Legislative communities. Conceptualising and mapping international parliamentary relations

Abstract

Besides the increasing scope of transnational activities of civil society actors, international relations of national legislatures have long been expanding, yet without attracting substantial scholarly attention. We can observe that national Members of Parliament meet in various bi-and multilateral organisational forms within and beyond international organisations to fulfil parliamentary functions. We present a conceptual framework differentiating between two forms of international parliamentary relations: multilateral vs. bilateral organisation. We argue that multilateral participation is mostly driven by the supply of such organisations and can mainly be found in Europe and Africa. On the contrary, the capacity of chambers can explain the realisation of bilateral channels. We test our claims with data for the international relations of 144 national parliaments. Our explorative empirical study is the first to jointly analyse bi- and multilateral transnational parliamentary relations and shows that international parliamentary cooperation varies over legislatures and regions, generating genuine clusters of institutionalised communities. Our findings help to embed the existing research on international parliamentary institutions and diplomacy in a larger context of international relations. Furthermore, our global relational account of national parliaments speaks to research on diverse topics of domestic outcomes, such as democratisation, norm and legal diffusion, and governmental control.

Introduction

Located somewhere in the grey zone between governmental (state-centric) and non-governmental (citizen-centric) perspectives in International Relations (IR), the international activity of national legislatures¹ remains an under-investigated field of study. This lacuna is undue. Normatively, national legislators are the main democratic representatives of citizens, and their stronger involvement in international politics could be an effect of the pressures of ‘globalisation’ that citizens perceive nowadays. Empirically, international parliamentary cooperation has been a vivid and expanding phenomenon on the global political scenery for years. In fact, amidst the growing dynamics of international cooperation and governmental delegation of national policy competencies to international levels, national parliaments have long been engaged in international ‘parliamentary diplomacy’ (Malamud and Stavridis 2011; Stavridis and Jančić 2016). Examples include international activities of foreign affairs committees or other committees engaging in international affairs, ad hoc bilateral visits of parliamentary presidents and Members of Parliament (MPs), International Parliamentary Friendship Groups (IPFGs), and membership in International Parliamentary Institutions (IPIs) (see Fiott 2011; Malamud and Stavridis 2011; Lüddecke 2010; Crum and Fossum 2013a; Alger and Kill 2014).

However, despite this plethora of international parliamentary cooperation, we know surprisingly little about the empirical realisations of these diverse activities and their different modes of institutionalisation. How does the global distribution of parliamentary cooperation look like? Does it differ across world regions? Can factors like the type of political regime and chamber characteristics relate to parliaments’ international activities? How do different forms of organising interact? In short, what are the drivers and patterns of organising transnational parliamentary relations? We contribute to closing this gap with an explorative analysis of bi- and multilateral international parliamentary cooperation data for 144 national parliaments and legislatures. We approach these questions from the perspective of the national parliament (as opposed to research that starts from international organisations [IOs]). We argue that bi- and multilateral structures provide different cooperation dynamics and interact to generate clusters with varying intensities of institutionalised parliamentary relations. We conceptualise these clusters as legislative communities, map empirically how national parliaments currently cooperate, and analyse how this cooperation varies over legislatures, regime types, and regions. In so doing, our analysis provides the first empirical exploration of the combination of bi- and multilateral transnational parliamentary cooperation and integrates this strand of literature into more extensive research on international cooperation, organisation, and domestic politics.

Our results show significant variation between chambers in bi- and multilateral activities. On the first level—the frequency of participation—we find the emergence of strong regional groupings in the multilateral dimension. Two world regions—Europe and Africa—are highly institutionalised, while the Americas, Asia, and Oceania do not provide as many opportunities for multilateral exchange. On a bilateral dimension, there is no clear regional trend that parliaments establish most bilateral exchanges. However, we can observe that the capacity of parliaments influences the likelihood of being very present on a global stage. Only larger chambers realise a great many bilateral contacts. On the second level—the clustering of parliaments based on their combination of bi- and multilateral relations—we show that the multilateral environment determines strong regional interaction. Our results also suggest that some parliaments use their bilateral relationships to bridge certain world regions, such as France, Turkey, and Morocco. With these findings, we provide a first global exploration of multifaceted transnational parliamentary cooperation and show that the main channels of representative democracy are connected beyond their region, political system, or ideology.

Our approach and empirical findings complement previous studies on IPIs that generally start from the macro-IO perspective and ask why IOs set up parliamentary institutions in the first place. It also extends the existing literature to include and combine IPIs and parliamentary friendship groups conceptually and empirically into a global network of legislative communities with varying degrees of density. This perspective can stimulate future research on international parliamentary relations and political interactions between international and domestic levels.

We structure this paper as follows. The first section reviews the literature on international parliamentary relations and finds a missing link between various institutional forms of parliamentary cooperation and their interaction. We approach this lacuna in the next section with our concept of legislative communities. Building on this conceptual framework, we develop hypotheses to explore institutional dynamics of frequency and relations of transnational parliamentary cooperation. We detail the findings of our analysis after presenting a novel data set on bi- and multilateral parliamentary relations of 144 national parliaments. We discuss these results with their current limitations and how they contribute to the emerging literature before summarising and concluding the paper.

International parliamentary relations in IR literature

All over the world, we can observe legislators being active beyond their nation-state: international organisations (IOs) established institutions to let MPs, rather passively, participate in their activities, parliamentarians created international institutions on their own to handle specific global or regional policy problems, and MPs meet on a bilateral level to exchange views and advice. A nascent strand of literature in political science and IR has been emerging to explore these phenomena. However, it departs either from the angle of the international organisation or from concrete policy issues.

The IO perspective mainly focuses on multilateral institutions of national parliaments in international politics and the European Union as a particular case of a directly elected international parliament. This literature first focuses on the development of typologies of IPIs by referring to stages of institutional development (Cutler 2001, 2013), formal decision-making competencies to influence adjacent inter-governmental bodies (Kraft-Kasack 2008), or organisational characteristics (Cofelice 2012; Costa et al. 2013b; Kissling 2011; Šabič 2008). These typologies are often employed for operationalising functions and powers when focusing on IPIs’ emergence or empowerment (Cofelice and Stavridis 2014) beyond the high-profile case of the European Parliament (see Rittberger 2003, 2007, 2012; Hix and Høyland 2013).

Based on these findings, we know that IPIs have been spreading regionally and globally since the end of World War II, especially since the fundamental transition of the international systems with the democratic revolutions around 1989/1991 (Šabič 2008). A few single case studies find a combination of internal and external factors leading to institutionalisations of multilateral IPIs depending on geographical and political region, time, and political contexts (Navarro 2010; Verdoes 2020). More recently, some studies build on proliferating data set collections on international organisations to supplement these findings. One study group finds that regional (general purpose) IOs tend to establish parliamentary bodies for legitimation purposes—rather than despite their often weak institutional power (Rocabert et al. 2019; Schimmelfennig et al. 2020). Other research identifies degrees of the formal authority of adjacent IO secretariats, regional diffusion, and close cooperation with the European Parliament as key driving factors (Lenz et al. 2019; Lenz 2021). Even more sceptical of IPIs legitimation benefits, Jetschke and Münch (2020) find no association between domestic levels of democracy and establishments of IPIs. Rather, functional dimensions such as IO policy scope as well as trade and conflict-related variables appear strongly associated.

In addition to the IO perspective, the policy perspective emphasises legislative ‘soft power’ by investigating parliamentary involvement in international politics, primarily in normative issues. The IPU, for example, constantly puts issues like democratic best practices, human rights, and women’s rights on its agenda (Slaughter 2004). Flockhart (2004) reports about the NATO Parliamentary Assembly’s ‘Rose-Roth seminars,’ which were used to familiarise parliamentarians from former socialist countries with ‘Western’ norms after 1989. Raube et al. (2019) assemble various studies on parliamentary cooperation and diplomacy, especially within the European Parliament and EU’s external relations. Soft power analyses often run on the bilateral level under the term ‘parliamentary diplomacy,’ which is defined as an autonomous international activity between parliaments or direct interaction between legislators across national borders (Stavridis and Jančić 2016; Pigman 2010). Wang (2016), for example, describes the parliamentary diplomacy of the Chinese National People’s Congress as part of the overall diplomatic policy of China to export its norms and interests. An overview of this multitude of single cases highlights that domestic and international parliaments have become very active in shaping international policies in many fields.

Another policy view on parliamentary exchange highlights the role of information instead of norm socialisation. Here, Ringe and Victor (2013) focus on interlegislator networks and spread policy information. Thompson (2006) reports how IOs can spread information about military interventions to parliamentarians to foster support or resistance. These examples show that parliaments can be part of the agency that carries and transmits international (policy) norms and practices.

A structural perspective that explores the variety of transnational institutions that provide MPs and national parliaments channels of international interaction is missing in the existing literature. Despite expanding research, we do not have systematic knowledge about larger patterns and dynamics of international parliamentary organisation and their interaction, focusing on bilateral and multilateral exchanges. We argue that these conceptually and empirically isolated approaches can gain analytical strength by jointly exploring international parliamentary relations of IPIs and IPFGs. Measuring exchange structures promises to deliver a first necessary step for a more systematic understanding of flows of legislative policy information and norm diffusion, for example, between parliamentarians and IOs. In addition, this literature can also benefit from a bottom-up perspective that explores how national parliaments and MPs position in the emerging relations and that the ultimate power resource of these formally largely powerless institutions emerges from the institutional relations they generate for national parliaments (see similarly Šabič 2013: 27–28).

Conceptualising legislative communities

We understand legislative communities as emergent institutional phenomena. The constituent parts are national parliaments that establish connections amongst themselves. Connections can be either direct bilateral exchange or co-participation in multilateral parliamentary institutions (IPIs). Legislative communities are the most densely connected parts within this multiplex network, constituting clusters (or subgroups) of national parliaments within the global network of transnational parliamentary relations (see Malang and Leifeld 2021).

Transnational legislative communities unfold as national parliaments face an increasingly complex international environment in which they need to fulfil their parliamentary functions like passing legislation or control governments (Kinski 2020; Kinski and Crum 2020). Experience and information from peer legislatures become thus increasingly important for individual chambers (Malang 2019). As an effect, parliaments have been turning their parliamentary unorganised (international) environments more and more into organisations based on parliamentary principles by creating meta-organisations of different organisational models beyond the national stage (see Ahrne and Brunsson 2005: 447, 2008: 62–91).

We conceptualise legislative communities from the perspective of the national parliament. Each national parliament can choose to develop international contacts and which of the two ideal types of transnational parliamentary organisations (IPIs and IPFGs) it wants to emphasise. A parliament, in our view, decides to develop its legislative community with different degrees of intensity and international integration. From this perspective, informal bi- and formal multilateral parliamentary relations offer parliaments different institutional dynamics, benefits, and costs. Consequently, the emerging structure, institutional composition, and regional density of legislative communities can vary.

Legislative communities build on two major forms of organisations: multilateral and bilateral parliamentary organisations. We understand multilateralism and bilateralism as two specific organisational forms rooted in certain rules, normative ideals, and practices of international relations. Most multilateral parliamentary organisations are International Parliamentary Institutions (IPIs),² and major bilateral parliamentary organisations are International Parliamentary Friendship Groups (IPFGs).³

IPIs enable enduring interaction among three or more national parliaments based on non-hierarchical, collegial, and parliamentary principles and are composed of either directly or indirectly elected individuals (usually elected by and members of national parliaments).⁴ Due to their multilateral nature, diffuse or undirected reciprocity characterises IPIs’ parliamentary work, which is based on and generates an indefinite, sequential exchange pattern grounded on a generalised principle of conduct (Ruggie 1992: 571; Keohane 1986). This cooperation pattern among a relatively large number of actors induces comparatively high transaction costs that demand relatively strong institutionalisation levels, especially for information exchange (Keohane 1984, 1982; Martin 1993). Thus, IPIs often materialise as formal organisations with (at least some form of) secretariat, assembly, and founding treaty.

In contrast, IPFGs⁵ are bilateral parliamentary organisations that facilitate enduring cooperation by groups of MPs between two national parliaments.⁶ Contrary to multilateral parliamentary organisations, chambers in bilateral structures explicitly choose target peer legislatures and are comparatively weakly institutionalised, which usually materialises in informal cooperation patterns without secretariats and founding treaties (Vabulas and Snidal 2013; Roger 2020). Specific or directed reciprocity characterises this relationship between two chambers—if one chamber selects a given country as a target, the target chamber’s cooperative (re)actions are likely but not necessary (Axelrod 1984; Keohane 1986). Table 1 summarises our conceptualisation.

Table 1 Multi- and bilateral forms of transnational parliamentary organising

IPIs have been an increasingly common feature in international relations since World War II and draw increasing scholarly attention (for an overview, see Costa et al. 2013a; Schimmelfennig et al. 2020). Today, at least 60 IPIs (Šabič 2008; Kissling 2011) spread throughout all global regions and engage in diverse policy fields. Many IPIs are part of international governmental organisations, but numerous cases exist that form an international parliamentary organisation on their own, without intergovernmental elements.⁷ Their political purpose generally extends from information exchange and promoting international cooperation to the (co-)development of international policies with their intergovernmental counterparts.

IPFGs arose likewise after World War II in democratic parliaments and have been promoted since the 1970s by the Inter-Parliamentary Union (IPU) as a tool of parliamentary diplomacy (Inter-Parliamentary Union 2005: 8). Unlike IPIs, bilateral friendship groups are not permanent. Rather, MPs decide which partner countries they want to set up or maintain groups at the beginning of each domestic legislative term. Hence, the emergent groups are based on a specific interest of MPs, parties, or chambers in a given target country (Fiott 2011; Wang 2016). The purpose of these bilateral exchanges reads almost identically in most parliaments: promoting ongoing dialogue with the parliamentary institutions of a partner state and enhancing mutual trust via personal relationships and shared practices. Often mentioned practices involve exchanging information and opinions and organising meetings with the partner country’s government and civil society representatives. Furthermore, the promotion of democratic parliamentary structures, strengthening human rights, contributions to the management of crises, and the parliamentary scrutiny of the government’s foreign policy are additional widely shared enterprises (see Inter-Parliamentary Union 2005).

Theorising parliamentary involvement in international politics

We argue that clusters of legislative communities emerge through interactions of national parliaments’ membership in IPIs and IPFGs. Both forms of organising transnational parliamentary relations are conceptually independent but empirically interact when forming legislative communities. We explore such emerging legislative communities on two different levels of transnational parliamentary relations: (1) frequency of participation and (2) the clustering of resulting connections. The first level denotes each national parliament’s general international activity as realised in memberships in multi- and bilateral organisations (IPIs and IPFGs). The second level concerns each parliament’s realised interaction with peer legislatures via these international channels as a relational phenomenon. Whereas the first aspect is one of frequency (How much activity does a given chamber realise?), the second aspect is relational (With which other chambers does a given chamber exchange?).

We analyse the frequency dimension with formal hypotheses and a quantitative ‘test’ set-up. As the relational aspect emphasises the qualitative dimension of relations, we employ an explorative network analysis in the second step after the quantitative analysis. Therefore, in the remainder of this section, we develop hypotheses about national parliaments’ international activities and specifically about the frequency of their IPI and IPFG memberships.

Hypothesising frequency of international parliamentary activities

How can we meaningfully describe and explore the variation in international activities of national parliaments? First, we ought to know the depth of structural integration measured by the number of IPIs and IPFGs membership each national parliament realises.

Country and parliamentary characteristics that indicate formal capacities of national parliaments form a natural starting point for the theory-guided explanation for membership in these institutions. If we interpret parliamentary activity analogous to state activity (realised in diplomatic ties and IGO involvement), we could follow that classic IR dimensions like military and economic power affect the frequency of international involvement (Neumayer 2008; Kinne 2014). More powerful countries normally comprise more institutional capabilities to be active on the international stage. Also, international representation is particularly important for promoting trade interests (James 1980).⁸ We thus hypothesise that the more powerful and economically strong a country is, the higher the frequency of its national parliament’s international involvement.

Second, we assume that internal and formal parliamentary capacities condition a legislature’s position in transnational institutional integration. The concrete practices of parliamentary activities beyond the nation-state boil down to certain MPs’ abilities and willingness to participate in bi- and multilateral meetings. Transnational parliamentary activities take time and energy. On an aggregate level, we expect national parliaments with large numbers of MPs in their ranks to have higher abilities staffing national delegation to IPIs and IPFGs more sufficiently. Thus, these parliaments have a higher propensity to be more integrated into legislative communities as they are more apt to distribute transnational tasks more efficiently among the remaining legislative workload.

Third, we hypothesise that formally stronger parliaments (vis-à-vis the government) play a more active role and thus have more realised international contacts than weaker chambers. We assume all parliaments, by virtue of their nature, to have vested interests in executive control. However, formally strong parliaments comprise more abilities to translate these incentives into actual executive control than weaker parliaments. High levels of international activity pay off more for formally strong parliaments than weaker ones. As international activities are costly, we expect stronger parliaments to be more internationally active than others.

A control dimension that we include in our regressions is the type of political regime and political system. Parliamentary roles in autocracies and democracies are different, which could translate into different international activities. The same difference can be assumed for the type of political system, where parliaments fulfil different functions in presidential, semi-presidential, and parliamentary systems (Shugart 2005).

Hypothesising frequency of IPI and IPFG membership

Beyond the sheer frequency of international activity, our concept of legislative communities needs to explain which specific channels of international activity parliaments choose. Thus, we differentiate our arguments between IPI and IPFG membership and focus on external, institutional supply instead of internal, parliamentary capacity.

The reality of multilateral IPIs is a regional one. Although global IPIs such as the prominent IPU exist, they are outlier cases, as all IPIs in our data set have a median of 14 country memberships. Thus, a region’s multilateral institutional supply strongly conditions the chance of becoming a member in multiple IPIs. For example, the parliament from Costa Rica simply cannot participate in the highly institutionalised European environment with IPIs like the NATO or OSCE PA. If a region is generally weakly institutionalised with multilateral IOs, the propensity of being part of adjacent IPIs or establishing stand-alone IPIs is likewise lower. Thus, we expect to find variation in the depth of institutional integration across global regions. We take world regions as a proxy for multilateral institutional density (Börzel and Risse 2016) that generally reflects the supply of multilateral IPIs, and hypothesise that the region has a significant effect on the realisation of the number of IPI memberships of a given parliament.

In contrast, regions are less likely to constrain national legislatures’ propensity for establishing bilateral friendship groups. This is because the above theorised weak levels of institutionalisation in bilateral parliamentary organisations are more apt to facilitate global, transregional cooperation, potentially extending a parliament’s interaction beyond its (strongly) institutionalised multilateral region. Hence, we hypothesise that the world region should not significantly affect the number of realised IPFGs over all parliaments. Suppose the supply of IPIs within regional contexts proves to be the demarcation line between the multilateral and bilateral logic of international activity. We hypothesise more generally that country and chamber attributes are better predictors for IPFGs than IPIs.

Beyond the regional scope that might influence the establishment of legislative communities, we want to explore the relationship between the bilateral and the multilateral dimension of parliamentary activity. Membership in multilateral IPIs can facilitate multiple relations between two national parliaments. As theorised above, the more national legislatures share joint IPI membership, the more likely they are to institutionalise a dense legislative community. Legislative communities’ interactions facilitated by bilateral friendship groups follow a different network logic. They do not facilitate multiple but single relationships and can run parallel to joint IPI membership. Generally, two different relationships are thinkable. On the one hand, the number of IPI involvement is positively correlated with the number of IPFGs, because IPFGs are an additional dimension of a strong legislative community. On the other hand, IPFGs could be interpreted as a substitute for parliaments with low IPI involvement, which could be due to the low supply of IPIs in a given region. Since both explanations are conceivable, we refrain from formulating a hypothesis with a direction but simply assume that the number of IPIs and IPFGs should be correlated.

Data

In order to capture legislative activity (the dependent variable), we code formal membership of parliaments in multilateral IPIs and bilateral connections through IPFGs for all national parliaments where data was available in 2017.⁹ The unit of observation in our study is the national parliamentary chamber. We only code chambers of independent states and exclude overseas territories, dependencies, and the like. In the case of bicameral systems, we take the lower chambers. We focus on the lower house as it is commonly the legislatively predominant chamber and emphasises these transnational communities’ (national) legislative dimensions. However, this limitation often has less severe consequences for the case selection as it might seem from the outset. Many bilateral parliamentary delegations in bicameral systems combine members from both houses, and membership in IPIs can be likewise mixed. The number of IPIs that exclusively accept membership from upper houses is small.¹⁰ Finally, we collected crucial information about IPI memberships and friendship groups for 144 national chambers.

For multilateral IPI memberships, we coded the cumulated memberships, which is the sum of all memberships per country in IPIs with a formal connection to a larger parent IO and stand-alone IPIs without any such connection.¹¹We exclude IPIs with directly elected MPs since they are not members of national parliaments. To be sure, this omission does not neglect that directly elected IPIs and their MPs often facilitated transnational relations and can be part of wider parliamentary networks. However, they do not drive national parliaments’ transnational parliamentary relations as understood in our conceptual framework. The European Parliament, for example, is a special case in the global population of IPIs (Cofelice and Stavridis 2014) and facilitates increasingly important parliamentary cooperation with national parliaments in the EU (Lupo and Fasone 2016; Griglio and Stavridis 2018; Gianniti and Lupo 2016) and with regions beyond Europe (Raube et al. 2019). However, from our vantage point, these relations are located on a different institutional level than transnational cooperation among national parliaments.

For bilateral friendship groups, we coded every group that one parliament has to another country. A difference between the two types of relationships is the directionality of the data. Whereas IPI memberships are, per definition, undirected, IPFG membership is directed. This means that when China and Cuba are part of the same IPI, they have one undirected connection. However, China can have an IPFG for Cuba, whereas Cuba might not have one for China. This difference is neglectable for the statistical analysis, but we can see the directionality of the IPFGs in our final exploration in Fig. 4. The violin plots in Fig. 1 show the distribution of both data. Almost all national parliaments are part of at least two IPIs, and the maximum is twelve IPI memberships (realised by the parliaments of Albania and Greece). On average, a parliament is part of six IPIs. Furthermore, the distribution is quite normal, with most legislatures realising between five and eight IPI memberships. With regards to bilateral parliamentary relations, the median legislature realises 27 groups. However, every fifth parliament does not have bilateral groups at all, including some prominent examples like the US Congress. The legislature that realises most groups is the French Assemblée Nationale (118 out of 144), followed by the Grand National Assembly of Turkey (111 groups).

Fig. 1

Distribution in IPI and IPFG Membership

We use standard IR and political system variables to test if the country and parliamentary factors affect the frequency of parliamentary activity. For grasping power, we use the Correlates of War ‘Military Capabilities Index’ and the population size (COW version 5, originally Singer 1988). For economic strength, we take the GDP from the World Bank indicators. Chamber attributes are operationalised by the size of a parliament taken from the IPU’s ‘Parline’ database (Inter-Parliamentary Union 2015) and Fish and Kroenig’s (2009) Parliamentary Powers Index Scores to capture its institutional strength. Furthermore, we use the Polity IV measure to control for the regime type in which a parliament resides, and we use COW data on formal memberships in intergovernmental organisations to control for the general international embeddedness of a nation state. In addition, we understand each legislature’s regional affiliation as defined by United Nations Statistical Division (UNSD 2019).

Results: legislative communities in international parliamentary relations

This section describes the three legs of our exploration strategy. Firstly, we focus on individual chamber frequency and describe the distribution of participation in bi- and multilateral international parliamentary organisations. Secondly, we try to explain the frequency of participation by country and chamber attributes. And thirdly, we explore variation in relations that enable the creation of diverse (formal) legislative communities.

International integration of national parliaments

Figure 2 shows that national parliaments vary strongly in their level of international integration facilitated by bi- and multilateral international parliamentary organisations. The x-axis sorts the number of bilateral IPFGs, whereas the number of multilateral IPI memberships is shown on the y-axis. The median of each dimension’s distribution separates this two-dimensional space into four levels of international parliamentary integration.

Fig. 2

Bi- and multilateral dimensions of national parliaments’ international integration. Node shape indicates world region (according to UNSD 2019)

The lower left corner assembles legislatures below both dimensions’ median and thus the least internationally integrated cases. We hardly see a common pattern of shared characteristics, rather a plethora of parliaments with almost no IPFGs and some IPI memberships. Each global region hosts some parliaments that are integrated below average. The list includes parliaments of larger size such as Japan, India, and the US Congress and smaller parliaments such as New Zealand, Paraguay, and Burundi. Hence, there appear to be some decisions that affect low international integration that we cannot explain for the time being or path-dependencies that deserve further qualitative case studies.

The lower right quadrant indicates high numbers of friendship groups but lower involvement in multilateral IPIs. We understand this reliance on bilateral friendship groups as an international practice with low formalisation. Here, we can find more well-defined patterns. Two regions dominate: (South and East) Asian and South American legislatures. Concerning (South and East) Asia, the low multilateral involvement corresponds with findings from comparative and regional integration studies generally identifying this region to host weakly institutionalised (governmental) multilateralism (Börzel and Risse 2016: 629). Contrary, South America shows lower levels of formal multilateral integration than the global and other regions’ averages. This is striking since this region generally exhibits even higher institutionalised regional integration than its actual (economic) interdependencies (Börzel and Risse 2016: 629).

Parliaments with the converse pattern—high participation in multilateral IPIs and low level of friendship groups, hence strong formalisation of contacts—can be found in the upper left corner. The cluster consists of 30 chambers, and exactly half of them are from Africa. The other 15 are from Europe (mainly northern Europe) and Asia. There is no chamber from the Americas in this cluster (and overall, only one [the Canadian] chamber in the upper half of the graph). Here we can observe the institutional ‘supply’ effect: there are not many American IPIs. Hence American chambers hardly can reach the global mean of IPI membership frequency.

Lastly, European parliaments dominate the upper right quadrant and are most strongly internationally integrated on both institutional dimensions (23 European chambers of 33 in this quadrant). This high integration level reflects most European parliaments’ high international parliamentary activities and the highly institutionalised multilateral environment present in greater Europe as defined by membership in the Council of Europe and OSCE (see Börzel and Risse 2016: 628–29; Schimmelfennig 2016). Notable extreme cases are the French Assemblée Nationale, the Grand National Assembly of Turkey, and the Hellenic Parliament.¹²

Explaining frequency of participation

The general mapping of all chambers in four quadrants already foreshadowed some potential explanations for the frequency of parliamentary activities. Now we undertake a more systematic test of potential explanatory factors. We have two dependent variables, the number of IPFG and IPI memberships per chamber, which translates into two separate models in Table 2. As noted, we first want to test explanatory factors that should work for both dimensions of international parliamentary activity then turn to the differences and interdependence between the two dimensions. We use a standard regression with the national chamber as the unit of analysis. All standard tests and the respectable R squared give no reason to assume model misspecification.¹³

Table 2 Regression results

Our first block of explanatory variables hosts standard IR factors that explain the general international activity of countries. However, neither military capabilities, population size, nor economic strength has a significant unified effect on the frequency of parliamentary activity. Only military power is negatively related to the number of IPFGs (model 1). However, this effect diminishes in the robustness check (Appendix A.2). National parliaments seem not to mimic government behaviour in international politics.

The second explanation put forward focuses on the capacity of chambers. We hypothesise that larger chambers equipped with more power in the national political system have a higher frequency of international activity. The highly significant results for the IPFGs support this claim. However, we do not find a similar effect for the multilateral IPIs in model 2. Going back to Fig. 2, a possible explanation trend is that smaller parliaments use the institutional capacities that IPIs provide to make their international appearances as efficient as possible. For example, some smaller African parliaments have only a few friendship groups (but not none), which indicates that they generally try to establish international contacts. However, since their capacity is low, they instead opt for the multilateral way.

Turning to the control factors for political regimes, we can see that more democratic regimes realise fewer IPFGs (the negative Polity IV factor). However, the regime type is once again insignificant for IPIs. As an illustration, we want to highlight an autocratic regime cluster in Fig. 2 that consists of China, Cuba, Pakistan, and Iran in that sector. Regime type might be uncorrelated to participation in multilateral IPIs, where parliamentarians are likely to interact with democratic MPs—a scenario an autocratic leader might want to prevent. In contrast, the bilateral, specific reciprocity enables autocratic regimes to select partners and only allows for exchanges with like-minded regimes. Anecdotal evidence from Cuba and China that try to pursue the goals of their leaders in African dictatorships illuminates this possible pattern underlying our statistical results.¹⁴ Additionally, we can see that parliaments in presidential systems are more active in IPFGs than parliaments from parliamentary systems.

We now turn to the differences between IPIs and IPFGs as well as their interrelationships. Regarding the frequency of IPI and IPFG membership, we argued that IPI supply should influence the participation of national chambers. As a proxy, we include world regions in the regression table and investigate their effect on IPFG and IPI membership. Most IPIs (75 percent) are geographically speaking regional. Hence, MP’s participation is restricted to chambers from that given region. Models 1 and 2 show the results with Europe as the reference region for the interpretation. The results of model 2 corroborate our descriptive hints. All world regions participate significantly less likely in IPIs than European chambers (model 2), where most IPI exist. This effect is not so nuanced for IPFGs (model 1), where parliaments from Oceania are no different from Europe. Especially when we control for IGO membership (model 3), the effect diminishes completely for Asia and the Americas in the case of IPFGs, whereas it stays robust for IPIs (model 4).

To understand the relationship between IPI and IPFG participation, we also include the number of either bi- or multilateral participation in the respective equation in models 3 and 4. This helps us understand whether IPIs and IPFGs reinforce strong legislative communities or whether these two dimensions are considered (exclusive) alternatives or substitutes. We can see that the two dimensions are positively correlated. Hence more IPI membership also comes with more IPFGs. Additionally, we see that IPFGs are not dependent on the general multilateral endowment of states as measured by the number of IGO participation. Again, this variable is only positively and significantly correlated with the frequency of IPI memberships.

Exploring clusters of legislative communities

After the frequency of parliamentary participation in international encounters, we turn to ask how parliaments formally connect with each other. Are there communities of strong parliamentary cooperation facilitated by shared bi- and multilateral organisations that we define as legislative communities? To detect these areas, we shift from the simple frequency of international activity to the connections between parliaments via bi- and multilateral organisations. Therefore we take the networks of formal inter-parliamentary cooperation facilitated by joint membership in multilateral IPIs. Here, we increase the threshold for joint memberships (the strength of ties between two respective parliaments) to a level that our overall network decomposes into distinct components. Afterward, we describe how these isolated components of the multilateral network connect via bilateral friendship groups.

For the next step of visual exploration of the legislative communities, we plot a sub-graph of the global network with six or more joint IPI membership¹⁵ to zoom in on clusters with high cooperation intensity and a greater likelihood of enabling practiced legislative communities described above.¹⁶ This top-down approach focuses on the whole network and looks for ‘sub-structures’ as locally denser parts than the field as a whole. In a sense, this rather macro lens is looking for ‘holes’ in the overall structure of our network. These holes define lines of division or cleavages in the parliamentary network and point to how they might be decomposed into smaller units. This top-down perspective leads us to think of dynamics that operate at the group selection level and focus on the constraints under which actors construct networks. The institutional supply of IPIs is such a constraint. In contrast, bottom-up approaches would focus on the individual chamber and ask for micro-logics of partner selection as we did in the first regression analysis (see Wasserman and Faust 1994).

Figure 3 shows five isolated clusters for the relation ‘more than six shared IPI memberships.’ All these clusters align with global regions: Central Asia, two separated clusters in Sub-Sahara Africa, one North African cluster extending into West Asia, and one large European cluster stretching from Scandinavia to the Black and Caspian Sea. Node colours indicate betweenness centrality: how often a parliament rests on the shortest path between any two other parliaments. Parliaments with high betweenness centrality (darker nodes) are assumed to bridge otherwise unconnected pairs of parliaments or sub-clusters, thereby controlling the information flow between two unconnected chambers (Hafner-Burton et al. 2009: 564; Scott 2012: 42).

Fig. 3

Clusters of above-average joint IPI membership. Node shape indicates world regions (according to UNSD 2019). Node color indicates betweenness centrality

Based on our theoretical framework, we understand each networked cluster to illustrate a legislative community’s (formal) multilateral dimension in today’s international parliamentary relations. We already showed that spatial contiguity to be most likely to form a cluster of shared multilateral parliamentary organisations and, thusly, create thick legislative communities. However, beyond the ‘regional dummy’ in our regression, we highlight the concrete relations and regional nuances of clusters in the following.

The small Central Asian cluster centres on the triangle between Tajikistan, Kirgizstan, and Kazakhstan and extends to Russia and Armenia. At the centre of this cluster are IPIs of regional international organisations in the post-Soviet space. This includes particularly the Parliamentary Assembly of the Collective Security Treaty Organization (PA-OCST) and the Interparliamentary Assembly of Member Nations of the Commonwealth of Independent States (IPA-CIS). The triangle is additionally shaped by the three parliaments’ membership in the Parliamentary Union of Organization of Islamic Cooperation (PUIC), thus excluding Russia and Armenia, which are a member of the Interparliamentary Assembly on Orthodoxy (IAO). Therefore, this cluster expresses some remaining geopolitical, historical, and cultural legacies. Russia is particularly using and promoting these international parliamentary organisations as a vehicle for its foreign policy. It aims to protect the existing regimes via developing common positions on international issues, establishing its own (alternative) cooperation and legitimation narratives mostly vis-à-vis (Western) European models, and harmonise domestic legislation as a tool of soft power (see Petrova 2019).

The first African legislative community is set up exclusively by French-speaking West African parliaments. These are primarily connected via the Assemblée parlementaire de la Francophonie and related IPIs. The focus here is on normative issues like women’s rights and cultural issues, especially the francophone heritage. We can see that this is a strongly closed legislative community, where every member is connected with every other member via the IPIs but to no other cluster.

In contrast, the other African clusters have several subclusters. On the one hand, we can see that the North-African parliaments from Tunisia, Morocco, and Algeria form a completely connected clique with shared similarities in religion and political regime. They are connected through shared memberships in African IPIs and through the IPIs of the Mediterranean dialogue, which bring together MPs from Southern Europe and North Africa to solve regional problems like migration or water pollution. The parliament from Mauretania is the most connected and bridges the North-African clique and the other three chambers of least developed countries, Djibouti, Somalia, and Sudan. Jordan only becomes a member of the larger community via Lebanon.

The large European cluster is generally connected via many task-specific organisations like the NATO-PA or the OSCE-PA and many more subregional IPIs like the Baltic Sea Parliamentary Conference. We can see that Greece’s Hellenic Parliament plays a vital role as a cut point connecting West-European and (South) Eastern European parliaments (as the dark shade indicates). The Polish Sejm is essential in bridging the otherwise weakly connected clusters of the Northern European and Baltic states with the Central, Eastern, and Southern European clusters.

If we add the bilateral connections on top of these multilateral clusters, we get a multiplex network consisting of strong multilateral and binary bilateral relations. We visualise the communities as follows in Fig. 4. The thick black edges plot ties only based on shared IPI membership (above the threshold of six common memberships, similar to Fig. 3); thin grey edges show bilateral connection via friendship groups. If two parliaments share six or more IPI membership and a bilateral friendship group, we only plot the friendship group to highlight multilateral and bilateral clusters separately. As an example, take Kirgizstan and Kazakhstan on the very upper left of Fig. 4. We know from Fig. 3 that they are part of a multilateral legislative community (they share more than six IPI memberships). We can see that Kirgizstan established an additional bilateral friendship group, visualised through a directed grey tie that substitutes the undirected tie from Fig. 3. Contrary, Kazakhstan does not have a bilateral relationship to Kirgizstan, which is indicated by the undirected thick black tie, which now means ‘multilateral relationship only’. Hence, whereas shared IPI memberships are undirected (now two black lines without an arrow between two parliaments), the bilateral friendship groups have a sender (the initiator of a group) and a receiver (the target of a given group, hence they are directed, as represented by arrows.

Fig. 4

Joint clusters of IPI and IPFG membership. Black lines indicate shared IPIs, grey ties indicate IPFGs

Overall, 850 of 2053 IPFG links between parliaments do not share a common IPI community (that is, there is an IPFG link between two chambers but no IPI link). Hence, the bilateral friendship groups do add a global component to the regionalised pattern of IPI clusters. We can see that the French Assemblée Nationale, the Grand National Assembly of Turkey, and to some extent, the Moroccan legislature are central actors in this combined network of dense international parliamentary relations. All three parliaments are marked by comparatively high betweenness centrality (as indicated by the dark shading). That is, each is comparatively well-equipped to provide fast connections between parliaments in separated communities. In each case, this is due to its membership in a multilateral legislative community and simultaneously numerous bilateral friendship groups reaching beyond this community to parliaments of other communities.

Three predominantly multilateral driven (sub-)communities stand out (represented in the thick black lines). The francophone African legislative community has almost no bilateral ties; the community is strongly based on multilateral exchange. This means that the organisational capacity of IPIs is sought after and provides a powerful tool to structure the agenda of the legislative community. Besides, the African cluster of the least developed countries does not rely on bilateral exchange. We take this again as evidence that national legislators need some level of resource endowment before engaging in bilateral relationships, which they must organise and pay on their own. However, the third exclusive multilateral cluster immediately contradicts the resource thesis. As we can see, also the wealthy north-European parliaments, together with the Baltic legislatures, create a strong multilateral community with almost no bilateral exchange.

For the larger European legislative community and the Central-Asian cluster with Russia, we find many connections on a bilateral basis. These communities seem to attribute an additional value to the specific reciprocity of bilateral relations. We highlight this potential value-added through some examples for some of the most prominent chambers.

As a first example, we can see that the UK adds to all their multilateral ties also a bilateral relation through IPFGs (they call them All Party Parliamentary Groups). One of the values attributed to these groups is to share democratic practices and the specific tradition of the Westminster system. For example, during the heated Brexit debate, Speaker John Bercow explicitly addressed this function. When he called for ‘order,’ he mentioned ‘to take account of the fact that we are visited by a distinguished group of Lebanese parliamentarians, at the invitation of the Inter-Parliamentary Union and the All-Party Parliamentary Group on Lebanon […]. We would like to set them a good example; I am not sure at the moment how impressed they will be’ (House of Commons 2019: 172).

The Turkish parliament serves as an example of the political dimension of target selection. Turkey is generally very active on the bilateral dimension (only France has more groups). However, they choose their targets on purpose. If we compare Figs. 3 and 4, we can see that Turkey complements virtually every multilateral relation with a bilateral one. However, they do not do this for Greece mainly because of the hostility which centres around Cyprus. In addition, there are no IPFGs for Armenia and Israel (with whom they do not share strong IPI overlap). Turkey seems to represent either a normative or a realistic perspective on group selection. We can see that the bilateral specific reciprocity does not follow a functional necessity like the IPI memberships. Instead, parliaments are free to dismiss targets whom they ‘do not like as friends’. Strikingly, after tensions between Turkey and the Netherlands ahead of the Turkish presidential referendum emerged in 2017, the Turkish parliament did not re-initiate their IPFG with the Netherlands in the new parliamentary term 2018 (Anadolu Agency 2018).

Azerbaijan serves as the last example of how bilateral contacts could be used to gain direct influence in the international sphere. Again, Azerbaijan is very active on the bilateral dimension, whereas they only have one strong community tie with Turkey multilaterally. As recent investigations under the headline ‘caviar diplomacy’ revealed, they used both legislative communities as a direct way to bribe parliamentarians from other states in their favour (Rankin 2017). German MP Karin Strenz, for example, was linked to Azerbaijan both through the multilateral PACE and the bilateral IPFG. In this function, she maintained stable relations with politicians of the autocratic regime and received monetary donations (Deutsche Welle 2019). Additionally, she was part of the electoral observatory mission to Azerbaijan. Hence, the exchanges could also be very material and serve some concrete interests.

In sum, multilateral parliamentary organisations such as IPIs are prime institutional conduits for generating clusters of strong regional legislative communities. In contrast, bilateral parliamentary organisations such as friendship groups are more suitable in facilitating communities beyond regional horizons.

Discussion

What does our concept of legislative communities and our explorative findings add to understanding general patterns and dynamics of transnational parliamentary relations? Regarding the literature on international parliamentary relations, our findings call for not treating IPIs and other features of parliamentary diplomacy in isolation (see Cofelice 2016; Costa et al. 2013a; Wagner 2013; Crum and Fossum 2013b). Rather, we claim it is more insightful to understand these as pillars of larger dynamics in organised (parliamentary) world politics. We embed international parliamentary organisation in the wider structures of transnational parliamentary relations, in contrast to approaches aiming to explain specific phenomena of formal, multilateral international parliamentary organisations (see recently Rocabert et al. 2019; Lenz et al. 2019).

The presented network approach grasps structural and relational patterns of international parliamentary activity that proved fruitful in related fields such as human rights INGOs (Murdie and Davis 2012) and peace research on IGOs (Lupu and Greenhill 2017; Hafner-Burton and Montgomery 2006). Building on this top-down, structural approach, scholars interested in (single) case studies might find it fruitful to explore international relations of the most active chambers. Insightful examples from our analysis can prove to be the French Assemblée Nationale, the Grand National Assembly of Turkey, the Moroccan parliament, Greece’s Hellenic parliament, and the Polish Sejm.

This further research can also extend the remaining limits of our approach. Due to the global empirical scope of our analysis, we remain focused on the formal and partially informal structures of transnational parliamentary relations. However, we cannot explore the actual intensity or the degree of practiced participation in the various legislative communities. The existing research highlights the variations in MPs’ attendance rates in IPIs (Wagner 2013; Lipps 2021). Our framework can help future research to zoom in on legislative communities at the level of MPs and explore how transnational parliamentary relations shape domestic legislative work and vice versa. This research can also incorporate other institutional branches of parliamentary cooperation, for example, in the multiplex realm of the European Union or other parliamentary initiatives (see Schade and Stavridis forthcoming).

The existing International Relations literature provides ample opportunities to apply our (such enriched) concept of legislative communities. Firstly, membership in IOs can positively affect the democratisation of (already) transitioning countries (Moravcsik 2000; Pevehouse 2005; Pevehouse and Russett 2006; Poast and Urpelainen 2015). From the perspective of international parliamentary relations, this literature can be enriched by asking if an important part of this democratisation dynamic might also stem from the support of IPIs and bilateral exchanges for non-democratic, transitioning parliaments. Previous research found some anecdotal examples, for instance, NATO parliamentarians socialising newly minted member states’ parliamentarians into new institutional set-ups (Flockhart 2004; Selden and Oehman 2019). And indeed, it is legislatures that most vividly embody democratic representation and change, normatively and factually speaking, and we should thus expect them and their international cooperation to have a substantial influence on regime transitions. In the same vein, further interesting questions can explore how autocratic regimes (try to) use international parliamentary relations to curb such democratisation efforts (Debre 2021).

Secondly, on larger theoretical levels, our legislative community framework and findings add a transnational and parliamentary perspective to often studied inter-governmental and civil society actor organisations and institutions (Carlsnaes et al. 2013; Cogan et al. 2016). Our concept of legislative communities emphasises the relational interaction of actors as found in the networked dimensions of organisations (see McCourt 2016).

For one, in our analysis, we highlight the network facilitating dimension of IPIs and friendship groups among the spectrum of organisational capabilities. However, especially IPIs should not be reduced to pure network facilitators but also understood as genuine international organisations with degrees of autonomy (Cutler 2013; generally see Brechin and Ness 2013). Moreover, not all national parliaments are equally well-positioned to engage in international affairs—be it due to their region, internal capacities, regime environment, and many more factors. Legislative communities are communities of strong ties, dense integration, and the capacity to influence international and domestic (legislative) decision-making. Thusly, members of legislative communities find themselves in different transnational theatres, exposed to different environmental influences, and equipped with different capacities to shape international and domestic policies.

Conclusion

This study begins with observing a plethora of international activities of national parliaments and parliamentarians, largely centring on bi- and multilateral forms of organisations. Thus, we ask, what are the drivers and global or regional patterns of organising transnational parliamentary relations? And do national parliaments from different regions or with different political regimes use transnational parliamentary institutions differently? We develop a conceptual framework and derive a typology differentiating bi- and multilateral dimensions of international parliamentary organisations to understand the varying positionality and clustering of national parliaments engaging in organised international parliamentary cooperation. We argue that these bi- and multilateral parliamentary organisations cluster national parliaments with different cooperation intensities and, thus, enable the creation of diverse legislative communities. Legislative communities are transnational groups of national parliaments united in common transnational parliamentary organisations and practices.

The subsequent empirical analysis provides some explorative insights. Firstly, there is substantial variation in national parliaments using bi- and multilateral modes of organising international parliamentary relations. Secondly, the capacity of parliaments shapes their membership in IPFGs, but not IPIs, and small legislatures tend to use the institutional capacity of formal IPIs instead. In general, the external institutional conditions prove relevant for the structure of legislative communities: national parliaments’ membership in IPI and IPFG reinforces each other, the regional multilateral environment conditions IPI membership, and IPFGs can facilitate more easily cross-regional relations than formal, multilateral IPIs. Thirdly, from these conditions emerge transnational clusters of diverse national parliaments generated by both organizational modes with different cooperation intensities. This is indicative of the existing legislative communities with different degrees of institutionalisation and formalisation.

As we detail in our discussion, this theoretical approach and empirical findings provide ample opportunities to extend the existing research. Starting from this institutional repertoire is the first step to detecting the institutional vertebrae generating transnational parliamentary practices. The next steps require different research designs, additional levels of analysis, and more data across time and on informal parliamentary cooperation and practices.

Research exploring legislative communities connects the international parliamentary activity to processes of domestic politics, especially the three dimensions of voting, scrutiny, and communication. Further work should investigate at the domestic level if legislative communities directly affect behaviour in the house. Can realised legislative communities influence voting on specific bills? Can we find variation between parties, especially in relationship with the target countries of parliamentary activity? What is more, do MPs ask more control questions in the house in countries where they share a legislative community? In the German Bundestag, there seems to be a trend toward this specialization. For example, questions about Chinese human rights violations are almost exclusively asked by members from the German-Chinese parliamentary group. Additionally, relevant questions should concern if MPs use their expertise to develop their own initiatives or legislation apart from the government.

Our concept of transnational legislative communities systematises and connects the existing research on IPIs, international cooperation, and domestic politics. Our findings provide a useful framework for future studies, identify potential case studies, and embed international parliamentary relations in the larger context of international relations.

Appendix 1: Robustness of OLS

A.1: Replication of Model 3 and 4 of Table 2 of the main Text with Poisson regression (with robust standard errors)

	(1)	(2)	(3)	(4)
	Bilateral OLS	Bilateral Poisson	Multilateral OLS	Multilateral Poisson
Population log	1.50	0.07	− 0.19	− 0.03
	(2.58)	(0.10)	(0.19)	(0.03)
GDP log	0.25	0.00	− 0.03	− 0.00
	(0.53)	(0.02)	(0.04)	(0.01)
Military power	− 4.54***	− 0.23***	0.06	0.01
	(1.60)	(0.06)	(0.12)	(0.02)
Parliamentary power	57.82***	2.11**	0.03	0.01
	(20.98)	(0.78)	(1.57)	(0.22)
Size parliament	0.06***	0.00***	− 0.00*	− 0.00**
	(0.01)	(0.00)	(0.00)	(0.00)
Polity IV	− 1.51**	− 0.07***	− 0.06	− 0.01
	(0.60)	(0.02)	(0.04)	(0.01)
Monarchy	− 17.77	− 1.14	− 1.03	− 0.19
	(12.46)	(0.70)	(0.91)	(0.16)
Parliamentary	− 18.62**	− 0.62**	0.66	0.11
	(7.30)	(0.31)	(0.54)	(0.09)
Base: presidential	0.00	0.00	0.00	0.00
	(.)	(.)	(.)	(.)
Semi-presidential	− 0.16	− 0.12	0.92*	0.14*
	(7.09)	(0.28)	(0.51)	(0.08)
Other	− 7.27	− 0.23	− 2.40*	− 0.40***
	(19.04)	(0.68)	(1.36)	(0.09)
Asia	− 14.25	− 0.42	− 1.59**	− 0.23**
	(8.79)	(0.26)	(0.63)	(0.10)
Africa	− 29.48***	− 1.23***	− 0.76	− 0.09
	(8.77)	(0.36)	(0.66)	(0.12)
Base: Europe	0.00	0.00	0.00	0.00
	(.)	(.)	(.)	(.)
Americas	− 9.69	− 0.27	− 2.67***	− 0.40***
	(9.14)	(0.24)	(0.62)	(0.13)
Oceania	26.34	0.98**	− 4.11***	− 0.71***
	(18.80)	(0.47)	(1.32)	(0.08)
N IGOs	− 0.07	0.00	0.05***	0.01***
	(0.23)	(0.01)	(0.02)	(0.00)
N IPIs	2.79**	0.08**
	(1.25)	(0.04)
N IPFGs			0.01**	0.00**
			(0.01)	(0.00)
Constant	− 18.40	1.48**	6.79***	1.88***
	(21.53)	(0.67)	(1.43)	(0.23)
Observations	135	135	135	135
R 2	0.412		0.477

1. Standard errors in parentheses
2. *p < 0.1, **p < 0.05, ***p < 0.01

A.2 Stepwise introduction of predictors to check for overfitting

Regression result for bilateral part

	(1)	(2)	(3)
	Bilateral	Bilateral	Bilateral
Population log	6.68***
	(2.37)
GDP log	− 0.07
	(0.48)
Military power	− 0.56
	(1.24)
Parliamentary power		52.16***
		(19.72)
Size parliament		0.03***
		(0.01)
Polity IV		− 1.43**
		(0.60)
Monarchy			− 23.22*
			(11.90)
Parliamentary			− 8.80
			(7.23)
Presidential			0.00
			(.)
Semi-presidential			4.80
			(7.46)
Other			1.87
			(20.27)
N IGOs			0.30*
			(0.16)
N IPIs			2.06
			(1.28)
Asia	− 22.35***	− 16.27**	− 3.37
	(6.81)	(7.22)	(7.72)
Africa	− 32.41***	− 20.42***	− 29.07***
	(7.43)	(7.62)	(8.35)
Europe	0.00	0.00	0.00
	(.)	(.)	(.)
Americas	− 20.82***	− 8.66	− 13.45
	(7.59)	(7.28)	(9.61)
Oceania	2.76	11.38	17.41
	(19.54)	(18.81)	(20.22)
Constant	− 13.93	11.70	8.07
	(16.31)	(11.87)	(15.37)
Observations	135	135	138
R2	0.235	0.283	0.236

1. Standard errors in parentheses
2. *p < 0.1, **p < 0.05, ***p < 0.01

Regression results for multilateral part

	(1)	(2)	(3)
	Multilateral	Multilateral	Multilateral
Population log	0.02
	(0.17)
GDP log	0.01
	(0.03)
Military power	− 0.09
	(0.09)
Parliamentary power		2.12
		(1.44)
Size parliament		− 0.00
		(0.00)
Polity IV		− 0.06
		(0.04)
Monarchy			− 0.27
			(0.83)
Parliamentary			0.46
			(0.50)
Presidential			0.00
			(.)
Semi-presidential			0.79
			(0.51)
Other			− 2.29*
			(1.38)
N IGOs			0.03**
			(0.01)
N IPFGs			0.01
			(0.01)
Asia	− 2.82***	− 2.82***	− 2.09***
	(0.48)	(0.53)	(0.50)
Africa	− 1.55***	− 1.33**	− 0.80
	(0.53)	(0.55)	(0.60)
Europe	0.00	0.00	0.00
	(.)	(.)	(.)
Americas	− 3.25***	− 3.10***	− 2.79***
	(0.54)	(0.53)	(0.62)
Oceania	− 4.39***	− 4.24***	− 4.29***
	(1.39)	(1.37)	(1.34)
Constant	7.80***	7.55***	5.43***
	(1.16)	(0.86)	(0.94)
Observations	135	135	138
R2	0.343	0.355	0.412

1. Standard errors in parentheses
2. *p < 0.1, **p < 0.05, ***p < 0.01

A.3 Test of models for subsample of countries with at least 4 IPI-Memberships

	(1)	(2)
	Bilateral	Multilateral
Population log	− 1.30	− 0.22
	(3.27)	(0.23)
GDP log	− 0.35	− 0.05
	(0.55)	(0.04)
Military power	− 3.33*	0.13
	(1.78)	(0.13)
Parliamentary power	51.95**	− 0.08
	(21.31)	(1.54)
Size parliament	0.12***	− 0.00
	(0.02)	(0.00)
Polity IV	− 1.70***	− 0.06
	(0.63)	(0.05)
Monarchy	− 17.35	− 2.00**
	(14.34)	(1.00)
Parliamentary	− 21.60***	0.40
	(7.38)	(0.54)
Presidential	0.00	0.00
	(.)	(.)
Semi-presidential	− 0.46	0.68
	(7.07)	(0.49)
Other	− 6.36	− 2.85**
	(18.58)	(1.28)
Asia	− 14.35	− 1.00
	(8.83)	(0.62)
Africa	− 35.46***	− 1.04
	(9.07)	(0.67)
Europe	0.00	0.00
	(.)	(.)
Americas	− 12.06	− 2.85***
	(9.40)	(0.61)
Oceania	36.76**	− 3.93***
	(18.39)	(1.26)
N IGOs	− 0.07	0.05***
	(0.23)	(0.02)
N IPIs	2.47*
	(1.35)
N IPFGs		0.01*
		(0.01)
Constant	20.73	7.91***
	(27.05)	(1.75)
Observations	125	125
R 2	0.472	0.470

1. Standard errors in parentheses
2. *p < 0.1, **p < 0.05, ***p < 0.01