The Existence of Courts and Parliaments in Regional Organizations: A Case of Democratic Control?

Why do states design regional organizations with courts and parliaments? Is it indeed the case that states establish them because they expect these organs to exert some kind of democratic control over executives? Undoubtedly, this is an important question given that politicians and political scientists alike regularly lament the lack of democratic control of many international organizations. We tackle this question empirically. Based on an original data set of 72 regional organizations and by using simple logistic and ordinal logistic regression analyses, this article tests for the association between domestic regime type and the existence of regional courts and parliaments. These organs were selected because they are associated with dimensions of democracy, namely constitutionality and inclusiveness. The most consistent correlates of the existence of each of these institutional bodies and the aggregate of them are functional ones: policy scope, trade-related variables, and conflict-related variables. There is no significant association between any measure of democracy and the existence of these institutions. These results are discussed the context of debates about the democratic deficit of international and regional organizations and the question of whether democratic standards are applicable to regional organizations.


Introduction
In the summer of 2019, European citizens were presented with a choice in the European parliamentary elections: For the second time since the existence of the European Union parliament, they voted not only on political parties but the political party leaders were also presented as candidates for the position of President of the EU Commission. The expectation among domestic publics was that the outcome of the election procedure would also determine who would preside over the EU Commission as one of the most influential decision-making organs in the EU. Although there was a clear winner, that person did not become the President of the Commission. In fact, none of the public contenders did; instead, the heads of governments decided to select a completely different individual. The European Parliament condemned the selection as "an undemocratic stitch-up by national governments" (Stone 2019). The social media platform "Debating Europe" suggested that instead of an open election for a leadership position, Europe was returning to "horsetrading in back rooms" (Debating Europe 2019). Another news outlet described the process as "hypocrisy" (Taylor 2019). The discussion surrounding the selection procedures demonstrates quite vividly that relevant publics evaluate the EU by standards of democratic control similar to those that exist in member states.
The EU is certainly not alone in this regard. Alexandru Grigorescu argues that there is an increased expectation that intergovernmental organizations (IGOs) adopt decisions "in a democratic fashion" and that the term "democratic deficit" by now expresses the "lack of democratic mechanisms in almost every existing IGO (Grigorescu 2015: 4). Many international and regional organizations have addressed this challenge by establishing organs that are usually associated with dimensions of democracy (inclusion and the rule of law), regional parliaments, and courts (Tallberg et al. 2014;Heldt and Schmidtke 2019). In 2018, altogether 28 regional organizations (ROs) had established a parliament, either in the form of a supranational parliament with delegated members or an intergovernmental parliament, while more than half of them had established a regional court, many of them having jurisdiction over human rights.
The observation that international and regional organizations are routinely measured against the standards of democratic polities is puzzling for two reasons. First, on the descriptive level of the universe of ROs, the expectation that the existence of parliaments and courts follows a liberal logic of the externalization of democratic standards poses a puzzle: Even regional groups that do not have democratic membership have established such organs, such as the Shanghai Cooperation Organization. Second, a thorough review of the literature on the democratic quality of ROs demonstrates that the externalization logic is itself highly contested. There is no consensus as to whether the existence of these two organs actually increases the democratic quality of ROs. More importantly, it is debatable whether domestic standards of democratic control can be applied to international and regional organizations at all. This article tackles these two puzzles empirically. What explains the existence of parliaments and courts as prima facie organs of democratic control? Does the existence of organs such as parliaments and courts follow a democratic externalization logic, or is it better explained by other factors? The democratic externalization thesis holds that member states apply domestic standards of democratic control when they design ROs. Hence, it is likelier that ROs with democratic membership feature organs of democratic control. Based on an original dataset of 72 ROs and by using simple logistic and ordinal logistic regression analyses, this article systematically tests the democratic externalization thesis against two other explanations derived from democratic lock-in and functional demand theories of the design of international institutions.
The article makes several original contributions to the existing literature: First, based on one of the most comprehensive datasets on ROs, the Comparative Regional Organization Project (CROP), the article provides descriptive statistics regarding the growth of organs associated with the democratic control of executives, in particular courts and parliaments, for the full set of ROs. It thus goes beyond existing studies, which worked with about half of the organizations in our dataset (Lenz et al. 2019;Rocabert and Schimmelfennig 2018). Second, the article provides the first comprehensive test of alternative explanations for the existence of courts and parliaments in ROs. It thus contributes more generally to the literature on the design of international institutions (Koremenos 2007(Koremenos , 2008Hooghe and Marks 2015). Across all cases of ROs, the most consistent correlates of the existence of each of these institutional bodies and the aggregate of the two are functional variables related to the policy competences delegated to an organization, trade-related variables, and conflict-related variables. Surprisingly, there is no significant association between any measure of democracy and the existence of these organs. Given the absence of significant associations between any indicator of democracy and parliaments and courts in ROs, concerns about the democratic quality of such organizations do not appear to be a major driver for establishing parliaments and courts within ROs on the global level. ROs follow international organization-like logic, not state-like logic, so applying the democratic standards of states to international organizations (IOs) is a category mistake. 1 The remainder of the article is structured as follows: In the next section, and as a precursor to our test of competing hypotheses, the debate on whether international and regional organizations can be judged by the same standards of democracy as states will be summarized. As will be demonstrated, it is by no means self-evident that the existence of courts and parliaments as important dimensions of democracy, namely constitutionality and inclusiveness, can be taken as instances of democratic control or increase the democratic quality of ROs. A large part of the literature is rather skeptical about such bodies. Despite the difficulty of clearly associating the existence of these organs with an increase in democratic control, this article takes their existence as prima facie indicators of "more democratically" designed ROs. Section 3 provides an overview of the alternative approaches to explaining the establishment of such organs. Section 4 outlines our research design, and Section 5 presents the results of our statistical analysis.
2 State of the Art: Debating the Need for the Democratic Control of Regional Organizations There appears to be an emerging consensus in the literature that international and regional organizations face a "democratic deficit" (Zürn 2019(Zürn , 2014Hooghe et al. 2018). The lack of organs of democratic control is even cited as a major reason for the crisis of such organizations (Zürn 2019(Zürn , 2014Hooghe et al. 2018). A broader look at the literature reveals, however, that this is by no means self-evident, and there are voices who dispute that international organizations face a democratic deficit (Majone 1998;Moravcsik 2004). The democratic control of ROs is, as one author has remarked, "a question of standards" (Majone 1998: 5). 2 Some dispute that international institutions can be held to the same standards as states (Dahl 1999). Others argue that these standards need to be adapted to the international level, and still 1 A category mistake occurs when actors attribute specific characteristics to a class of phenomena (e.g., democratic standards to regional organizations) that violate these phenomena's general nature (ROs are not state-like institutions but rather instances of international cooperation). 2 In a similar vein, but regarding the question of the legitimacy of regional organizations, Rittberger and Schroeder (2016) also argue that the legitimacy of ROs is a question of standards. They distinguish between a state-and society-based conception of legitimacy.
others that they should even be developed with regard to the status quo that can be observed on the international level (Moravcsik 2004). In the following, the debate will first be organized around the question of those standards: Whether one considers ROs to be state-like and therefore applies the standards of domestic democratic control, or instead looks at them as international organizations to which this standard is not applicable. 3 And finally, we look at ROs as multilevel arrangements, where the international or regional and domestic levels are interdependent and influence each other. This debate will, in the section thereafter, guide our operationalization of democratic control mechanisms within ROs.

Domestic-Level Standards for ROs
In the debate on the democratic control of ROs, many authors directly apply the standards of democratic states. One key argument for why ROs should be subjected to similar standards as states is the state-like-threshold argument: Many ROs have reached a sufficient degree of autonomy and capacity to be able to interfere in the domestic affairs of states, which justifies applying such standards (Zürn 2018). Wolfgang Wessels and Richard S. Katz (1999) argue that the EU has crossed the border from horizontal interstate cooperation to vertical (i.e., hierarchical) policymaking. "This naturally raises the democratic question of how the system of institutions exercising this power is to be controlled and held accountable" (Wessels and Katz 1999: 5). 4 Dominik Zaum states that the growing scope of international organizations and their intrusiveness into areas considered by many states to be within their sovereign domain increases the need for their legitimization (Zaum 2013: 3). Michael Zürn assumes that ROs increasingly become contested or politicized due to their growing authority (Zürn 2014; also, see Keohane 2005). For the EU, Marcus Höreth summarizes the dilemma as follows: whereas there has been an increasing delegation of competences and decision-making powers to the EU, the development of democratic institutions has lagged behind (Höreth 2002; see also Scharpf 1996;Follesdal and Hix 2006). From the state-like perspective, it becomes quite apparent that the overwhelming majority of ROs does not yet meet the standards of democratic control. Consequently, a clear demand to establish such organs of control or to expand the competences of existing ones follows. Two organs emerge as key contenders here, because they can be linked directly to dimensions of democratic control: parliaments for the dimension of inclusion and courts for the dimension of constitutionality or the rule of law. For example, Wessels and Katz argue that "the European Parliament has continuously increased its powers, and, by default, it is the key institution at the European level with regard to popular participation and representation" (Wessels and Katz 1999: 6). In a similar vein, Joseph Weiler (2000) holds that one of the key functions of the European Court of Justice (ECJ) is constitutional: it reviews European legislation regarding the protection of the human rights of European citizens (for a similar argument, see Alter 2014). 5

International-Level Standards of Democracy for ROs
From the perspective of international relations, the picture becomes more complicated because here the international organization, normally an intergovernmental organization with clearly defined competences, is the key concept. Hence the standard against which ROs are being assessed is their efficiency in solving specific cooperation problems, not their democratic quality. ROs are designed to deal with the negative externalities of interdependence. They are problem-solving institutions jumping on the case where unilateral action and noncooperation yield suboptimal benefits (Keohane 1984;Zürn 1998;Axelrod and Keohane 1986;Koremenos et al. 2001). States delegate to ROs precisely those competences that are functionally necessary to address these cooperation problems. Where the costs of monitoring are high and cheating is likely, strong monitoring and sanctioning institutions are established, such as regional courts (Koremenos 2007;Moravcsik 1997).
From a functional perspective focusing on efficient problem-solving, neither regional parliaments nor courts are established to address concerns about democratic control. Regional organizations are not even comparable to states. Hence, there can be no "democratic deficit." According to the limited autonomy thesis promoted by Andrew Moravscik (Moravcsik 2004) this is also true for the extreme case of the EU. Unlike some authors who argue that regional actors behave opportunistically and use functional demand (the need to create a common market) only as a justification for acquiring more extensive competences (Kielmansegg 2009; Scharpf 2019: 250), Moravcsik regards the competences of ROs as restricted. Even in the case of the EU, competences are limited to the regulation of policy externalities and the securing of the four economic freedoms. The EU cannot levy taxes; it cannot implement regulations or directives without member states; and in many areas, such as education, it does not have authority at all. Its budget is restricted to agriculture, structural adaptation, and development aid. He further argues that the level of delegation to ROs, especially the EU Commission, reflects a wider trend in the domestic realm of delegating. The claim that ROs have a democratic deficit overlooks the constitutional constraints on exercising authority that exist.
From such an international perspective, it might even be dysfunctional to create organs of democratic control. According to the technocratic rule thesis, ROs are essentially technocratic organizations created to further economic integration among states. The key organs tasked with furthering economic integration are meant to be insulated from member states' governments, as they would otherwise not be able to fulfill their tasks. According to Giandomenico Majone (1998), it is impossible and even counterproductive to provide regional parliaments with competences com-parable to those of states. It is impossible because specific processes, such as the legislative one, cannot be meaningfully reproduced on a regional level. It is counterproductive, meanwhile, as the delegation of competences that would meet democratic standards would ultimately require a full federation, which is not preferred by member states. Majone argues for the EU, however, that the implementation of the four economic freedoms is at least bound by human rights standards, which is an important dimension of democratic control. This is, arguably, a key competence of regional courts (for a similar argument, see Neyer 2010). In sum, the key argument here is that ROs are still qualitatively different from nation-states and that domestic standards of democracy are not applicable.

Multilevel Governance Standards of Democracy in ROs
The picture becomes even more complicated if one looks at the democratic control of ROs from a multilevel governance perspective. Some authors have consistently pointed out that the increasing transfer of competences precisely to such organs as courts with compulsory jurisdiction and binding judgements as well as regional parliaments leads to the hollowing out of democracy on the domestic level (Scharpf 1999(Scharpf , 2010. Governments might gain more control over transnational problems by delegating authority, but citizens' capacity to control their governments becomes likewise diminished (Dahl 1999;Vaubel 2006). This is especially discussed with regard to the European Court of Justice, which is seen as having interpreted the competence of the EU in the area of the creation of a common market in distinctive ways that undermine the constitutional characteristics of domestic market economies (Höpner and Schäfer 2010). Martin Höpner and Armin Schäfer (2010: 345) list a number of examples where proposals for directives by the EU Commission were voted down by the EU Parliament but were then enforced by the European Court of Justice. The Court of Justice of the Southern African Development Community's judgment on the expropriation of farmers in Zimbabwe is seen as encroaching on that country's sovereignty (Van der Vleuten and Van Hulse 2016).
A number of authors have pointed to the ineffectiveness of democratic controls in the EU. Even if citizens have a right to elect the members of parliaments in some ROs, the competences of parliamentary organizations are not extensive enough to lead to the dismissal of the executive-the Commission, in the case of the EU (Kielmansegg 2009;Follesdal and Hix 2006). Likewise, critics of the parliamentarization of ROs have argued that regional parliaments undermine the strength and competences of national parliaments. Increasing regional integration shifts the legislative process to the supranational level, but supranational parliaments-introducted to control the executive-seldom have legislative or control functions that are comparable to domestic parliaments. This leaves the latter with fewer formal powers. This was one of the reasons for the intervention of the German Constitutional Court after the Maastricht Treaty had been signed (Herdegen 1994). According to the parliamentary decline thesis, parliaments have generally lost competences to the executive branch since World War II, and regional integration is just one area where this becomes evident (Malamud and Stavridis 2016). The parliamentary co-evolution thesis argues instead that the empowerment of supranational parliaments leads to increasing competences for domestic parliaments too, as they start competing with regional parliaments (Winzen et al. 2015). From this perspective, there is an inherent tension between the efficiency standards of international organizations and standards of democratic control within states.
As becomes apparent from these different analytical perspectives, whether or not ROs can be evaluated against the democratic standards of states is disputable. It is also a matter of controversy to what class of phenomena courts and parliaments belong: Are they organs established to democratically control executives, or are they merely ones designed to promote compliance? This article provides no answer to the question of which standard is the most appropriate one. The question of standards is tackled only indirectly by asking the following: What determines the existence of courts and parliaments in ROs? Do these organs exist in ROs because democratic governments expect international organizations to be as democratically controlled as they themselves are? Or, is their existence better explained by a functional demand for them? If this is the case, applying domestic-level standards to what are essentially international organizations might be a category mistake. In the following, we develop this and alternative hypotheses to account for the existence of courts and parliaments in ROs.

Explaining the Institutional Design of Regional Organizations: Courts and Parliaments
In the following, three different explanations for why ROs feature courts and parliaments are developed. Two of them associate the design of international institutions with domestic regime-type variables, and one of them explains it functionally. The theoretical approaches are first being described generally and their specific explanations for courts and parliaments subsequently.

Externalization: Democracies and Regional Courts and Parliaments
Democratic membership has emerged as a key factor for explaining the existence of courts and parliaments in ROs. The assumption is that democratic states externalize their domestic norms and structures and create international institutions mirroring themselves (Tallberg et al. 2016;Heldt and Schmidtke 2019). The externalization thesis has a long tradition in international relations. Liberal theories of democratic peace have long argued that democracies externalize their domestic norms when interacting with other states in the international system. Originally formulated in the context of the democratic peace literature (Russett 1998;Owen 1994), the theory has been extended to the establishment of international institutions more generally. Eugenia Heldt and Henning Schmittke (2019) argue that the democratic status of member states shapes their expectations about democratic control, and that established democracies therefore also create international institutions featuring more democratic control organs-in their case, civil society organizations. Jonas Tallberg et al. (2016) argue that democracies and autocracies "hold competing polity preferences, rooted in their respective domestic political systems. Governments in K liberal democracies work to extend the constitutive principles of democracy to global governance" (Tallberg et al. 2016). A number of more recent studies have focused on the role of parliaments in the execution of the democratic peace. These studies generally point to the important constraining function of domestic parliaments in decisions over international interventions. They find that parliaments with more extensive "war powers" are significantly less likely to agree to foreign military interventions (Dieterich et al. 2014;Wagner 2018). Given the crucial role of parliaments in domestic settings, the more generalized expectation that democratic states externalize parliamentary organs as institutions of democratic control is reasonable. Everything else being equal, autocratic member states seek to constrain such mechanisms or at least are indifferent about establishing them. The democratic externalization thesis thus expects democratic member states to also establish characteristic institutional designs, meaning ROs with parliaments, courts, and organs of civil society representation (not included in our analysis).
We expect that:

H1
The higher the level of democracy of the member states of an RO, the more likely the existence of a regional court, a regional parliament, or both.

Democratic Lock-In: Democratizing States and Regional Courts and Parliaments
A well-established literature on democratic lock-in argues that there is a relatively strong relationship between democratizing states and the setting up of democratic institutions or institutions with strong monitoring powers. New democracies promote quasi-judicial bodies as a tactic to lock in and consolidate democratic institutions (Moravcsik 1995). Edward Mansfield and Jon Pevehouse show more generally that democratizing states are much more likely to become members of regional integration arrangements (Mansfield and Pevehouse 2008). Here, it is not established democracies but the democratizing ones who have the biggest incentives to establish regional courts. Again, the expectation is not that these institutions act as ones of democratic control over executives within the RO but as institutions guaranteeing the democratic status of member states. The body of literature building on Moravcsik's democratic lock-in theory especially sees a strong association between democratization and the establishment of regional (and international) courts (Danner and Simmons 2010). For the Latin American organizations Southern Common Market (Mercosur) Mercosur and the Union of South American Nations (UNASUR), Carlos Closa and Stefano Palestini (2018) show that the demand for the stabilization of democracies motivated the establishment of democracy clauses in both. And, in fact, most of the regional courts in existence today have jurisdiction over human rights (Osiemo 2014;Murungi and Gallinetti 2010;Yerima 2011). The African parliaments are attributed a key role in fostering democracy within Africa (Nwosu 2018). Latin American countries "conceived ROs as a means of domestic democratic consolidation" (Bianculli 2016: 162). From this literature it follows that the establishment of regional courts and parliaments is associated with the existence of democratizing states. Note that these organs are deliberately designed to monitor the development of domestic structures in member states and therefore to intrude on the sovereignty of member states, but they are not necessarily designed to control the executives of ROs. We expect that: H2 The more members of an RO are democratizing, the more likely the existence of a regional court, a regional parliament, or both.

Delegation: Rising Competences and Democratic Control Within ROs
The extant literature on delegation to international institutions provides a third explanation for the existence of regional courts and parliaments in ROs. This literature assumes a clear association between the demand for regional courts and parliaments and their existence within such institutions. As pointed to above in the discussion on international standards for ROs, parliaments and courts are not considered to represent organs of democratic control. Instead, they are problem-solving institutions for mixed-motives games in which actors have incentives to cooperate and defect alike (Keohane 1984;Zürn 1998;Axelrod and Keohane 1986;Koremenos et al. 2001). They help monitor member states' behavior and sanction noncompliance. Where the costs of monitoring are high and defection from cooperation is likely, strong monitoring and sanctioning institutions are established, such as regional courts (Koremenos 2007;Moravcsik 1997). Consequently, a extensive literature focuses on explaining the existence of dispute-settlement mechanisms in general and courts in particular. Works on preferential trade agreements consistently find a correlation between the depth of those agreements and the existence of more legalized disputesettlement mechanisms (Baccini 2018;Jo and Namgung 2012). Functional theories have a hard time explaining the existence of regional parliaments, as these promise few efficiency gains "while generating potentially significant sovereignty costs" (Lenz et al. 2019(Lenz et al. : 1096. Some of the literature explains their existence by their function as symbolic expressions of modern representative democracy borrowed so as to increase democratic control within ROs (Rocabert and Schimmelfennig 2018). The state-like-threshold thesis discussed above proceeds from a similar assumption (Zürn 2014). 6 Jofre Rocabert and Frank Schimmelfennig argue that general-purpose organizations generally have a greater demand for such democratic legitimation and are therefore more likely to have such an organ. The authors find support for this theory in a statistical analysis of 31 ROs, where the existence of an interparliamentary institution is the dependent variable, and the distinction between task-specific and general-purpose organizations is the main independent variable of interest. The difference between the democratic externalization thesis and the delegation thesis is that the latter assumes that the need for democratic legitimation follows from a functional demand triggered by an increase in the competences of an international institution. Hence the existence of parliaments (and courts) should vary with the competences of ROs, not necessarily with the level of democracy of its members.

H3a
The larger the policy scope of an RO, the more likely the existence of a regional court, a regional parliament, or both.

H3b
The greater the share of intraregional trade, the more likely the existence of a regional court, a regional parliament, or both.

The Correlates of Regional Courts and Parliaments: Statistical Analysis
In the following, these explanations for the establishment of regional courts and parliaments are tested using a sample of 72 ROs. 7 First, the operationalization of the variables is introduced; then the descriptive distribution of them is presented, followed by the presentation of the results from various logistic regression models. Finally, the results in relation to the theoretical expectations are discussed.

Operationalization of Variables
The dependent variable is the existence of either a court, a parliament, or both as possible organs of democratic control. The analysis excludes other organs, such as ones of stakeholder representation, because these have, in most cases, even less control powers than courts and parliaments do. For courts, the analysis does not distinguish between economic and human rights courts. Some courts have jurisdiction over both, whereas others have either human rights jurisdictions or act as dispute-settlement mechanisms for economic communities. The dependent variable is conceptualized as an ordinal variable that can take on one of three values: 0 if neither a court nor a parliament exists, 1 if either a court or a parliament exists, and 2 if both kinds of organs exist. "Existence" means that an RO has established the respective organ during its lifetime. Each RO is assessed only once with regard to the existence of these organs, based on the most recent version of its institutional design. On the one hand, this results in a relatively low number of positive observations (see Fig. 1 and Table 1). On the other, it prevents the analysis from overestimating positive cases by taking into account all years in which the respective organs have existed and were not abolished. While taking each RO year of existence as a unit of analysis, it ignores the fact that the chance of these organs of being abolished is basically zero: once established, RO courts or parliaments have never been abolished as of today. 8 The data for the dependent variable originate from the Comparative Regional Organizations Project (Jetschke et al. 2019). This project collects data on more than 7 See Appendix, Table A.I for the list of these ROs along with their predecessors and lifespans. 8 The Southern African Development Community Tribunal was briefly suspended between 2010 and 2012, but resumed its work with a change of jurisdiction in the latter year.  To test the first two hypotheses on the relationship between the democratic membership of ROs and courts and parliaments' existence, this article relies on data from the Polity IV project (Marshall et al. 2014). The externalization hypothesis (H1) was tested with the independent variable democracy level, using the average aggregated 9 The analysis in this article includes only 72 ROs due to missing data on the independent variables. K democracy scores from all RO member states since the RO's establishment as the indicator for the regional level of democracy. The democratic lock-in hypothesis (H2) was tested using democratization as the independent variable. Because democratization is rather a process that occurs in waves and not necessarily in a linear fashion, the share of democratizing states is close to zero in most years. Yet if it occurs, the share of democratizing states across regions varies considerably. Therefore, the average of the share of democratizing states would not entail much information being generated. 10 Instead, we used the maximum share of democratizing states on the RO level, whereas a country was deemed as democratizing by the Polity IV scale if its democracy score had been constantly rising during the three years prior to measurement.
The functional demand hypotheses (H3a and 3b) were tested using, on the one hand, the variable degree of delegation, which relies on the scope of policy areas covered by an RO as an indicator (3a). The data for this indicator originates from-CROP. Policy scope thereby represents the average number of policy areas in which RO member states formally agreed to cooperate, as indicated by the respective RO treaties. 11 The exogenous economic dimension of delegated competences, on the other hand, was tested with the variable trade relations (H3b). This variable uses the average intraregional trade volume between RO member states in relation to the total trade volume by an RO's member states as an indicator. The country-level data for this indicator were taken from the CoW trade dataset (Barbieri et al. 2016).
As controls, regional conflicts, membership scope, and RO age were included as additional variables. The variable regional conflicts includes the respective countrylevel data on intrastate and interstate conflicts within the region aggregated to the RO level and then averaged out across the RO's lifetime. The country-level data were sourced from the CoW dataset (Sarkees and Wayman 2010). For membership scope and RO age, data from CROP was taken measuring the average number of member states for each RO and the number of years since the founding of the RO, 12 with 2018 as the cut-off year.
As a robustness test, we ran the same models with alternative indicators for the independent variables. Instead of taking the average democracy score for the democracy level, we specified democratic regime type as a dummy variable that indicates whether the dominant regime type within a region is democracy or not. 13 As an alternative indicator for the democratization variable, democratic homogeneity of ROs' member was used, which is based on the individual country data provided by Polity IV, aggregating the distances from the mean of each RO member's democracy score. In order to test the robustness of the delegation degree, a dummy variable 10 Its mean (0.04) and standard deviation (0.02) would both be close to 0. 11 Included policy areas are economic development, security, migration, environment, good governance, social welfare, culture, human rights, infrastructure, sustainable development, trade, science and technology, agriculture, police and judicial cooperation, and human resources. 12 Including predecessor organizations; in the case of the African Union, for example, the establishment of the Organisation of African Unity is taken as its starting point. 13 Democracy includes the categories democracy and full democracy from Polity IV, while the other regime-type categories (open anocracy, closed anocracy, autocracy) are deemed nondemocracies. was generated indicating whether ROs are authorized to make "binding decisions"; using data from CROP, decisions were coded as binding if at least one organ of the RO had the right to make such decisions, as indicated by their treaties. Moreover, intraregional trade share as an indicator for trade relations with the data from the KOF globalization index (Dreher 2006). Here, information on countries' relative economic globalization was used and aggregated to the RO level to account for delegation demand generated by integration into the international trade system.

Descriptive Statistics
This analysis, focused as noted on 72 ROs. For each organization, it was examined whether a supranational court, an interparliamentary assembly, or a supranational parliament was part of the formal institutional design. The two types of parliaments were differentiated according to their composition: interparliamentary assemblies are composed of delegations of national parliamentarians from the RO's member states, and supranational parliaments assemble delegates that are elected exclusively for this purpose-either by the member states' legislatives bodies or directly by universal direct suffrage within the member states. Figure 1 above shows the aggregated numbers of formally established supranational courts, interparliamentary assemblies, and supranational parliaments over time, as well as the total number of existing ROs. The story of ROs increasing in number over time between 1945 and 2010 is already well known (Mansfield and Pevehouse 2008): During the Cold War years, 48 ROs were established almost all over the world within 45 years, 14 and after 1990, the number increased to more than 65 existing ROs within 20 years. 15 Yet the number of democratic organs developed rather slowly during the first period, resulting in 12 democratic organs in total by 1990. After the Cold War, democratic organs proliferated, as the total number of democratic organs had increased to 45 by 2010. Thereof, 19 are interparliamentary assemblies, 16 17 are courts, and nine are supranational parliaments.
The first interparliamentary assembly was the Council of Europe in 1949, the first court was the European Court of Justice, and the first supranational parliaments were established in 1979-namely the European Parliament and the Andean Parliament. The most recent instances of democratic organs are the (interparliamentary) Assembly from the Organization of Eastern Caribbean States as well as the Eastern Caribbean Court of Justice of the same organization, and the newest supranational parliament, parliament, Mercosur's Parlasur, formally established in 2005. Looking 14 Six ROs were dissolved during this period: the Carribean Organization, the East African Community, the Central Treaty Organization, the African and Mauritian Common Organization, and the Council of the Entente. Organizational transformations are not counted as dissolutions, for example the evolution from the EC into the EU. 15 Three ROs were dissolved during this period (Arab Cooperation Council, COMECON, WEU). 16 In the case of six ROs the establishment supranational parliaments was preceded by interparliamentary assemblies. These are: AU, Community of Andean Nations, Central Market for Central Africa, East African Community, EU, South American Common Market. As we only include the most recent institutional design for each of these ROs in the statistical analysis, these interparliamentary assemblies are not listed here.
Source: Own compilation based on the Comparative Regional Organizations Project; see Appendix, Table  A.I for full names of regional organizations at the regional distribution of these distinct organs, there are quite a few differences across regions and across the different types of organs. Up to now, 17 supranational courts have been established. While Europe was the first region with such a court, Africa is the continent where more than one-third of all regional supranational courts are situated, followed by the American continents, in this case exclusively in South America, with five courts. Asia has only one court in the Eurasian subregion. Interparliamentary assemblies are somewhat more frequent, with 19 of them having been formally established as of 2018. By far most of them are to be found among European ROs. Most of the supranational parliaments are located in Africa, three of them are in South America, and one is in Europe. In Asia, there are as yet no supranational parliaments. 17 17 Historically, there have been more interparliamentary assemblies. Six ROs (more precisely their predecessors) had established interparliamentary assemblies before they agreed to establish supranational parliaments. These include: African Union, Central Market for Central Africa, East African Community, EU, Central American Integration System, South American Common Market. As we only include the most recent institutional design for each of these ROs in the analysis, these interparliamentary assemblies are not listed here. The descriptive statistics for the dependent and independent variables used for the regression analyses are displayed in Table 3. With regard to the dependent variable, there are four dummy variables used for the different binary logistic regression models: court, parliament, interparliamentary (IP) assembly, and supranational (SP) parliament. These variables indicate whether, with regard to its most recent institutional design, an RO has established the respective type of organ. The other dependent variable, organs combined, is an ordinal variable with two values that we employed for the ordinal regression model. In this variable, ROs with neither of these organs were assigned a value of 0, while ROs with a court or a parliament were assigned a value of 1, and those with both a court and a parliament were assigned a value of 2.

Results from Multivariate Statistical Analysis
The statistical analysis does not lend support to the argument of a correlation between democracies (H1) or democratizing states (H2) and the existence of regional courts and parliaments. The statistical analysis yields four important findings. The argument that the degree of delegation to the RO is positively associated with the existence of courts and parliaments (H3a) is strongly supported, as is the hypothesis about a positive association between intraregional trade shares and the existence of regional courts and parliaments (H3b). The results from the binary logit models (see Appendix, Tables A.II.a-d) show that the existence of the two types of parliaments is associated with different variables: Whereas courts and supranational parliaments are associated with policy scope, regional conflicts, and RO age, interparliamentary assemblies appear to be correlated significantly only with intraregional trade.
K While the establishment of each of these organs individually were tested with binary logit regression models, the reporting is confined to the results of the ordered logit regression models focusing on the existence of courts, parliaments, and both organs in combination. Afterward, the findings on the individual organs are summarized by reporting the results from the binary logit models. Finally, the results from the robustness check are presented the indicators for our independent variables are replaced with alternative indicators.

Results from Ordered Logistic Regression Models
As the main dependent variable of interest is a categorical ordinal variable, ordered logistic regression models are employed to test our assumptions. The results from the different models are summarized in Table 4, which shows the coefficients as log odds with the standard error in parentheses for each variable. As a proxy for R 2 we used McFadden's R 2 along with the log likelihood and Akaike's information criterion (AIC) as indicators to assess the models' goodness of fit.
The domestic regime type variables, measured with the average level of democracy and the highest share of democratizing member states within the ROs, do not show any significant correlations with the dependent variable. Overall, the contribution of both variables to the goodness of fit is almost negligible, as becomes obvious if Models 6 and 8 are being compared. On this basis, both hypothesis 1 on the externalization of regime types and hypothesis 2 on the democratic lock-in relationship between democratizing member states and RO institutions are not supported.
The association between delegated competences and the existence of courts and parliaments finds strong statistical support. The higher the number of policy competences delegated to the RO (delegation degree; hypothesis 3a), the more likely the existence of a court and parliament-as indicated by the positive and highly significant correlation across all model specifications. It can be concluded that widening policy scope increases the chances of ROs establishing courts and (supranational) parliaments. Judging from the log odds for policy scope, 18 an expansion of the policy competences toward one more policy area 19 increases the odds of the RO having a court and a parliament by 36 percent, if the other variables remain constant.
The hypothesis that trade relations (hypothesis 3b) are positively related to ROs having courts and parliaments does find support, as as the average intraregional trade share is slightly significantly correlated with the dependent variable.
Taking the controls into account, the relationship between the number of regional conflicts and the existence of courts and parliaments is negatively correlated. This means that courts and parliaments are more likely to exist among ROs whose members experienced fewer conflicts. However, it is important to note that the question of whether these organs contribute to more peace or whether peace is a condition (3)  K for the establishment of these organs cannot be assessed via this kind of analysis. And while the number of RO members is not related to the RO having a court or parliament, the age of the RO is significantly correlated with having such organs. Therefore, the likelihood of courts and parliaments being established increases as an RO gets older.

Results from Binary Logistic Regression Models and Robustness Checks
In the models presented above, the combination of different organ types is modeled as an ordinal variable. In order to account for the fact that the different types of organs, including the different types of parliaments, might be associated with different variables, we ran several binary logistic regression models (see Appendix, Tables A.II.a-d) where we tested the same groups of variables against binary dependent variables that represented the existence of the individual organs introduced by Table 1 above (courts, parliaments, subtypes of parliaments). In addition, we substituted independent variables in another series of ordered regression models with alternative indicators as a robustness check (see Appendix, Tables A.III.a-b). Our first finding, the absence of any significant correlations between democracy level and democratization and the existence of courts and parliaments, also holds for each type of individual organ analyzed separately with binary logit models. These results do not change if we use alternative indicators for these two variables (dominant regime type and democratic homogeneity; see Appendix, Tables A.III.a).
The association between delegation degree and the existence of courts and parliaments is positive and significant. But this relationship remains intact only for the models with courts and supranational parliaments as binary dependent variables, while it fades away if tested against interparliamentary assemblies individually. As a robustness check for the relationship between delegation degree, the ordinal models were tested with a dummy variable that assessed whether ROs were authorized to make binding decisions on behalf of their treaties. The results from the robustness checks yield significant correlations for this variable as well, while the overall model fit is similar to for the models with policy scope as an indicator for this variable.
The second variable that measures functional demand is trade relations, which correlates slightly significantly in the ordinal logistic models. However, this significance vanishes if this variable is tested against courts and supranational parliaments individually, while it becomes highly significant if tested against interparliamentary assemblies. Using the economic globalization index as an alternative indicator for trade relations in the robustness check yields similar results, with the variable being slightly significantly related and significantly correlated only in the case of interparliamentary assemblies.
Focusing on the control variables, the negative association between regional conflicts and courts and parliaments also holds only partially true if tested against the organs as binary variables. Looking at courts and supranational parliaments separately, the correlation remains equally moderately significant. But focusing on interparliamentary assemblies alone, this significant correlation vanishes. The same change in pattern can be observed for the relationship between RO age and the respective organs. Whereas older ROs are more likely to have established courts and supranational parliaments, the existence of interparliamentary assemblies is not related to the age of ROs.
These observations of different correlation patterns between the independent variables and courts and supranational parliaments on the one hand and interparliamentary assemblies on the other shows that different types of organs apparently follow different establishment logics. Yet all three organs are more likely to be established if functional demand increases, while the likelihood of their establishment is not affected by the domestic regime types of their member states.

Discussion of Results
Our analysis shows that the existence of regional courts and parliaments is rather correlated with variables associated with the findings of the literature on the institutional design of international institutions rather than with domestic regime type variables. This finding also contrasts with other studies that have found that democracies positively affect the democratic design of international organizations. While this might partially be due to differences in operationalization, it casts doubts on the assumption that democratic states are more likely to establish certain types of organs.
The results allow for three different interpretations: The first acknowledges that one is only looking at correlations, not necessarily causal relationships. Statements about causality are generally difficult in analyses that do not use longitudinal data or measurements at different points in time. Considering this, the results can be interpreted in the way that regional courts and parliaments exist where ROs have more competences delegated to them (policy scope), where member states trade more with each other (intraregional trade), and where the regional context is more peaceful (lower number of regional conflicts). These would be the conditions generally conducive to transnational ties and exchanges on the level of parliamentarians, for example. This interpretation is in line with the finding that democracy or democratization does not affect the chances of having a court and a parliament. A broad policy scope, high trade levels, and a peaceful environment are conducive to transnational contact, no matter whether states are authoritarian or not. And even ROs with predominantly autocratic membership profit from courts if they help in creating common markets.
The underlying mechanism could be the one described by the democratic peace literature: Bruce Russett et al. (1998: 441) describes the expectation that: 1) democracies would constrain governments in waging war; 2) commercial spirit in the form of trade and interdependence would reinforce those structural constraints by creating transnational ties that encourage accommodation rather than conflict; and 3) international law and organizations-building on an understanding of the legitimate rights of all citizens-would provide the basis for the peaceful resolution of conflicts. The relevance of the "third leg" of international organizations lay in their power to coerce norm-breakers, mediate between conflicting parties, and contribute to problem-solving. The surprising element here is that one crucial leg of the "democratic tripod" is missing: the association with any measure of democracy. Where there is greater trade and more peaceful relations, the executives coming together in ROs will be K more lenient toward transnational actors and allow greater debate on democratic standards.
A second, bolder interpretation of the results focuses on the functional demand mechanism. Here, the findings strongly support the interpretation that the existence of courts and parliaments on the regional level is due to the specific functions that these organs assume. Increasing economic interdependence also creates externalities for other policy areas, leading to more delegation to the RO and ultimately the need for strong monitoring and sanctioning mechanisms in the form of regional courts. In this interpretation, the existence of courts (and parliaments) has very little to do with the aim of establishing organs of democratic control and everything to do with the demand for international institutions. The first and second interpretations bear the implication that applying standards of democratic control to ROs would indeed be a category mistake.
A third interpretation combines insights on the main correlates of courts and parliaments with the finding that the bindingness of decisions is highly significant. In particular, the strong association between this variable and the existence of courts and parliaments might indicate that this feature indeed creates the demand for regional parliaments as organs of democratic control. This is completely in line with the state-like-threshold thesis, arguing that the degree of autonomy and capacity to interfere in the domestic affairs of states makes international organizations more state-like than international organization -like and hence creates the demand for organs of democratic control (Zürn 2018). This interpretation is also supported by the differential finding on interparliamentary assemblies and supranational parliaments. While stronger trade ties increase the likelihood of ROs having an interparliamentary assembly, higher degrees of delegation, fewer regional conflicts, and the advancement of age make the existence of courts and supranational parliaments among ROs more likely. In this case, domestic standards of democratic control would at least be applicable to those ROs that issue binding decisions.

Conclusion
This article started with the observation that international and regional organizations are increasingly evaluated according to the standards of democratic states, and that there is an empirical literature arguing that the latter are also more likely to design international institutions featuring organs of democratic control. This article first provided a more detailed discussion of why such a comparison is controversial. It then developed and explored different variables, using a dataset including the universe of ROs since 1945. The results do not support the claim that democracies in particular externalize their democratic norms and establish regional courts and parliaments out of concern for democratic control. Instead, and perhaps surprisingly, the existence of regional courts and parliaments is significantly related to policy scope, intraregional trade levels, and low levels of interstate conflicts within ROs-and therefore with variables that we associate with functional demand theories of delegation to international organizations.
Our results also help put the discussion about the "democratic deficit" of regional-and, by extension, international-organizations into perspective. Courts and parliaments first and foremost have to be seen in light of what they have been created for: institutions that monitor the compliance of member states. They were not initially created to democratically control the executives of ROs. It is only in a second step-if decisions become binding-that they can also be seen as institutions of democratic control. Hence, the example of the EU selection process for the EU Commission introduced at beginning of the article is instructive insofar as it might be an example of too high expectations for an organization that is also still an international organization and not a state.