder acknowledged the benefits of trade liberalization, they criticize, in particular, the imbalance between these provisions and enforcement mechanisms that ensure the promotion of labour, social, environmental, consumer and human rights, along with several more specific concerns applying to Latin American countries.
During the same Bundesrat session in June 2013, the Bundesrat also passed its first resolution on TTIP (Deutscher Bundesrat, 2013e), this time calling on the federal level and the EU to respect and protect sub-federal jurisdictions for culture and media, followed by another TTIP resolution in 2014 on the implications of investor-state dispute settlement (ISDS) mechanisms. These and other sub-sequent resolutions are very similar to the four resolutions passed by the Austrian LHK. Sub-federal units express strong reservations, or even reject individual agreements, because they anticipate constraints on their authority to provide essential social services and public goods like housing, energy, water supply, or public transport. They also share the view that new trade agreements may negatively affect high environmental-, consumer-and social standards. Investment protection mechanisms and regulatory cooperation, in particular, surface as one of the most pressing concerns as the Länder expect irreversible, negative longterm effects for their right to regulate and provide public services. In addition, recent Land resolutions on Brexit (Deutscher Bundesrat, 2017a, 2017b, 2018b and international trade and steel (Deutscher Bundesrat 2018a) suggest that economic concerns over protectionist measures also motivate especially those Land governments whose regional economy is highly export-dependent. 10 Overall, these resolutions indicate strong sub-federal preference intensity, especially since 2013. Concerns over the impact of trade agreements on the political autonomy appear to be most important in this respect, followed by a desire to promote (or protect) their regional economy. However, cultural interests, which are the third type of interest suggested by Blatter et al. (2010), barely seem to play a significant role.

Party politics
The influence of party politics on the federalization process also reveals variation. The first major difference pertains again to Switzerland on the one hand, and Austria and Germany on the other hand. Regarding the former, and with the notable exception of occasional referenda on highly salient issues, the role of party politics is rather limited in Swiss foreign affairs (Schneider 1998). This is also true for trade policy. The continuous, but silent and pragmatic participation of cantons in trade politics has been largely immune from the dynamics of partisan competition. 11 In Austria and Germany, by contrast, party politics are influential. On the one hand, new trade agreements seem to polarize political parties along the left-right continuum. While centre-right parties are generally in support of agreements like TTIP, CETA or TiSA, centre-left parties have either major reservations or oppose them altogether. The same applies to the far-right parties FPŐ (Freiheitliche Partei Österreichs) in Austria and the AfD (Alternative für Deutschland) in Germany. However, since supportive and opposing parties have been part of coalition governments on both governmental tiers (with the exception of the AfD), the decision to approve or reject the ratification of individual agreements can create a dilemma.
In Germany, this dilemma posed a serious problem for the SPD and the Green Party. The SPD has been in a coalition with the Christian democrats (CDU/CSU) on the federal level since 2013, and was also a junior or senior coalition partner in several Land governments. The Green Party has formed coalition governments as a senior partner with the CDU since 2016 in Baden-Wurttemberg, and as junior partner in Hesse since 2014. While coalition governments in the Länder usually adopt the rule that they abstain from voting in the Bundesrat on controversial issues, this counts, de facto, as a vote against a bill. Since it has been far from clear whether CETA, for example, would enjoy the necessary support from Land governments, this problem became further aggravated. And although former Minister of Economic Affairs Sigmar Gabriel (SPD) was able to gain support of the federal party organization during a special party congress on CETA held in September 2016, this decision did not resolve the dilemma on the Land level where several SPD-led governments remained silent on how they would vote in the Bundesrat (Broschek, Bußjäger, and Schramek 2020). The interplay of federalism and polarized party politics in trade policy followed one typical pattern of joint decision-making in Germany: deliberate neglect in order to avoid deadlock. In anticipation of a possible ratification failure, the federal government, has been reluctant to introduce a ratification bill, formally pointing to a still pending decision by Germany's constitutional court.
The case of CETA ratification in Austria is remarkable in so far as it represents a deviation from the typical pattern. 12 Since political parties in Austria display a high degree of vertical integration, sub-federal party organizations usually comply with the position taken by the federal umbrella organization. Recent developments in trade policy, however, indicate that territorial interests can trump party politics: The Christian democratic ŐVP (Österreichische Volkspartei) ultimately ratified CETA in 2018 with the support of their junior coalition partner FPŐ on the federal level although the party was, at the same time, a coalition partner in seven of the nine Länder, and the FPŐ in three Land governments. And even before CETA was ratified, trade policy had divided political parties in Austria less in terms of their ideology on the left-right continuum, but more in terms of whether they were federal or sub-federal organizations. Accordingly, party politics is not unimportant to understand the Austrian case, but its immediate effect on trade policy related decisions is surprisingly low.

Social mobilization
Social mobilization also suggests important differences between Switzerland on the one hand, and Austria and Germany on the other. To be sure, trade policy is not entirely uncontroversial in Switzerland. GATS prompted responses from several Swiss municipalities and cantons, like Basel-Stadt, to declare themselves as GATS-Free Zones (Regierungsrat des Kantons Basel-Stadt, 2006). Since about 2015, TiSA again emerged as a trade agreement with the potential to mobilize broader opposition. However, with the negotiations on TiSA being stalled, issue salience abated. The degree of social mobilization in Austria and Germany, by contrast, was considerable higher and shaped political discourses within Land legislatures more profoundly.
In Germany, NGOs began to mobilize against the EU and Columbia and Peru Free Trade Agreement, and deliberately targeted the Bundesrat as a political arena with the potential to prevent its ratification. In April 2013, shortly before the Bundesrat adopted its resolution on the agreement, a broad alliance of 45 humanitarian organizations, environmental groups and unions published an open letter, urging the Bundesrat to reject the ratification of the agreement in its current form (Kolko, 2013). Mobilization gained further momentum as NGOs and (often local) party organizations organized an unforeseen level of protests against TTIP and CETA in several German cities. At the peak of this wave, more than 150,000 people rallied in Berlin in October 2015. Campact, Foodwatch and Mehr Demokratie joined forces and, backed with more than 125,000 signatures of supporters, filed a case against CETA at the Constitutional Court in August 2016 (Campact et al., 2016). Austria witnessed a similar surge of protests during that period. Thousands of protesters flocked the streets demonstrating against TTIP and CETA in 2016, and public opinion surveys showed that a large majorityaround 75%of Austrians opposed CETA. 13 These protests resonated on the sub-federal level as opposition parties submitted questions and stirred debates about how the respective Land government intends to position itself towards agreements like TTIP, CETA, EPA or TiSA. In Bavaria, the SPD, the Green Party and the Free Voters were particularly active and kept the debate alive through numerous parliamentary initiatives since 2013. The Green Party-SPD coalition government of Baden-Wurttemberg proactively established an advisory body in September 2015, the TTIP-Beirat, and commissioned a legal expertise to assess the potential impact of CETA on the political autonomy of Länder and municipalities. In Austria, numerous parliamentary resolutions critical of recent trade agreements have been adopted on the Land level since 2014. For example, Land parliaments of Salzburg, Lower Austria and Upper Austria passed several resolutions, often carried by all or a large majority of political parties on agreements like TTIP and CETA or more specific issues entailed in new agreements, most notably their impact on agriculture policy, genetically modified products or ISDS mechanisms (Broschek, Bußjäger, and Schramek 2020). Social mobilization, therefore, mattered. Yet, as will be shown in the final section, it is equally important not to overestimate its influence, especially if the role of social mobilization is analyzed in the broader causal configuration of factors.

Discussion and conclusion
This comparative study of trade policy federalization in Switzerland, Austria and Germany set out to address three questions. First, what explains noticeable differences of these processes in three otherwise rather similar cases? Second, what does this mean for the broader theoretical debate about the emergence of sub-federal actors in international trade policy? And third, what can we conclude about the possible long-term implications for trade politics?
In order to address the first question, this paper adopted a most similar case design and used process-tracing to examine how individual factors generate variation. Pattern variation was established with respect to three properties: The dynamic of federalization, the degree of cooperation and conflict in intergovernmental relations, and the impact of federalization.
To explain these differences, the paper started from the rather uncontested observation that the changing nature of trade policy represents the main cause of heightened sub-federal engagement since about the 1990s. But this observation tells us very little about why sub-federal units embark on different pathways of federalization. A closer analysis of how institutional resources, preference intensity, party politics and social mobilization interact in individual cases promises to deliver a more fine-grained picture of how this causal configuration generates variation (Table 2).
From a comparative point of view, first of all, institutional resources matter. The Swiss case stands out here as the cantons enjoy broad and direct participation rights beyond information and (non-binding) consultation. Perhaps most importantly, and different from Austria and Germany, the cantons can influence trade policy negotiations early in the policy cycle through their participation in the formulation of the negotiation mandate and, in some cases, by even participating in the negotiation process itself. Potentially conflictual issues can thus be identified and resolved before negotiations begin. This is an important difference to the German case, where Länder can become veto players in the ratification process, often after years of negotiations. The Austrian Länder have a constitutional right to be informed and consulted, similar to the German Länder. However, they lack the veto power of the German Länder through the Bundesrat. In addition, intergovernmental capacities, especially to coordinate horizontally, are an important prerequisite for subfederal units to position themselves in all three cases.
But institutional resources alone cannot explain how political actors utilize these in practice, which is why it is necessary to examine their interaction with more dynamic factors. In Switzerland, preference intensity remained moderate to high, and barely interacted with party politics and social mobilization. This explains why federalization has been a rather unspectacular process: The cantons have been cooperating with the federal level on an ongoing basis, eventually raising concerns about the implications of trade provisions for their capacity to regulate and deliver public services, but without much conflict. In Austria and Germany, by contrast, preference intensity has increased with the emergence of new trade agreements since 2013. However, the events in early 2013 suggestand this is importantthat social mobilization and party politics did not precede increased sub-federal engagement. It was the first reading of the EU-Central America Association Agreement in February 2013and not TTIPthat put the Länder on alert, regardless of the partisan composition of individual Land governments. This event was reinforced through the requirement to ratify the EU-Colombia and Peru agreement just a few weeks later and with TTIP on the horizon. Within a very short time period, the Länder realized that these new trade agreements require closer attention. This finding supports the importance of preference intensity to explain why sub-federal units become active: Concerns over their political autonomy and regional economy are major driving forces, while this study found no support for the argument that cultural interests motivate their actions. Party politics and social mobilization are nevertheless important to explain variation, with respect to the dynamic of federalization, the relative degree of cooperation and conflict as well as the impact of federalization. First, as catalysts rather than causes, the timing of social mobilization in Austria and Germany amplified high preference intensity among the Länder, and sharpened divisions among the Länder in Germany. For example, KOLKO's open letter to the Bundesrat in April 2013 did not trigger heightened Länder engagement, but certainly reinforced and amplified this trend. Second, the interaction of party politics and institutional resources explains variation. In Germany, partisan conflict can lead to stalemate in the ratification process, which is the main reason for why CETA's ratification is still pending. In Austria, by contrast, party politics wassurprisinglyless important as Land governments unanimously rejected CETA and other pending agreements across all party lines. Third, the unprecedented level of social mobilization between 2014 and 2016 can continue to affect how federal and sub-federal governments position themselves in future trade agreements. Decision makers are wary of a potential resurgence of protests. As well, NGOs and trade unions continue to mobilize their ties with social democratic and Green parties in sub-federal parliaments and governments on trade-related matters. The EU-Mercosur trade agreement, for example, will be an important test case in this respect.
The broader theoretical debate about the emerging role of sub-federal units and, possibly, regions in trade politics should therefore be careful not to isolate individual variables at the expense of others. Rather, this study suggests that complex processes of institutional change warrant a closer examination of how individual factors interact. Institutional arrangements obviously matter, and this factor interacts with preference intensity, which in itself can be reinforced through partisan dynamics and social mobilization. However, considering that sub-federal units in all three federations have been engaging in trade politics since the 1990s, for the most part in absence of larger mobilization, these factors should be understood as catalysts rather than causes in their own right.
Finally, what does this case study suggest for the more general question of how the emergence of sub-federal actors in trade politics will affect future negotiations? As economic integration has become deeper, interdependencies between governmental tiers have increased through trade policy. Even countervailing trends like Brexit or the resurgence of protectionism did not reverse this trend. On the contrary, the Swiss cantons and German Länder have been actively engaged in these processes as well. It is thus reasonable to assume that interdependencies will consolidate sub-federal participation. Although sub-federal units can become veto players under certain circumstances, it is rather unlikely that they will prevent future trade agreements altogether. CETA may never be fully ratified but remain in place as a provisional agreement. Considering the concerns expressed by German and Austrian Länder, and the Belgian Regions and Communities as potentially powerful sub-federal stakeholders, federal governments and the EU have an incentive to negotiate more 'progressive' trade policy agreements, and to promote a more transparent trade governance structure like in case of the new Investment Court System. Notes 1. For example, in their authoritative attempt to develop a new theory on the political foundations of new preferential trade agreements, Mansfield and Milner (2012) shed light on the important role of domestic factors without, however, even mentioning federalism as a potential variable. 2. The study builds on an extensive analysis of primary documents, secondary literature and supplementary interviews with government officials in all three federations. 10 interviews were carried out by email, using a semi-structured questionnaire consisting of five questions that were adjusted for each case. I refer to eight responses in this study, which are listed in the online appendix. The purpose of these interviews was to gain deeper insight into the motives and perceptions of involved actors, and to gain information that was not readily accessible from primary documents such as resolutions, press releases or parliamentary proceedings. 3. All primary documents are listed in the online appendix. 4. Interview 1. 5. Interviews 2, 3, 4, 5. 6. Interview (by email) Land Rheinland-Pfalz, April 3, 2020. 7. Interview 6 and 7. 8. Four resolutions deal with Brexit and, therefore, address more than just trade related matters. 9. Interview 3. 10. Almost all interviewees from German Länder mentioned the importance of economic interests as well. 11. Interview 6 and 7. 12. Interview 8. 13. According to one survey from September 2016: Mehrheit ist gegen CETA, https://oesterreich.orf.at/v2/stories/2796984/ [last accessed April 20, 2020].