International organizations, European Union access, and authority

ABSTRACT

In recent years, the European Union has gained access to several international organizations (IOs). This is a noteworthy and novel development, as IOs not only formalize but also deepen their interactions with the EU by granting it formal rights to participate in their policymaking processes. Building on the literature on the EU’s role in global governance and the opening-up of IOs, this paper investigates the conditions under which IOs grant the EU access to their decision-making processes. While overlapping policy mandates between the concerned IO and the EU are certainly an important explanation, they are not the entire story. As this paper suggests, much depends on the authority of the IO granting access to the EU. This is because the members of authoritative IOs recognize in the EU a highly authoritative IO and might thus be more inclined to take it on board in order to mitigate negative externalities, enhance their own effectiveness, and avoid intra-institutional conflicts. Using a novel dataset on the EU’s formal access to 33 IOs and addressing important inferential concerns, the statistical analysis shows that EU access depends on the authority of IOs. That authoritative IOs establish formal relations between themselves provides the basis for increased interactions and opportunities to jointly shape global outcomes.

KEYWORDS:

• European Union
• access
• authority
• International Organizations

1 Introduction

The European Union’s (EU) influence in the global governance system and its involvement in international organizations (IO) are controversially debated in the literature. On one side of this debate, some argue that the EU’s role in contemporary IOs remains below its potential due to internal policy-making difficulties, member state heterogeneity, or insufficient resources (Emerson et al. 2011; Pisani-Ferry 2009). On a different view, the EU wields significant influence on third parties in both formal and informal institutions, not least due to its commitment to multilateralism, its large economic might, and its regulatory strength (Jupille and Caporaso 1998; Jørgensen 2009; Blavoukos and Bourantonis, 2010; Gehring, Oberthür, and Mühleck 2013; Bradford 2020).

While previous accounts have focused on the constraints and opportunities that determine the EU’s interest and ability to engage in other organizations, less attention has been paid to the IOs that are expected to welcome the EU into their ranks. Other IOs and their member states must decide whether and which formal status they wish to grant to the Union. Since the EU is not a traditional subject of international law (Govaere, Capiau, and Vermeersch 2004), member states and bureaucracies of other IOs usually have to change their statutes to accommodate the EU. This entails adjustment costs. These costs increase even further if the IO in question grants the EU a formal status with policy influence, allowing a major non-state actor to influence its decisions. The EU’s ability and possibilities to influence world affairs thus depend to a considerable extent on other IOs (see Jørgensen, Oberthür, and Shahin 2011; Dai and Martinez 2012). Therefore, this paper asks under what conditions do IOs grant access to the EU. While EU-IO relations can take several forms, I focus on the EU’s formal policy influence in other IOs, capturing thereby an instance of deep and formalized IO-EU interaction (see introduction). Specifically, I use the concept of access that has been developed in the literature on the opening up of IOs to civil society actors (Tallberg et al. 2014). In contrast to being just represented in an IO, access is defined as the explicit grant of formal participation rights in an IO’s policy-making processes to non-state actors. The Union’s access to other IOs, while not encompassing all types of influence, strengthens its position in global governance institutions as an autonomous actor and enables it to influence global outcomes through explicit decision-making rights.

Why should other IOs grant access to the EU? In addition to overlapping policy competences, I suggest to take the authority of IOs as a predictor of EU access into account. Students of global governance have observed that many IOs exercise authority over their member states (Zürn 2018; Hooghe, Tobias, and Gary 2019), defined as an IO’s competence to make binding decisions and policies that may even go against the short-term interests of some of its members (Zürn, Binder, and Ecker-Ehrhardt 2012, 87). The extent to which an IO exercises authority depends on the formal rights and institutional properties that states have delegated to it (Abbott and Snidal 1998; Hooghe, Tobias, and Gary 2019; Zürn 2018).¹ Organizations have formal authority if they enjoy some policy autonomy from their member states and if their decisions and policies can bind their members to some extent (Zürn, Tokhi, and Binder 2021). Authority varies among IOs, some of which have considerable organizational autonomy and policy bindingness, while others have little – if any – such institutionalized influence over their members. Out of all IOs, the EU is the most authoritative organization (Hooghe, Tobias, and Gary 2019; Zürn, Tokhi, and Binder 2021).² Its high formal authority, backed up by its market power and its regulatory reach beyond Europe, generates negative externalities for some IOs and their members (Jupille and Caporaso 1998; Damro 2012; Bradford 2020). Other IOs, by contrast, may depend on the Union’s resources to enhance their own effectiveness and avoid rule inconsistency in an increasingly fragmented global governance system (Gehring and Faude 2013). I argue that more authoritative IOs are more likely than less authoritative IOs to grant access to the EU. Although IOs may lose some of their autonomy, they also gain influence over the Union by binding it to their policy-making processes and thus constraining it to some extent. Such constraints are stronger in more authoritative IOs, enabling them to more effectively influence the Union in order to mitigate negative externalities (e.g. restricted market access), to enhance their own effectiveness, or to pre-empt intra-institutional conflicts.

To test my expectation about IO authority and EU access, I generate novel quantitative data. Specifically, I code whether or not the EU has access – measured as being either an observer or a full member for each year between 1957 and 2013 – to 33 IOs from the International Authority Database (IAD, Zürn, Tokhi, and Binder 2021). The EU has access to half of all sampled IO. While representing a substantial involvement, the EU has access to about 85% of all global IOs, but only to few regional organizations. A similar pattern holds with respect to issue areas. Using this broad set of IOs and the considerable variation in EU access, special care is taken in the statistical analysis to distinguish the effect of IO authority – the main independent variable – from closely related factors that could confound the relation between authority and access. I find that IOs with more authority are significantly more likely to provide the EU with access to their decision-making processes than IOs with less authority. IO authority is an important predictor for IO-EU interactions that should be considered in addition to explanations about overlapping issue areas and policy competences. More generally, the findings imply that authoritative organizations grant access to the authoritative EU, thus providing the institutional basis for inter-organizational cooperation to affect global policies.³

2 The EU in the global governance system

Students of global governance have observed that formal international organizations have become important carriers of authority beyond the nation state (Abbott and Snidal 1998; Lake 2010; Zürn, Binder, and Ecker-Ehrhardt 2012; Zürn 2018; Hooghe, Tobias, and Gary 2019). States have increasingly delegated rights and competences to IOs and tasked them to contribute to the solution of transboundary challenges and the provision of global public goods. As a result, IOs have become influential actors in global governance that often exercise substantial degrees of authority over their member states. To exercise authority, IOs must act with a certain degree of autonomy from their member states and have the capacity to bind members to a particular course of action. Secretariats with some policy discretion, international courts, but also governing bodies in which a subset of members decides per majority vote for the entire membership are examples of IO autonomy. An IO can bind its members when it has the formal capacity to limit states’ policy discretion by using instruments that may be costly for states. The available instruments vary in type and costs, such as legally more stringent rules vs. recommendations or on-site monitoring vs. self-reporting of compliance. The more autonomous and binding an IO is, the more authority it exercises. Authoritative IO can thus act relatively autonomously from their member states and use their formal competences to limit state discretion.

Importantly, however, IOs have typically sector-specific authority, which is either geographically (e.g. EU in Europe) or thematically limited (e.g. WTO [trade], WHO [health], IOM [migration]). Often, the membership and mandates of IOs overlap, creating regulatory conflicts that might compromise the provision of global public goods (Gehring and Faude 2013; Kreuder-Sonnen and Zürn 2020). Because the global governance system lacks a meta-authority that would coordinate sector-specific IOs and settle conflicts, IOs may be pushed to either compete with each other or try to mutually adjust their policies (Zürn 2018).

Given the characteristics of global governance, this paper investigates the conditions under which formal international organizations institutionalize and deepen their relations with the EU. While other forms of IO-EU interactions exist (e.g. Paper 4), I focus on formalized relations – reflecting an explicit and thus binding commitment by a formal IO to engage with the EU – that accord policymaking rights to the EU. In line with the introduction, this is an instance of deep interaction as it grants the Union an institutionally recognized influence that goes beyond its mere diplomatic representation in IOs. In the following, I first discuss the interest and incentives of the EU to seek access to authoritative IOs before developing an account for why authoritative IOs may grant access to the EU.

2.1 EU incentives for authoritative IOs

Since its creation, the EU has been the most authoritative organizations in world politics (Zürn 2018; Hooghe, Tobias, and Gary 2019).⁴ In principle, this high authority confers on the EU the capacity to act relatively autonomously vis-à-vis its member states and establish and maintain relations with third parties, such as IOs and transnational non-state actors (Jupille and Caporaso 1998; Jørgensen 2009; Gehring, Oberthür, and Mühleck 2013). Being an organization with sufficient organizational autonomy and a strong commitment to multilateralism, it is hardly surprising that the EU wants to shape outcomes in world politics and thus to engage with other IOs. Indeed, through its market power, the combined weight of its member states, and stringent regulations (Damro 2012; Manuel 2021), the EU already possesses significant global influence, which it may seek to seal through its engagement with other IOs, formalizing and projecting its unilateral power to more jurisdictions (Bradford 2020; Jarlebring 2021).

However, the Union may not want to or be able to join all IOs. Rather, and assuming that the EU wants to influence global policies through IOs, it has to weigh its opportunities and limitations to achieve that goal. An important determinant in that regard is the EU’s competence and policy overlap with another IO. Although the Union is overall the most authoritative organization, that authority is not equally distributed across the Union’s policy domains (see Gehring, Oberthür, and Mühleck 2013). EU competences are high, and even exclusive, with respect to market regulation (e.g. external trade, competition, or agricultural policy). Here, the Union wields considerable resources (financial, political, legal) in addition to, and sometimes even exceeding, its member states. By contrast, in matters of Common Foreign and Security Policies (CSFP), the EU has typically only supporting competences, facilitating coordination between its member states, but not authoritatively legislating and enforcing a common non-proliferation policy, for example. Indeed, qualitative accounts have repeatedly shown that the EU’s involvement in other IOs is driven by the unequal distribution of EU competences across its policy domains (Frieden 2004; Gehring, Oberthür, and Mühleck 2013).

Given an imbalance in its competences and thus limited resources, the EU must choose in which organizations to become involved. To the extent that it seeks to maximize its global influence and that issue areas overlap, it should engage with authoritative IOs as these wield the instruments and capacity to affect global regulations and policies. Moreover, authoritative IOs typically decide per majority vote and not unanimity (Haftel and Alexander 2006; Hooghe, Tobias, and Gary 2019; Zürn, Tokhi, and Binder 2021). This provides additional incentives for the EU as it might benefit more from a unified representation by pooling EU votes, rather than risking to undermine its influence in unauthoritative IOs where unanimity voting prevails (see Frieden 2004). The authority of an IO thus appears to be a further important factor that influences EU incentives to seek access to IOs. This could also explain why the Union does not seek access to the relatively unauthoritative Pacific Island Form and the Central African Economic and Monetary Community, despite significant competence and policy overlap.

2.2 Why authoritative IOs grant access to the EU

While the Union might want to join IOs, the concerned IOs have to make a decision whether or not to grant access to the EU. The Union and its institutions cannot join other IOs as ‘traditional subjects of international law’ (Govaere, Capiau, and Vermeersch 2004, 158). Unlike states that can join IOs either freely or if they meet certain membership criteria, IOs must amend their statutes or find equivalent solutions to welcome the EU into their ranks. This requires legal, procedural, and administrative preparations by the IO bureaucracy (that communicates with the EU) and, most importantly, the agreement of the members of an IO.⁵ As a result, whether the EU can exercise its influence therefore depends on the willingness of other IO, including their members, to recognize it and grant it certain participation rights. While the EU may be involved in many different ways in another IO (see Gehring, Oberthür, and Mühleck 2013; see SI introduction; see Papers 4, 6, and 9), a particularly consequential one for its role in global governance is the formal recognition of policymaking rights by another IO. The concept of access, theorized by Tallberg et al. (2014) to capture the increasing involvement of transnational actors in IOs, travels well to the case of the EU. Going beyond the representation in IOs, access encapsulates a specific formal status for non-state actors who acquire the right to partake in the IO’s policy process. Granting access to non-state actors entails costs for the IO in question, which are arguably higher in the case of the EU. Each IO has to individually decide whether or not to open up to the EU. This generates adjustment and coordination costs, both for the bureaucracy and the membership of an IO. More importantly, given the authority of the EU and its economic and political weight, granting access to the EU may also impact the autonomy of an IO by complicating rule making and even shifting the distribution of power within the IO. For example, were the Union to become a full IMF member, it would acquire de facto veto power in the Fund at the expense of the United States (Pisani-Ferry 2009).

Therefore, other IOs and their member states must weigh the costs and benefits of EU access. To better understand such decisions, I start from the baseline assumption that IOs are more likely to grant access to the EU when their relevant regulatory issue areas overlap.⁶ When policies are fully harmonized within the EU, such as in trade or agriculture, it is more expedient for an IO within the same issue area to directly engage with the EU, rather than trying to negotiate with each of its individual member states. The EU’s full membership in the World Trade Organization (WTO) or the Food and Agriculture Organization (FAO) can be partly explained by overlapping competences between these IOs and the Union’s exclusive competences (Gehring, Oberthür, and Mühleck 2013).

While overlapping competences form the baseline expectation, they are not the entire story. Some IOs in issue areas that do not fall within the EU’s exclusive competences do seek close ties with the Union. The International Criminal Court (ICC) or World Health Organization (WHO) are two cases in point where the Union has certain participation rights. I therefore suggest that an additional factor that influences the likelihood of EU access is an IO’s authority.

But how is the authority of an IO related to its decision to grant access to the EU? The formal authority of the European Union and its economic and regulatory power is not without effect for other organizations. In particular, I suggest three rationales that might propel IOs to grant access to the EU: mitigating negative externalities, furthering one’s own effectiveness, and avoiding intra-institutional regulatory conflict. Authoritative IOs are more likely than less authoritative IOs to address these issues, because by granting access to the EU, they not only give up some of their autonomy, but also bind the Union to their policymaking procedures and rules. I next discuss the three rationales and the role of IO authority.

Concerning the first rationale, it has been argued and shown that the European Union (and its uni- or plurilateral action) generates significant negative externalities for IOs and their members (Jupille and Caporaso 1998; Damro 2012; Bradford 2020; Manuel 2021). For example, producers from third countries have often to comply with European product standards to export to the largest single market with little influence over what Brussels decides (Bradford 2020). Extending over a range of other issues, such as digital technologies, environmental affairs, or labor rights, the EU’s unilateral power often adversely affects state and non-state actors outside Europe. It therefore comes at little surprise that affected parties that are organized in other IOs might want to ‘discuss, modify, or simply understand’ European regulations and policies (Jupille and Caporaso 1998, p. 215–216). Negative externalities can be larger for authoritative IOs, when unilateral EU action undermines their own rule setting and implementation. Anticipating such developments, authoritative IOs might grant access to the EU to not only pre-empt such conflicts through joint deliberations, but also to constrain the Union to some extent. When the EU accedes to another IO, it not only acquires policy-making rights, but is also subject to the same constraints and obligations as any other (state) member (Govaere, Capiau, and Vermeersch 2004; Dai and Martinez 2012). These constraints are higher in authoritative IOs and hence increase their options to effectively influence the EU and thereby mitigate negative externalities. For example, both the FAO and the International Coffee Organization (ICO), two relatively authoritative organizations that are directly affected by EU policies, have changed their statutes to grant full membership rights to the EU.

With respect to the second rationale, the EU’s governance resources and authority may appear attractive for IOs wishing to enhance the implementation of their own policies (Gehring, Oberthür, and Mühleck 2013). More generally, international institutions grant access to non-state actors in exchange for their resources in monitoring and enforcing international rules (Mitchell 1998; Dai 2002; Tallberg 2002; Tallberg et al. 2014). However, mobilizing EU resources for one’s purposes can be a double-edged sword. For organizations with little autonomous and binding decision making this can mean that the EU dominates their policy process (see Paper 9). More authoritative organizations can guard themselves against undue influence of the EU, while still trying to engage the Union’s resources. For example, the UN, and particularly the UN Security Council, actively seek the cooperation of EU member states to enforce Chapter VII sanctions (Govaere, Capiau, and Vermeersch 2004), attempting thus to give full effect to the Council’s considerable formal enforcement authority. To the extent that such functional needs arise, granting access to the EU can enhance the effectiveness of decisions adopted under the rules of the concerned IO. Moreover, because authoritative IOs have greater policy discretion than less authoritative IOs, they can more easily decide to involve the EU in order to expand their resources and enhance their regulatory effectiveness. For example, the ICC, which is vested with significant powers but without the critical participation of the United States, has granted access to the EU, relying in part on Brussels for the global promotion of the Rome Statute and its own internal operation (Groenleer and Rijks 2009). For authoritative IOs, taking the EU onboard facilitates the implementation of their rules and helps to bring their formal authority into full play, while containing a disproportional influence of the EU on their internal operation and policymaking.

The third rationale for why IOs grant access to the EU springs from the structure of global governance. The lack of a central meta-authority forces IOs to interact on a decentralized basis with each other (Zürn 2018). As a result, inter-organizational conflict or cooperation may arise. In that regard, authoritative IOs should be more likely than less authoritative organizations to adjust their policies, adopt binding rules, and effectively cooperate with each other. In authoritative IOs, the control of each single member over the organization is reduced, minimizing thus possibilities to veto the establishment of formal relations with other influential organizations. State concerns over their sovereignty thus may play a less prominent role, facilitating the opening up of an IO to the EU (Tallberg et al. 2014). In turn, this allows IOs to coordinate with the EU and avoid regulatory conflicts that tend to become more salient with the increasing fragmentation of global governance (Gehring and Faude 2013; Zürn 2018; Kreuder-Sonnen and Zürn 2020). Moreover, by pooling their resources, the EU and authoritative IOs might jointly strengthen their policy discretion vis-à-vis states, reducing thus inter-institutional competition that could undermine their influence in global governance.

The discussion of the three rationales and the role of IO authority results in the following testable expectation: More authoritative IOs are more likely than less authoritative IOs to grant access to the EU.

3 Data and method

To systematically test my hypothesis, I use a sample of 33 IOs from the IAD (Zürn, Tokhi, and Binder 2021). The sample is representative of the geographic and thematic distribution of IOs in the 21st century.⁷ The IAD’s single measure of authority varies considerably, comprising highly authoritative IOs and also those that have barely any formalized influence. The time frame of analysis spans the years from 1957, the creation of the EU’s predecessor, the European Communities (EC), to 2013. The unit of analysis is the IO-year.

In the remainder of this section, I first outline the measurement of the dependent variable – EU access. Then, I present a set of independent variables used for the quantitative comparative analysis. Finally, I discuss model choice and specification.

Dependent variable

To measure EU access, I code the Union’s formal participation status in the 33 IOs of the IAD. The focus lies on the EU as an autonomous organization and not on its member states. I consider any EU institution in another IO, such as the Commission, the Council, or the European Central Bank, as an instance of EU involvement. I code the formal status the EU has in each of the 33 IOs for each year. Unlike informal relations, the formal status of the Union expresses an explicit recognition by the other IO and comes with specific participation rights and obligations. Moreover, because the formal status is an explicit and observable measure, it allows me to systematically compare the EU’s involvement across organizations and over time.⁸

The Union’s formal status can take various forms. Following Tallberg et al. (2014), I distinguish access – that involves formal policymaking rights in other IOs – from other forms of engagement with IOs. Obviously, not having any official relations with other organizations leaves the EU without any formal influence. Often, however, the EU has official diplomatic relations with other organizations by sending a delegation to their headquarters (e.g. African Union) or regularly consulting with staff from other IOs (e.g. NATO). By accepting the Union’s diplomatic delegation, an IO recognizes the EU, enabling thereby also informal ways of influence. However, this status does not attribute specific rights to the Union. By contrast, the Union gets access to another IO when it enjoys (enhanced) observer status or, even more so, when it becomes a full member. Being an observer comes not only with diplomatic recognition, but also with the right to raise agenda items or to position oneself on policy matters. For example, at the United Nations General Assembly, where the EU is recognized as an observer since 1974, the Union can propose its own draft resolutions. Being a full member, in turn, not only includes all rights associated with observer status, but also gives the Union full voting rights.

Organizations thus grant access to the EU when they accept it as an observer or full member, reflecting a deep and highly formalized IO-EU interaction (see SI-Introduction). It is deep because the Union is not only recognized as an autonomous actor, but also acquires institutionalized policymaking rights in another organization. It is highly formalized because IO member states have to explicitly consent and IO bureaucracies have to adjust their statutory procedures to reflect the institutional access of the EU. Other forms of IO-EU interactions, such as bilateral consultations or diplomatic exchanges, may offer informal ways of influence, but do not recognize and institutionalize the Union’s right to partake in decision making. Although observer status and full membership are different, I consider both as indicators of access for the following reasons. First, both statuses confer explicit rights to the Union in contrast to a mere diplomatic presence. Second, many IOs that are willing to grant access to the EU, but whose statutory amendments are pending, have given enhanced observer status with policymaking rights to the Union. In many cases, this even amounts to de facto membership (e.g. International Whaling Commission, Organization for Security and Cooperation in Europe).

I dichotomously code EU access, distinguishing between instances where the EU lacks any formal options of influence from those where it has been granted the de jure means to partake in another IO’s policymaking. Accordingly, for each IO and year, the dependent variable receives the value one when the EU is either an observer or a full member and zero otherwise. The temporal range spans the years from 1957, the year of the creation of the European Communities, until 2013 when the IAD data end. The dependent variable thus describes not only variation in EU access across IOs, but also possible status upgrades within the same IO over time, such as when the EU’s formal status switches from no presence (or only diplomatic delegation) to observer or from observer to full member.

Concerning the distribution of the dependent variable, the EU has access in about half of the sampled IOs in the year 2013. Specifically, in 51% of IOs considered in this article, the EU is either an observer or a full member. While this is substantial, EU access differs across regions and issue areas. The Union has access to about 85% of all global IOs. By contrast , the EU has access to five out of 19 regional organizations (26%).. Moreover, access varies across substantive issue areas. The EU has access to about 75% of all multi-issue IOs. However, it has no access to the security IOs in the sample (e.g. NATO, Shanghai Cooperation Organization [SCO]). When considering both region and issue area, the contrast becomes even stronger. All global economic IOs gave access to the EU, while only ten percent of regional economic IOs did so.

Figure 1 plots the proportion of IOs that have granted access to the EU over time. Three years after the adoption of the Rome Statue, the European Communities (EC) join the OECD in 1960. In the late 1960s, the EC accedes to the Bank for International Settlement (BIS) and several UN-related organizations, such as the World Health Organization (WHO). After the end of the Cold War, more and more IOs grant access to the EU (Jørgensen 2009, 16) and by 2013 the EU has access to 51% of all IOs in this sample. The involvement of the EU in other IOs seems to follow the growing importance of IOs more generally (Jupille and Caporaso 1998; Zürn 2018).

Figure 1. Proportion

Figure 1. of IOs with EU access.

Independent variables

The goal of this article is to analyze the relationship between IO authority and EU access. The focus therefore lies on distinguishing the influence of IO authority from variables that are related to both authority and EU access.

The independent variable of interest is IO authority. I expect that more authoritative IOs are more likely to grant access to the EU. To measure an IO’s authority, I use the composite authority indicator from Zürn, Tokhi, and Binder (2021). This indicator captures both the extent to which an IO is autonomous from its member states and the degree to which it can bind its members to substantive policies. The indicator is based on the product of autonomy and bindingness, measured over seven distinct policy functions that IOs perform, and expresses in a single number how much authority IOs exercise over states. I recode the variable to vary between zero, denoting the lack of authority, and a theoretically maximum value of 10. Higher values denote more authoritative IOs. The sample mean for the 33 IOs between 1957 and 2013 is 2.43 (standard deviation = 1.8), where the United Nations is the most (6.05) and the Arab Maghreb Union (0), for example, the least authoritative IO.

As mentioned, EU access varies systematically between regional and global IOs. Because an IO’s geographic scope is related to its authority (Hooghe, Tobias, and Gary 2019), it is important to control for this variable as it might confound the association between IO authority and access. Accordingly, the variable Regional IO takes the value one if an IO has a regional scope (e.g. African Union) and zero if an IO has a global mandate. In the sample, almost half of the IO-year observations belong to regional IOs.

Another potential confounding variable is issue area that reflects the substantive governance domain of an IO. The number of IOs and their average authority differ across issue areas. Organizations have systematically more influence in some issue areas, (economy) than in others (human rights). Moreover, the overlap of EU and IO competences increases the likelihood that the IO will grant access. For example, the EU’s exclusive competences in trade make it more likely that the WTO will grant it access (Gehring, Oberthür, and Mühleck 2013). In other words, the specific issue area in which an IO is active will be correlated with the likelihood of access. To account for this, I use issue-area fixed effects. Following the IAD, I define four broad issue areas into which IOs fall based on their mandate: economy, security, human rights, and multi-issue. The economy issue area comprises both financial, trade, development, and commodity IOs. The multi-issue category covers large organizations, such as the United Nations or the African Union. The human rights issue area includes organizations that promote and protect human lives and livelihoods, such as the ICC, UNESCO, or the ILO. For each of the four issue areas, I use a binary variable taking the value one if the respective IO is in that issue area and zero otherwise. Issue-area fixed effects help to assess whether IO authority affects EU access irrespective of differently institutionalized issue areas and overlapping policy competences.

As suggested, EU access may vary with the dynamic of IO creation. More IOs reflect more options to become involved and might even lower the possibilities for EU participation when these have not yet opened up to non-state members. To capture these dynamics, I include the annual changes in the number of IOs (labeled Δ IOs).

This set of variables comprises important factors that might confound the relation between IO authority and EU access. The descriptive data suggested as much. Therefore, my modeling approach attempts to minimize inferential threats.⁹ Table B.2 in the Supplementary material shows descriptive statistics for the variables used in the analyses.

Modeling

Because the dependent variable is binary, I use a logit model for the statistical analysis. All independent variables that vary over time are lagged by one year.¹⁰ Moreover, to account for possible correlated observations within IOs, I cluster standard errors by IO. All models use issue area fixed effects.

With this baseline setup, I take a series of additional precautions to adjust the statistical analysis for both the potential confounders outlined above and other concerns. First, there is temporal dependence in the dependent variable. The EU’s status can be upgraded over time, as happened in 13 out of 33 IOs. On the other hand, once it has been granted access, it typically keeps it. These are time-dependent processes that the standard logit model cannot capture and therefore I define time polynomials to model this temporal dependence (Carter and Signorino 2010).¹¹

Second, several unobserved time-invariant IO characteristics could affect both the authority of an IO and the likelihood that it grants access to the EU. To rule out that such factors bias the analysis, I run a model with IO-fixed effects. Analyses with fixed effects focus on variation within organizations and hence time-invariant variables, such as Regional IO, are excluded from the estimation.

Finally, I use weighted regression to enhance comparability of IOs, reduce model dependence, and address endogeneity concerns. Because authority is not randomly distributed, some IOs will systematically differ on a series of factors that can affect their authority and the outcome. In the standard logit regression, I control for such confounding factors. Yet, this approach often risks to adjust the effect of IO authority away and relies heavily on correct model assumptions. To avoid this, analysts can use propensity-score weighting to remove imbalances between more and less authoritative IOs that are due to observed confounders (Rosenbaum and Rubin 1983). Weighting proceeds in two steps. First, analysts predict the probability of being an authoritative IO, conditional on the observed confounders, and take the inverse of that probability to derive weights for each observation. Using these weights, observable differences between IOs are leveled out. Second, the weights are used in the logit regression on the relationship between IO authority and access without including any further confounders. To compute the weights, I use entropy balancing that removes all imbalances induced through observed confounders (Hainmueller 2012). For the weighted regression, I dichotomize the authority variable, where IOs with an authority score above the sample mean receive the value 1 and IOs at or below the sample mean the value zero.¹²

4 Results

Table 1 presents results. Model 1 is the baseline logit model. Model 2 adds time polynomials to the specification of Model 1. Model 3 uses IO-fixed effects. Model 4 presents estimates from a weighted regression using entropy balancing (EB) weights.

Table 1. Logistic regression estimates.

	(1)	(2)	(3)	(4)
	Base	Time	FE	EB
IO Authorityt-1	0.322**	0.276**	2.531***	1.917**
	(2.04)	(1.97)	(7.74)	(2.28)
Regional IO	−0.463	−0.0380
	(−0.59)	(−0.05)
Δ IOs	−0.452***	−0.381***	−0.614***
	(−3.25)	(−3.76)	(−4.38)
Constant	−1.631**	−3.385		−2.215***
	(−2.51)	(−1.33)		(−2.89)
χ^2	14.87**	32.32**	165.2***	5.212**
Log.-Lik	−856.5	−829.2	−419.5	−805.1
BIC	1756.9	1717.1	852.5	1624.9
N	1531	1531	870	1531

Note: Logistic regression. Issue-area fixed effects and time polynomials (Model 2) omitted from output. FE = fixed effects; EB = entropy balancing. Note that IO authority is dichotomized in the EB model. IO-clustered robust standard errors. T-statistics in parentheses. Interpretation: To obtain percentage changes, exponentiate the coefficient on the independent variable e^b. For a one-unit increase in IO authority in Model 1, the probability of EU access grows by e^0.32 = 1.38, yielding a percentage change of (1.38–1)*100 = 38%. Significance: ⁺ p < 0.10, ** p < 0.05, *** p < 0.01.

I find that more authoritative IOs are more likely to grant access to the EU. Across all models in Table 1, IO authority is statistically significant and positively related to the probability of access. The higher an IO’s authority, the higher the likelihood that it grants the EU access in the form of an observer or full member. As all models include issue-area fixed effects, this finding holds regardless of the specific issue area in which an IO is active and of the associated competence overlap between the respective IO and the EU.

Based on the estimates of Model 1, the odds that an IO will grant access to the EU if that IO’s authority increase by one unit is 38%. This estimate is close in terms of substantive effects to the one from Model 2 where I control for the temporal dependencies in the data. Here, a one-unit increase in an IO’s authority raises the probability of EU access by about 32%. Surprisingly, and in contrast to the descriptive patterns presented above, the geographic scope of an IO does not appear to be systematically related to EU access. While in both Models 1 and 2 regional IOs tend to refrain from granting the EU access, the coefficient on Regional IOs is statistically insignificant. Concerning the pattern of newly-created IOs, a positive change in the number of IOs reduces the likelihood of the Union’s access to IOs. The coefficient on Δ IOs is negative and statistically significant. This is plausible because as the number of IOs increases, there are new actors, both IO bureaucracies and member states, that must first decide whether or not to grant access to the EU.

In Model 3, I use IO fixed effects to adjust the estimates for unobservable time-invariant IO characteristics. As opposed to Model 1 and 2, the interpretation changes in Model 3, because the focus now lies on the variation in access within IOs and not across them. However, substantive results are consistent: when an IO’s authority grows, it is significantly more likely to grant the EU access. Specifically, when considering the average marginal effect of authority, the probability of EU access grows by about eight percentage points for a unit-increase in an IO’s authority (0.081, 95% CI: 0.077,0.085). This is an important finding, because it shows that to the extent that an IO becomes more authoritative over time, it is also more likely to welcome the EU as either an observer or a full member, irrespective of its time-invariant institutional characteristics, such as its specific geographic sub-region or its substantive policy mandate.¹³ An example that illustrates these general associations is ASEAN. The European Union established official relations with ASEAN in the late 1980s, maintaining a diplomatic delegation at ASEAN’s headquarters. But it was not until ASEAN member states gave the organization more authority that it granted access to the EU. It could be the case, however, that the EU influenced ASEAN member states in their decision to delegate more or less authority to their IO, trying to affect the odds for EU access. The claim that the EU may influence the authority of other IOs is testable and Table C.4 in the Supplementary material shows that EU presence in other IOs, defined as including also mere diplomatic relations, is not related to the authority of IOs. Moreover, in a further robustness check I find that the EU’s diplomatic presence at another IO does not increase its chances of having access to that IO, while IO authority is still positively and significantly related to EU access (see Table C.5 in the Supplementary material). While the EU’s informal influence on ASEAN cannot be ruled out for that particular case, the statistical evidence on the full sample of IOs shows that IOs first tend to become more authoritative before granting access to the EU.

Finally, Model 4 in Table 1 presents estimates from the weighted regression. Recall that I dichotomized the IO authority variable. Being an authoritative IO raises the probability of granting access to the EU when compared to an unauthoritative IO. In terms of predicted probabilities, the probability of the EU having access is more than four times bigger for authoritative IOs than for unauthoritative IOs. Specifically, while the probability of EU access in low-authority IOs is 10%, the probability of EU access in high-authority IOs is estimated to be 43%.

To more closely inspect the association between IO authority and EU access, Figure 2 plots predicted probabilities based on the estimates of Model 1. Specifically, it shows how likely IOs are to grant access to the EU as a function of their level of authority. The x-axis divides the IO authority variable into four categories, reflecting the quartiles of its distribution. Intuitively, IOs with an authority value at or below the 25^th percentile have very low to no authority. The 50^th percentile value represents the median authority score, while the 75^th and 95^th capture high and very high IO authority levels. The y-axis plots the probability of an IO granting access to the EU.

Figure 2. Predicted probability of IOs granting access to the EU conditional on their authority.

The probability of EU access to IOs is steadily increasing for IOs with more authority. This pattern is also statistically significant as the confidence intervals in Figure 2 do not comprise the value zero. The probability that low-authority IOs (at or below the 25th percentile) grant access to the EU is estimated to be around 0.21. Organizations within this lowest percentile lack independent policy competences and are under the full control of their member states. Examples include the SCO or the Arab Maghreb Union. It is doubtful whether such organizations would benefit from EU involvement in any important ways. The externalities that the European Union generates for them might be too insignificant as to motivate the Union’s inclusion in the IO. Moreover, as member states control low-authority IOs, their policy preferences would matter more than considerations of the IO bureaucracy (if there is any) to tie the EU to them. For example, in the SCO it is authoritarian regimes that dominate the organization and whose priority might not be to allow a powerful organization of democratic states to partake in their IO. Organizations with an average authority level (50^th percentile) are more likely to grant access to the EU than their low-authority counterparts. That second, median, quartile comprises IOs such as the International Whaling Commission (IWC), the Organization of American States, or NATO. These IOs exercise authority only over very specific functions they are tasked to perform, such as for epistemic purposes (IWC) or in setting standards and rules (OECD). These IOs have a predicted probability of granting access to the EU of around 0.30. However, that estimated probability is also half as much as the probability of EU access among the most authoritative organizations. In that highest quartile, the predicted probability of EU access is around 0.60. The United Nations, the World Bank, the ICC, or the FAO, fall within that quartile. These are major, and predominantly global, organizations that exercise authority over a series of different policy functions to produce binding substantive policies for states. These IOs grant the EU most readily access and try to integrate it into their respective policy-making processes as either an observer, where statutory changes are too cumbersome (UN General Assembly), or even as a full member (FAO, WTO, International Coffee Organization). The rationale of authoritative IOs to grant access to the EU may vary from case to case. Yet the pattern is clear: It is authoritative IOs that are most likely to grant access to the EU. More generally, the findings suggest that IO authority is an important predictor for IO-EU interactions that should be considered in addition to explanations about overlapping issue areas and policy competences.

5 Conclusion

This paper has investigated the conditions under which IOs grant access to the EU. In addition to overlapping policy competences, I have argued that more authoritative IOs are more likely than less authoritative organizations to give the EU a say in their decision making. This is because authoritative organizations can more effectively address negative externalities caused by unilateral EU action, mobilize EU resources for their own effectiveness, and engage the Union to avoid inter-institutional regulatory conflicts. My statistical analysis supports my expectation and shows that organizations with more authority are more likely to grant access to the EU than those with less authority. Importantly, this association holds when controlling for competence overlap (issue area fixed effects) and unobserved IO characteristics (IO fixed effects). Taken together, my findings suggest that the EU has not only access to IOs in issue areas where competences overlap, but also in those organizations that possess the institutional capacity to act relatively autonomously and adopt binding rules.

A central implication of my findings is that authoritative institutions tend to establish formal relations with the highly authoritative EU, providing thus an institutionalized basis for increased interactions and possibilities to jointly shape global outcomes. This is a novel development in global governance (e.g. Zürn 2018) that requires, however, further work to better understand its analytical and political consequences. I therefore discuss two possible avenues for future research: refining causal logics and accounting for contextual factors.

First, while the findings demonstrate that IO authority is associated with EU access, the theoretical discussion suggested three rationales for why authoritative IOs would open up to the EU (see above). These rationales could all simultaneously influence decisions about access to the EU, or some could be more important in some cases than in others. This calls for more attention to the conditions under which each particular rationale works, spelling out the distinct causal mechanisms that propel authoritative IOs to open up to the EU and testing them. We thus may distinguish between cases in which other IOs simply try to avoid negative externalities caused by the EU and those in which influential organizations interact with the EU to pre-empt institutional conflicts and pool their resources, eventually allowing them to increase their joint leverage over states.

Second, given the scope and goal of this article, the role of the international environment has received less attention. Changes in the global distribution of power, transformations of the global governance architecture (e.g. growing fragmentation), the pace of economic globalization, or even macro-trends in domestic politics (e.g. democratic backlash) can affect the odds for institutionalized IO-EU interactions in both negative and positive ways. Formalized cooperation among authoritative IOs could fuel domestic and transnational contestation if they are viewed as non-majoritarian institutions that combine forces and where domestic constituencies have little influence on their policies (Zürn 2021). On the other hand, fundamental international crises often create new incentives for stronger IO-EU cooperation. For example, the cooperation between NATO and EU during Russia’s war on Ukraine reaches unprecedented levels of intensity, paving the way to a more formalized and deeper cooperation between the two IOs. As a result, future research should consider such external dynamics and events to develop more fine-grained theories and empirical tests of inter-institutional cooperation.

Acknowldgement

I am grateful to the editors of this special issue for their helpful comment. I would like to thank the anonymous reviewers for their constructive feedback and suggestions. Finally, I thank Rebecca Majewski and Asja Riggert for their research assistance.

Disclosure statement

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

Additional information

Funding

This work was supported by the European Commission [822654].

Notes

1. This is not to say that informal authority does not play a role. Yet informal practices often lack the binding power of formalized rules.

2. From a global governance perspective (see also the special issue's introduction), the EU fits the criteria of an IO. It differs from other IOs in that it exercises unmatched levels of authority over its member states over several policy areas.

3. On the possible tonality of such cooperation, see Paper 3.

4. This applies also to the European Communities, the predecessor organization of the EU.

5. The members of an IO have to decide also whether or not they accept applicant states if these meet the membership criteria. Yet as opposed to the case of the EU, they do not need to change their statutes to accommodate new state members.

6. Following the habit in comparative IO research, I use the term IO to designate both its member states and the bureaucracy (see Tallberg et al. 2014; Hooghe, Tobias, and Gary 2019).

7. See Supplementary material for a list of included IOs.

8. The decision to code the EU’s formal status in other IOs involves a trade-off as informal relations between an IO and the EU may precede the formalization of their interaction. Yet the gain in systematically comparing the conditions for granting a formal status to the EU – which can be seen as the conclusion of a possible informal engagement – outweigh the considerable efforts needed to capture the myriad and often unobservable ways of informal EU-IO relations.

9. In the supplementary material I also consider an IO’s prominence and policy scope. Including these has no effect on the main findings, see Table C.3.

10. To guard against reverse causality, I assess whether EU presence or EU access drive an IO’s authority. Results in Table C.4 suggest that they do not.

11. Specifically, I use the linear, quadratic, and cubic version of a variable that indicates analysis time.

12. I balance on these confounders (all lagged if time-varying): issue area, regional IO, the number of IOs, the prominence of an IO, and the policy scope of an IO.

13. Including an IO’s policy scope in fixed-effects models does not change substantive findings.