1 Introduction

Legitimacy is a critical governance resource for international organizations (IOs) because it taps into the “wellspring of voluntarism” (Reus-Smit, 2007, p. 163) that comes with viewing rules as binding for moral reasons. Legitimacy lowers the cost of rule, enhances the likelihood of compliance, and stabilizes the political order (Beetham, 1991). It is essential for IOs because they generally lack coercive enforcement mechanisms and rely on voluntary compliance. Once IOs lose legitimacy in the eyes of member states, civil society, and business actors, their ability to manage international cooperation problems is threatened, and they confront the prospect of decline and death (Debre & Dijkstra, 2021; Eilstrup-Sangiovanni, 2020). Consequently, self-legitimation—a set of practices aimed at enhancing relevant audiences’ belief that an IO’s authority is normatively appropriate—is a core activity that IOs undertake to retain their operability.

A growing literature argues that IOs tend to intensify self-legitimation when they face public criticism and delegitimation efforts (Börzel & Zürn, 2021; Ecker-Ehrhardt, 2018a; Tallberg & Zürn, 2019). This argument suggests that the increasing authority of IOs has triggered a process of politicization in which IOs have become the object of public debate, normative contestation, and delegitimation (Rixen & Zangl, 2013; Zürn et al., 2012). In the process, a broader set of actors have recognized that they are affected by IO decisions (Rauh & Zürn, 2020, p. 586) and use the public sphere to confront IOs with normative demands—for instance, in the form of mass protest and critical debates in the media (De Vries et al., 2021; De Wilde et al., 2016). Since public contestation has the potential to undermine an IO’s legitimacy, IO representatives are expected to intensify self-legitimation reactively to prevent detrimental effects on their ability to govern. In this view, self-legitimation intensifies when an IO faces the prospect of legitimacy loss and serves to defend the organization’s legitimacy.

We challenge this perspective by developing a novel theoretical argument that perceives self-legitimation as a proactive practice aimed at building the legitimacy of a growing organization. Drawing on organizational sociology and management studies—literature emphasizing the varying opportunities and challenges that organizations confront during their life cycle (Van De Ven & Poole, 1995, p. 515; for systematic reviews, see Habib & Hasan, 2019; Smith et al., 1985)—we propose that self-legitimation intensity varies over organizational life stages. As IOs transition from one stage to the next due to formal changes in the nature and functional scope of their authority, their representatives adjust self-legitimation efforts. We hypothesize that IOs’ self-legitimation intensity varies systematically in accordance with changes in IO authority: during stages of authority growth, IO representatives intensify self-legitimation to attract additional support for their expanding activities, whereas they reduce self-legitimation during stages of authority decline and maintain a continuous level of self-legitimation during stages of stable authority. From this perspective, IO self-legitimation serves the preemptive purpose of seeking to enhance legitimacy proactively by winning the support of new audiences when IO authority expands.

We examine both arguments based on a novel dataset that captures the legitimacy claims of 28 regional IOs from 1980 to 2019. Using fixed-effects models and analyzing the distinct dynamic implications of both arguments, we find that life stages characterized by higher pooled authority are linked to heightened levels of self-legitimation. A brief vignette on the African Union shows how a large expansion of majoritarian decision-making through the Abuja Treaty in 1993 led actors to intensify their self-legitimation efforts in the absence of public politicization. In contrast, we find little or even counterintuitive evidence for the reactive self-legitimation argument. In contradiction of a core expectation of this argument, we observe that the politicization of pooled authority through public protests is associated with less intensive self-legitimation.

These findings have important implications for research on IO legitimation and behavior more broadly. First, they suggest that IO representatives’ preferences more strongly shape self-legitimation than the literature recognizes. While much of the current work on these issues argues that self-legitimation is primarily a reaction to relevant audiences’ normative demands (see Bernstein, 2011; Bexell et al., 2021), we show that IO representatives proactively seek to build and expand an IO’s legitimacy. IO representatives are active shapers of self-legitimation rather than mere conduits that give voice to audiences’ normative demands (see Lenz & Söderbaum, 2023, p. 910). This finding aligns with recent work on the proactive role of elites in shaping the legitimacy beliefs of audiences through cues (Anderson et al., 2019; Dellmuth & Tallberg, 2021). Whereas the independent role of IO representatives is increasingly being taken seriously in work on individual legitimacy beliefs, it has been neglected in research on IO legitimation.

Second, the finding that pooled authority—but not delegated authority and policy scope expansion—is strongly associated with self-legitimation intensity suggests that sovereignty concerns provide an important impetus for IO self-legitimation. It is widely recognized that pooling has particularly high sovereignty costs for member states because it deprives them of the ability to veto unwanted decisions (Lake, 2007, p. 232; Rixen & Zangl, 2013, p. 368). Since pooling authority, therefore, brings about particularly high legitimation requirements, IO representatives appear eager to obtain support for departing from consensus or unanimous decision-making. Yet IO representatives only try to legitimize the loss of national sovereignty if they do not fear public backlash. The finding that pooling under conditions of public protest is associated with lower self-legitimation intensity may indicate that representatives of IOs with high levels of pooled authority tend to use different strategies to manage politicization (Schimmelfennig, 2020). Instead of intensifying self-legitimation to convince critics, they shy away from the limelight.

Third, our findings shed light on normative debates about the democratic responsiveness of IOs. Research on the politicization of IOs often emphasizes the potential for democratizing global governance through growing public awareness of IOs and a less elite-centered mode of operation (Deitelhoff, 2020; Zürn, 2014). According to these arguments, politicization can be a corrective to insulated, elite-centered IOs by raising their representatives’ awareness of public demands and creating pressure to integrate them. Our findings show, in contrast, that representatives justify an IO’s authority less intensively when it becomes politicized. This suggests that they try to avoid the pressures of public contestation. Instead of debating with public critics, they try to lower an IO’s public profile by “dimming” self-legitimation efforts.

The article consists of three substantive sections. Section 2 conceptualizes discursive self-legitimation, outlines the two theoretical accounts of self-legitimation, and derives testable hypotheses. Subsequently, we introduce the research design (Section 3) and present empirical analyses (Section 4). In the conclusion, we summarize our findings and outline their implications for research on IO legitimation, life cycles, and institutional change.

2 The discursive self-legitimation of international organizations: Concept and theory

The established literature links increased IO authority to politicization, which mobilizes IO representatives to defend the legitimacy of their organization by scaling up self-legitimation. According to this view, self-legitimation is a reactive practice aimed at protecting an IO’s legitimacy in light of public criticism. We develop a theoretically distinct argument that conceives of self-legitimation as a proactive practice aimed at expanding the legitimacy of a growing organization by addressing new supporters, including civil society, the broader public, new members, and other IOs.

Emphasizing the expected benefits of self-legitimation, both arguments start from the premise that self-legitimation is strategic and costly. Specifically, they assume that (1) IO representatives seek to shape the legitimacy beliefs of relevant audiences because they perceive legitimacy as valuable for governance and that (2) doing so entails transaction and credibility costs. Seeking to convince audiences of an IO’s legitimacy requires representatives to collate information on relevant audiences and the norms that inform their legitimacy beliefs, to agree among themselves on specific legitimation practices,Footnote 1 and to dedicate a share of limited communication opportunities and scarce public attention to self-legitimation instead of, for instance, offering information on the organization or promoting specific policies. To manage these tasks, many IOs invest considerable financial and human resources into communication departments (Ecker-Ehrhardt, 2018b).

Apart from these transaction costs, there are costs associated with maintaining the credibility of self-legitimation. As Saward (2010, p. 50) notes, audiences “may accept, [but may also] reject, contest, or ignore the claim[s].” The audience dependence of legitimacy claims generates two types of credibility costs. For one, there is a constant risk that different legitimacy claims will be in tension with or contradict each other. This may be the case, for example, when IO representatives highlight both the functional efficiency of an IO and opportunities for democratic participation in it (on the general dilemma, see Dahl, 1994). Moreover, if audiences perceive self-legitimation to be merely cheap talk, without consequences for behavior, IOs may damage rather than enhance their legitimacy (see Ashforth & Gibbs, 1990). Thus, the credibility of self-legitimation also depends on costly follow-up practices, such as institutional and behavioral change (Bäckstrand & Söderbaum, 2018).Footnote 2

Both arguments, developed below, share the premise that since IO representatives seek to contain transaction costs and avoid credibility costs, they will avoid “overclaiming,” which translates into a baseline preference for limited self-legitimation.Footnote 3 We assume this preference to be constant across IOs and over time, allowing us to theorize variation in self-legitimation intensity by emphasizing varying benefits. As the expected benefits from self-legitimation increase, we propose, more intensive self-legitimation will follow. Conversely, when the expected benefits decrease, self-legitimation will decline.

In this section, we first outline the concept of IO self-legitimation and discuss what we can learn from mapping and explaining its intensity. Next, we develop the two theoretical arguments, which differ in terms of the expected benefits of self-legitimation.

2.1 Conceptualizing discursive self-legitimation

IO legitimation denotes a set of practices aimed at enhancing audiences’ belief that an IO’s authority is normatively appropriate. IO delegitimation, by contrast, describes practices that challenge belief in the appropriateness of an IO’s authority (Bexell et al., 2022; Gronau & Schmidtke, 2016; Tallberg & Zürn, 2019). This conceptualization implies that (1) practices of (de)legitimation rest on norms of appropriate behavior rather than considerations of individual self-interest or coercion (Hurd, 1999); (2) practices of (de)legitimation may refer to the normative appropriateness of different aspects of an IO, such as the organization as a whole, specific institutional reforms, or concrete policies; (3) various actors can engage in such practices, including international bureaucrats, civil society actors, and governments; and (4) these actors may use discursive, institutional, and behavioral practices to shape audiences’ legitimacy beliefs (Bäckstrand & Söderbaum, 2018).

This study focuses on a subset of IO (de)legitimation by examining the intensity of discursive self-legitimation. We define discursive self-legitimation as generalizable norm-based justifications of an IO’s authority made publicly by IO representatives. Self-legitimation thus involves those actors who represent an IO. It takes the form of public communication through annual reports, the communiqués of meetings of heads of state and government, or the speeches of the secretary-general. This communication is the outcome of negotiations among member states and international bureaucrats as the key representatives of IOs. Self-legitimation entails reason-giving (justifications), and these reasons rest on the norms of appropriate behavior that an IO claims to espouse or embody. Justifications are generalizable in that they refer to the IO as an organizational whole, not to individual components, actors, or specific policies.Footnote 4 The following statement by the European Council in the context of the Treaty of Nice provides a quintessential example of this phenomenon: “This new treaty strengthens the legitimacy, effectiveness and public acceptability of the institutions and enables the Union’s firm commitment to the enlargement process to be reaffirmed” (European Union, 2000, p. 1).

Our focus on the intensity of discursive self-legitimation—that is, the share of public communication that IO representatives devote to legitimation—breaks with previous scholarship, which has focused primarily on the specific norms that IO representatives use in their justifications, such as democracy or technocracy (Dingwerth et al., 2019, 2020; Rauh & Zürn, 2020; Zürn, 2018). This research assumes that legitimation is invariably important to IO representatives and presupposes, rather than investigating empirically, that relevant legitimation activity exists. If we do not know how intensively IOs self-legitimize and what drives variation, insights into the normative substance of self-legitimation are difficult to interpret. To understand how prominent such claims are likely to be in public (de)legitimation struggles and whether such claims will shape the legitimacy beliefs of relevant audiences, the intensity of discursive self-legitimation is a central variable. We can only understand the sources and effects of self-legitimation by considering whether an IO is an intensive or weak legitimizer. Compare, for instance, the self-legitimation discourse of the Organization of American States (OAS) and the Central American Integration System (SICA). According to our dataset, both organizations use democracy-based justifications in approximately 15 percent of their self-legitimation. Yet when we consider that the self-legitimation intensity of SICA is almost twice as high as that of the OAS, we can see that the number of self-legitimation statements using the norm of democracy is also almost twice as high for SICA relative to the OAS. This is a considerable difference, which is likely to shape the impact of both organizations’ self-legitimation. Only by considering self-legitimation intensity can we see this difference.

Specifically, the intensity of discursive self-legitimation matters because it shapes the public visibility of an IO and, ultimately, the impact of its legitimacy claims. Legitimacy claims that are few and far between are more easily overlooked by audiences. In contrast, more frequent claims are more likely to be noticed and to shape audiences’ legitimacy beliefs (Dellmuth & Tallberg, 2023). The literature on advertising shows that more frequent messages are more likely to be absorbed by target audiences and thus affect their choices (Schmidt & Eisend, 2015). While repeating a message reduces its informational content, it enhances the chances that recipients will change their attitudes and behavior. In other words, repeated exposure to a message is more likely to lead to attitudinal and behavioral change than infrequent exposure, especially when the message is easy to process (Anand & Sternthal, 1990). In this context, the intensity of self-legitimation is a critical variable because IO supporters and critics are competing for discursive hegemony in the public sphere (Bäckstrand & Söderbaum, 2018; Dingwerth et al., 2019). In a market of competing messages, the intensity of discursive (de)legitimation determines whose voice is heard.

2.2 The conventional wisdom: Reactive legitimacy protection

The most prominent attempt to account for variation in discursive self-legitimation suggests an “authority-legitimacy link” according to which legitimacy problems, and thus attempts to rectify them through self-legitimation, are an indirect function of an IO’s authority (Börzel & Zürn, 2021; Ecker-Ehrhardt, 2018a; Tallberg & Zürn, 2019). This argument suggests that IO representatives intensify self-legitimation in reaction to a situation in which the broader public begins questioning their organization’s authority. It assumes a baseline of limited self-legitimation, because IO representatives take legitimacy for granted in “normal” times and only invest in this governance resource when it comes under threat, thereby linking the growth of IO authority to the intensity of self-legitimation in three steps.

The first step builds on the observation that IOs have acquired increasing authority through the pooling of decision-making rights, the delegation of competencies to supranational agents, and the expansion of membership and policy scope (Hooghe et al., 2019; Lake, 2007; Pevehouse et al., 2020). States pool authority when they make collective decisions by majority vote rather than unanimity or consensus. This undermines national sovereignty because states cede their national veto over IO decisions (Blake & Lockwood Payton, 2015). They delegate authority when they empower independent agents within IOs to act on their behalf (Hawkins et al., 2006). When general secretariats set the policy agenda, international courts adjudicate disputes between member states, and international parliaments participate in adopting policies, IOs enjoy delegated authority (Bayerlein et al., 2020). This constrains national sovereignty because agents not directly controlled by states participate in making and enforcing collective decisions (Pollack, 1997). At the same time, authority increases with growth in an IO’s membership and policy portfolio because this widens an IO’s geographic and functional reach and thereby restricts member states’ ability to set policy independently (Börzel, 2005; Poast & Urpelainen, 2018). As a result of authority growth, many IOs possess the right to make binding decisions in the expectation of obedience, which is necessary for legitimacy problems to arise. As Tallberg and Zürn (2019, p. 586) note, “legitimacy only becomes an issue once an institution possesses authority. In the absence of authority, there would be no legitimacy problem.”

The second step highlights the politicization of international authority as a mediating factor between authority growth and self-legitimation intensity. It builds on Weber’s (1978) argument that legitimacy requires that those subject to political authority regard the rules issued by that authority as valid and binding, independent of the fear of being punished or of the material benefits that may emanate from rule-following (see also Hurd, 1999; Steffek, 2003). Hence, proponents of the authority-legitimacy link expect the growth of international authority to trigger a process of politicization in which IOs become the target of public debate, normative contestation, and delegitimation (Rixen & Zangl, 2013; Zürn et al., 2012). In this process, a broader set of actors, including citizens, civil society, business, and political parties, recognize that they are affected by IO decisions since “the more intrusive international institutions are, the more is at stake for those affected” (Rauh & Zürn, 2020, p. 586). As these actors lack direct access to IO decision-making, they mobilize and confront IOs with normative demands in the public realm—for instance, in the form of mass protest and critical debates in the media (De Vries et al., 2021; De Wilde et al., 2016). This public contestation holds the potential to undermine an IO’s legitimacy and ultimately threaten its ability to manage international cooperation problems successfully.

In the third step, IO representatives recognize politicization and its detrimental effects on their IO’s legitimacy and address it by intensifying discursive self-legitimation. Whereas they largely take legitimacy for granted and try to avoid the transaction and credibility costs of intensive self-legitimation in “normal” times, the expected benefits of intensive self-legitimation increase when an IO becomes politicized. To appease public critics, IO representatives therefore explain more frequently why their authority is normatively appropriate. They hope that more intensive self-legitimation can counteract outbursts of delegitimation in the public sphere. This argument understands self-legitimation as a reactive practice on the part of IO representatives that is driven by public criticism. When public criticism approaches a critical threshold and threatens to undermine an IO’s legitimacy, its representatives intensify self-legitimation to defend their organization’s legitimacy and to avert legitimacy loss (Sommerer et al., 2022).Footnote 5

In sum, the reactive self-legitimation argument proposes that IO authority growth triggers politicization and that IO representatives react to this politicization by intensifying discursive self-legitimation.Footnote 6 This argument yields the following conditional hypothesis.

  • H1: When IO authority growth is mediated by politicization, IO representatives intensify discursive self-legitimation.

2.3 The alternative argument: Proactive legitimacy expansion

Complementing this explanation, we develop an argument that conceives of discursive self-legitimation as a proactive attempt by IO representatives to expand the legitimacy of a growing organization. Rather than drawing on the authority-legitimacy link, our argument is inspired by organizational sociology and management studies, which connect self-legitimation to organizations’ life stages. While the theoretical starting point of our argument—change in IO authority—is similar to the conventional wisdom, the life-cycle perspective does not require public politicization as a mediating factor. Instead of understanding self-legitimation as a defensive reaction to public contestation aimed at protecting an IO’s existing legitimacy, we suggest that IO representatives proactively intensify self-legitimation to win the support of new audiences when IO authority increases. This argument moves beyond the reactive self-legitimation perspective by suggesting that an IO intensifies self-legitimation to expand its legitimacy when it enters new domains of activity or when its pooled and delegated authority increases. From this perspective, IO representatives seek to anticipate the legitimation requirements of their IO during different stages of an organizational life cycle and to adjust the intensity of self-legitimation accordingly. Hence, self-legitimation does not aim to prevent legitimacy loss but is directed at building a stock of legitimacy to support the new tasks and activities that come with increasing authority and to motivate compliance with an expanding set of IO rules.

Organizational sociology and management studies suggest that organizations progress through life stages (Banerjee & Hofmann, 2022; Banner & Gagné, 1995; Boin et al., 2010; Habib & Hasan, 2019), defined as temporally limited phases associated with distinct tasks, problems, and behaviors (Cohen, 2018, p. 52). In the birth stage of an organization, the emphasis is, for instance, on developing a product and marshalling resources. The stages of growth and success demand the establishment of a functional management structure and the attraction of additional resources, whereas decline often requires renewal and innovation to avoid death (Lester et al., 2003; Quinn & Cameron, 1983; Van De Ven & Poole, 1995).

According to this literature, legitimacy management is a critical challenge that varies with organizational life stages (Ashforth & Gibbs, 1990, p. 182; Suchman, 1995, p. 586). Following Stinchcombe’s (1965, p. 148) suggestion that new organizations suffer a “liability of newness,” Freeman et al. (1983, p. 692) demonstrate, for instance, that young organizations die more frequently “because they depend on the cooperation of strangers, have low levels of legitimacy, and are unable to compete effectively against established organizations.” Mature organizations, in contrast, can rely on entrenched networks of support, a stock of legitimacy, and experience in withstanding competition from rivals. As a result, the expected benefits of self-legitimation vary in accordance with the life stages of creation, growth, continuity, and decline (Yang & Aldrich, 2017). In the early phases of an organization’s life, its representatives expect substantial legitimacy gains from proactively convincing relevant audiences that the organization is appropriate. This endeavor entails attracting new constituencies and persuading other organizations to lend support (Suchman, 1995, p. 587). Once the organization becomes established and enjoys a modicum of support, the expected benefits of self-legitimation decline, and representatives tend to direct resources elsewhere. While they still nourish existing legitimacy beliefs, their “legitimation activities become increasingly routinized” during this stage (Ashforth & Gibbs, 1990, p. 183). During stages of stability, self-legitimation intensity is thus likely to be constant.

This situation changes as soon as an organization expands its activities or authority. Like the creation stage, periods of growth induce organizational representatives to intensify self-legitimation to generate additional support for the new activities and increasing authority; the expected gains from self-legitimation increase again. Here, discursive self-legitimation may be useful for propagating an organization’s strengths, disguising weaknesses, and expressing intrinsic merits (Halliday et al., 2010). In contrast to growth, stages of organizational decline tend to be associated with less self-legitimation. For one, decline may obviate the benefits of self-legitimation because external actors are no longer required to provide the same level of support and resources. Moreover, declining organizations may be unable to engage in forceful self-legitimation or may decide to prioritize other activities, such as inward-looking reforms and organizational renewal (Haftel et al., 2020). In sum, organizational sociology and management studies suggest that self-legitimation expands when organizations grow, stages of continuity are associated with stable levels of self-legitimation, and decline is linked to contracting self-legitimation.

We transfer these insights to IOs’ discursive self-legitimation by proposing that changes in an (1) IO’s function and purpose, (2) organizational processes for achieving these goals, and (3) the realm and scope of its operations determine its life stages (Scott, 2001).Footnote 7 These organizational characteristics are enshrined in an IO’s formal mandate. We can think of them as an immanent program governing an IO’s behavior and development (Hooghe et al., 2017; Van De Ven & Poole, 1995, p. 515). More specifically, we suggest that an IO’s authority (pooled and delegated) prescribes the nature of organizational processes, while the territorial (membership) and functional scope of its activities (policy areas) specify the organization’s purpose, goals, and scope. These structural characteristics determine the respective life stage, because organizational stages are “distinguishable in form or function” (Van De Ven & Poole, 1995, p. 525). We conceive of a stage as stable and persisting if this fundamental structure remains unaltered. The metamorphosis of an IO to a qualitatively different stage occurs when its member states decide to increase or decrease its authority or its territorial or functional scope (Drori et al., 2009, p. 717; Kazanjian, 1984, p. 22; Smith et al., 1985, p. 801).

It follows that the intensity of discursive self-legitimation evolves in line with change in the nature and scope of an IO’s authority. When an IO increases its policy scope and membership or deepens authority through pooling or delegation, its representatives expect increased benefits from explaining more frequently why this organizational growth is normatively appropriate. By contrast, decreases in IO authority or scope mark a stage of organizational decline (Zürn, 2018, p. 89) and thus tend to be linked to less intensive discursive self-legitimation. Overall, these arguments suggest the following hypothesis:

  • H2: When IO authority increases (declines), IO representatives intensify (reduce) discursive self-legitimation.

The two arguments are complementary. We may, for example, observe a cumulative relationship in which authority growth is associated with more intense self-legitimation while politicization further increases IO representatives’ urge to explain the normative appropriateness of their organization to a critical public (Steenbergen et al., 2007).

3 Research design

This section explains our sample of IOs, outlines the content-analytical method used to map the discursive self-legitimation of IOs, and operationalizes dependent and independent variables.

3.1 Measuring IOs’ discursive self-legitimation

We measure self-legitimation intensity by drawing on the LegRO dataset (Schmidtke et al., 2023), which covers the discursive self-legitimation of 28 regional IOs from 1980 to 2019.Footnote 8 In line with common usage, we define a regional IO as a formal international organization composed of three or more geographically proximate states (Haftel, 2013, p. 394; Pevehouse et al., 2020). Our sample encompasses several IOs from four major world regions (Africa, the Americas, Asia–Pacific, and Europe) and cross-regional IOs (see Appendix A1 for the IO sample). Since these IOs are among the most resource-endowed organizations post-World War II (Hooghe et al., 2017), we expect their self-legitimation to matter in global governance. The sampling period is sufficiently long to capture significant changes in the selected IOs’ authority and politicization.Footnote 9

Due to their geographical demarcation, regional IOs offer distinct analytical advantages over global IOs that help us to examine our hypotheses. The IOs in our sample vary strongly with regard to politicization and authority, whereas global IOs are more homogenous in both dimensions. Our sample includes not only IOs that frequently face public delegitimation and protest, such as the EU and the Asia–Pacific Economic Cooperation (APEC), but also IOs that operate almost entirely under the radar of public scrutiny, including the Central American Integration System (SICA) and the Common Market for Eastern and Southern Africa (COMESA). Similarly, the sample includes IOs from different world regions that have experienced stability, decline, and growth in their authority. Across the four variables of pooling, delegation, policy scope, and membership, the sample includes 269 cases of authority growth and 138 cases of authority decline. For instance, the African Union (AU) has seen multiple changes in its authority, including declining and increasing levels of delegation, increasing levels of pooling, growth in policy scope, and member-state withdrawal and accession. The EU and SICA have experienced growth and decline in pooling and delegation and growth in issue and membership scope. The Commonwealth of Independent States (CIS) has undergone growth and decline in delegation and membership and growth in pooling and issue scope. Moreover, our focus on regional IOs complements existing work on IO legitimation, which has tended to focus on global IOs.

We have operationalized our definition of discursive self-legitimation intensity as the share of public discourse that IO representatives devote to enhancing audiences’ legitimacy beliefs as follows. First, we selected annual reports, typically issued by an organization’s general secretariat, and the final communiqués of heads of state and government meetings as representative sources of IOs’ public discourse. These documents have several advantages that make them preferable to other sources of IOs’ public discourse, such as press releases, speeches by IO representatives, or social media communication. First, member states and bureaucrats take the drafting of these documents seriously and invest considerable time and resources. The drafting process generally includes various IO bodies, the bureaucracy’s head (and other departments), and member-state representatives (Ecker-Ehrhardt, 2018a). Second, the documents are published at similar intervals, and their structure and intent are similar. Third, most documents are publicly available across organizations and over long periods. As a result, the selected documents are rich sources of self-legitimation discourse (Halliday et al., 2010, p. 84), and their broad availability and comparability are important advantages over other potential sources.Footnote 10

Second, we coded sample sections of these documents. We defined the paragraph as the “natural language” coding unit (Däubler et al., 2012). The number of coded paragraphs (n = 32,675) per IO-year (n = 974), thus, proxies the total amount of IO public discourse. Given the length of the documents in our corpus, we selected paragraphs that are particularly interesting from a legitimation perspective because they express commitments to basic principles and contain key elements of the organization’s philosophy, self-conception, and desired public image rather than merely documenting an organization’s activities. We followed the sampling procedure developed by Dingwerth et al. (2020), which is based on an initial screening of approximately 2 percent of all selected documents. The screening showed that self-legitimation discourse is clustered in a limited number of clearly identifiable sections—namely, general overviews, summaries, forewords, introductions, and conclusions. Hence, for most documents, we coded paragraphs from similarly structured introductory sections across IOs and time. Since the number of paragraphs in these sections varies, we calculated a 25 percent range around the mean number of paragraphs in these sections. As a result, we coded a minimum of 16 and a maximum of 28 paragraphs per document and, thus, a minimum of 32 and a maximum of 56 paragraphs per IO-year.Footnote 11

Third, we operationalized the number of distinct self-legitimation statements that IO representatives make in these selected paragraphs by identifying generalized justifications of an IO’s authority. We coded these statements based on two stylized grammars: the OES and the OIS grammar (see Table 1). O stands for legitimation object, S for normative standard, E for evaluation, and I for identity (Schmidtke & Nullmeier, 2011).Footnote 12 The legitimation object is the organization in general, excluding officials or policies. In the OES grammar, we identify positive evaluations of the legitimation object by searching for formulations such as “good,” “great,” or “improve.” The OIS grammar describes propositions that highlight an organization’s identity, purpose, and guiding principles (Dingwerth et al., 2020; Koopmans & Statham, 1999). Such statements are descriptive and do not necessarily contain an explicitly positive tone.

Table 1 Legitimation grammar and examples

To qualify as a distinct legitimation statement, the generalized justification of an IO’s authority must highlight a unique normative standard. Consequently, we counted one legitimation statement per highlighted standard but no additional statement(s) if a standard appears multiple times.Footnote 13 The examples in Table 1 illustrate this approach, as we consider the share of discourse devoted to legitimation to be higher in the paragraph from Asia-Pacific Economic Cooperation 2007 annual report than in the paragraph from the European Council’s 2018 communiqué. The European Council repeats the economic welfare standard four times by highlighting inclusive growth, job creation, investment, and global competitiveness but does not use additional normative standards. APEC, by contrast, explains that the organization aims to facilitate economic welfare and enhance a sense of community in the region. Our coding of normative standards builds on a list of 31 norms. The list includes liberal and technocratic norms that other scholars have highlighted as essential legitimation standards for IOs (Binder & Heupel, 2021; Dellmuth et al., 2019; Dingwerth et al., 2019; Tallberg & Zürn, 2019), communitarian norms that feature in more recent research (Spandler & Söderbaum, 2023; Wajner, 2022), and standards that are prominent in our empirical material (see Appendix A2.1, for further details).Footnote 14

Based on this procedure, we measure self-legitimation intensity as the number of legitimation statements divided by the number of coded paragraphs. For example, for an IO-year in which we coded 32 paragraphs, 16 legitimation statements result in a self-legitimation intensity of 0.5. As coded paragraphs may contain more than one legitimation statement, this variable has an absolute minimum of zero and a theoretical maximum of 31. Since the annual scores for self-legitimation intensity are sensitive to individual coder decisions and the drafting process of the documents analyzed, we use a two-year rolling mean in our estimations.

3.2 Independent variables

The reactive self-legitimation argument (H1) and our proactive self-legitimation hypothesis (H2) draw on the formal authority of IOs to account for variation in self-legitimation intensity. Hence, the first step of our empirical strategy for examining these arguments draws on variables included in an updated version of the Measure of International Authority (MIA) dataset (Hooghe et al., 2017, 2019; updated for 2011–2019 by Haftel & Lenz, 2022). These data capture three central dimensions of IO authority that we expect to be associated with self-legitimation intensity.

First, we refer to pooling, defined as the extent to which member-state bodies move from the unanimity principle towards various forms of majority voting across six decision-making areas: membership accession, membership suspension, policymaking, budgetary allocation and noncompliance, and constitutional reform. The variable is an aggregate index that ranges from zero to one (from low to high), with an empirical maximum of 0.56, reached by the AU since its 2003 reform. Second, we use delegation, defined as the extent to which member states empower supranational agents to set the agenda and make decisions across the six areas mentioned above. This aggregate index ranges from zero to one (from low to high), with an empirical maximum of 0.65 reached by the European Union (EU) with the enactment of the Lisbon Treaty in 2009. Third, we apply policy scope, counting the number of policies from a list of 25 policy areas in which an IO holds competence.Footnote 15 This variable ranges from zero to 25 (from narrow to comprehensive), with an empirical maximum of 25, reached by the EU with the establishment of the European Stability Mechanism in 2012. All three variables measure authority at the time of the entry into force of a new legal agreement, which in the case of policy scope may be based on secondary legislation that does not require domestic ratification (whereby the moment of adoption and entry into force are identical).Footnote 16

Finally, we operationalize membership scope with the help of the Correlates of War IGO Version 3.0 dataset, which provides information on the formal membership of the IOs in our sample (Pevehouse et al., 2020). For 2015 to 2019, we have added missing membership data with the help of IO websites. In principle, this count variable ranges from three to 218 states in the system, with an empirical maximum of 55, reached by the Organization for Security and Co-operation in Europe (OSCE) with the accession of Montenegro in 2006.

The main difference between reactive and proactive self-legitimation is that the former expects the public politicization of IOs to function as a mediating factor that links growing IO authority to self-legitimation intensity. In contrast, we propose a direct association between changing IO authority and self-legitimation intensity. To model the differences between these two arguments, we include a measure of publicly visible protest against IOs and interact it with the four measures of IO authority. We use this measure because protest constitutes the most severe form of criticism and delegitimation in the public sphere. We have generated this variable with the help of keyword searches for the IO name or acronym and the terms “protestor” or “demonstrator” in the Major World Newspapers corpus of the online newspaper database LexisNexis. We use the logarithmized count of hits per IO-year to limit the effect of outliers, such as the EU and APEC (for a similar approach, see Dingwerth et al., 2020; Tallberg et al., 2014). If the reactive self-legitimation argument is correct, we should observe statistically significant positive associations between self-legitimation intensity and the two-way interaction terms of protest and authority. By contrast, our proactive self-legitimation argument finds empirical support if we observe statistically significant positive associations between self-legitimation intensity and the four authority variables.

3.3 Controls

Finally, we include four control variables. Previous work has shown that IOs intensify discursive self-legitimation when facing exogenous shocks or crises (Reus-Smit, 2007; Widmaier et al., 2007). We include the following variables to capture this idea. First, turning to major political junctures in world politics over the past 40 years (Buzan & Lawson, 2014, p. 446), we distinguish between the pre-and post-Cold War eras by separating our data into that from before and from after 1989. Second, we control for economic crises that hit an IO’s member states (Davis & Pelc, 2017; Haftel et al., 2020). To this end, we utilize data gathered by Laeven and Valencia (2018) that codes whether a state was in a banking, currency, or sovereign debt crisis in a given year. Assuming that a crisis is an unusual situation that threatens significant harm to a group of actors and compels a response under time pressure and uncertainty (Lipscy, 2020, p. E99; Sommerer et al., 2022, p. 7), we code an IO-year as characterized by an economic crisis when the count of member states in an economic crisis was in the 95th percentile for a given IO and decade. Finally, we apply a similar logic to operationalize security crises. Using the Correlates of War Militarized Interstate Dispute (MID5) dataset (Palmer et al., 2022), we count interstate wars involving IO member states and code IO-years as security crises if this number was in the 95th percentile for a given IO and decade.

Moreover, we control for variation in the availability of our coded material. Ideally, our data would build on one annual report and one communiqué per IO-year. In certain cases, we were only able to code one of these documents. As our dependent variable pools information from both documents, having only one of them potentially decreases self-legitimation intensity (type of documents). Table 2 provides descriptive statistics for the dependent and independent variables.

Table 2 Descriptive statistics

4 Empirical analysis

How intense is the discursive self-legitimation of IOs? How do IOs alter the intensity of self-legitimation over time? And how can we account for this temporal variation? Our empirical analysis of these questions proceeds in two steps: First, we present descriptive patterns of discursive self-legitimation intensity. Second, we examine our hypotheses using OLS regression with IO fixed effects.

4.1 Patterns of discursive self-legitimation intensity

Descriptively, we make three central observations. First, discursive self-legitimation is a persistent feature of IO practices across all organizations in our sample and the entire observation period. Second, despite the continuous flow of discursive self-legitimation, the intensity varies considerably across IOs and over time. Third, substantive changes in self-legitimation intensity are rare. Despite some annual fluctuations, most IOs in the sample only shifted one or two times to considerably more or less intense self-legitimation over almost 40 years.

Figure 1 illustrates the general trend in our data. The bars show the annual average of self-legitimation intensity, and the curve represents the trend in the data.Footnote 17 The figure reveals that the intensity of discursive self-legitimation increased between 1980 and 2019. Specifically, the data suggest three distinct temporal phases of discursive self-legitimation. Stagnant or even contracting levels of self-legitimation characterized the first phase during the 1980s. Beginning with an average of 0.4 in self-legitimation intensity at the beginning of the 1980s, the IOs in our sample reduced their self-legitimation to an absolute minimum during the mid-1980s. Throughout the 1990s and early 2000s, this trend reversed, and we observe a phase marked by intensification. During this period, average self-legitimation intensity reached an absolute maximum of almost 0.6 in 2005. Finally, over the second half of the 2000s, self-legitimation intensity stabilized. While there was still annual fluctuation, the upward trend in the data flattened, indicating a period of continuity. This development suggests that IO representatives did not radically adapt the intensity of discursive self-legitimation on an annual basis. Instead, we find a pattern resembling a punctuated equilibrium, in which self-legitimation was comparably low during the 1980s, intensified considerably during the 1990s, and stabilized again at a higher level towards the end of the observation period.

Fig. 1
figure 1

Self-legitimation intensity of 28 regional IOs over time, 1980–2019

We further examine this pattern by disaggregating the data. Figure 2 displays smoothed trendlines for the development of self-legitimation intensity within IOs. Some annual fluctuation notwithstanding, substantive changes from high to low intensity are relatively rare. This observation further underlines that IO representatives did not radically adapt self-legitimation annually. Instead, all IO plots show that self-legitimation developed in a limited number of relatively long phases.

Fig. 2
figure 2

Phases in self-legitimation intensity across 28 regional IOs, 1980–2019.

Note: The four groups are the following: Group 1: single phase (increase, stable trajectory, or decrease); Group 2: two phases (increase, followed by stable trajectory or decrease); Group 3: two phases (stable trajectory, followed by increase or decrease); Group 4: more than two phases (wave-like trajectory). Full names of IOs mentioned in the figure (in the order in which they appear): Commonwealth of Independent States (CIS), East African Community (EAC), Gulf Cooperation Council (GCC), European Free Trade Association (EFTA), Common Market of the South (Mercosur), Nordic Council (NordC), Southern African Customs Union (SACU), Organization of Eastern Caribbean States (OECS), Pacific Islands Forum (PIF), Asia–Pacific Economic Cooperation (APEC), Organization of African Unity/African Union (AU), Andean Pact/Andean Community (CAN), Economic Community of West African States (ECOWAS), European Union (EU), Organization of American States (OAS), Organization for Security and Co-operation in Europe (OSCE), Central American Integration System (SICA), Association of Southeast Asian Nations (ASEAN), Caribbean Community (CARICOM), Central African Economic and Monetary Union (CEMAC), League of Arab States (LoAS), Inter-Governmental Authority on Development (IGAD), Organization of Arab Petroleum Exporting Countries (OAPEC), Shanghai Cooperation Organization (SCO), Council of Europe (COE), Common Market for Eastern and Southern Africa (COMESA), South Asian Association for Regional Cooperation (SAARC), Southern African Development Community (SADC)

Furthermore, IO-specific patterns suggest the presence of four major groups of organizations in our sample. First, we find a group of nine organizations from all world regions whose self-legitimation trajectory evolves in a single phase of increasing, stable, or declining intensity. The CIS, the EAC, and the GCC displayed continuous downward trends; the intensity of self-legitimation on the part of the Nordic Council, the EFTA, Mercosur, and the SACU (neglecting the outlier in the first year) is fairly stable; and the OECS and the PIF increased continuously over the entire observation period. The second group is characterized by intensifying self-legitimation during the observation period’s early years, followed by a second period in which legitimation intensity either declined or remained stable – this group evolves in two phases. It also encompasses heterogeneous IOs from different world regions, except for the Asia–Pacific region (APEC (cross-regional), AU, CAN, ECOWAS, EU, OAS, OSCE, and SICA). The third group comprises organizations from all regions except for Europe. These IOs also witnessed two phases, progressing from stable to increasing (ASEAN, CARICOM, CEMAC, LoAS, OAPEC, and SCO) or declining levels of self-legitimation intensity (IGAD). Finally, only four organizations displayed more than one substantive shift in self-legitimation intensity by displaying wave-like trajectories (COE, COMESA, SAARC, and SADC). These patterns suggest that substantive changes in self-legitimation intensity are relatively rare events in the life cycle of IOs.

Finally, if we zoom in on the specific times of change in self-legitimation intensity, IO-specific patterns confirm the observation from the aggregate picture that the first half of the 1990s and the second half of the 2000s constituted important turning points for many IOs in our sample. Of the 25 organizations in the sample that experienced one or two transitions, almost 70 percent progressed from one phase to the next during these periods. This group is heterogeneous and consists of IOs from different world regions with distinct phases in their self-legitimation intensity. The remaining set of organizations exhibited a shift in self-legitimation intensity during the first half of the 2010s.

4.2 Multivariate analysis

We specify a set of OLS regressions with IO fixed effects to test the hypotheses on reactive and proactive self-legitimation (Mummolo & Peterson, 2019). Both methodological and theoretical considerations have driven the decision to use this approach. First, our unbalanced panel includes 947 observations clustered within IOs, for which we have varying numbers of observations because not all IOs in the sample existed over the entire observation period. Second, our theoretical arguments focus on change over time and less on cross-sectional variation. We use the following fixed-effects model:

$${SLI}_{it}={\alpha }_{i}+{\beta\;pooling}_{it}+{\beta\;delegation}_{it}+{\beta\;policy\;scope}_{it}+{\beta\;membership\; scope}_{it}+{\beta\;protest}_{it}+{\beta\;pooling}_{it}*{protest}_{it}+{\beta\;delegation}_{it}*{protest}_{it}+{\beta\;policy\; scope}_{it}*{protest}_{it}+{\beta\;membership\;scope}_{it}*{protest}_{it}+{\varepsilon }_{it},$$

in which self-legitimation intensity (SLI) and the independent variables pooling, delegation, policy scope, and membership scope are observed for each IO i over multiple years t, and a mutually exclusive intercept shift, \(\alpha\), is estimated for each IO i to capture the distinctive, time-invariant features of each IO. This results in an estimate \(\beta\) that is adjusted for the influence of inter-IO time-invariant confounders.

We estimate a series of models using the two-year rolling mean of self-legitimation intensity as the dependent variable (see Table 3). Modeling the reactive self-legitimation argument, M1 includes two-way interaction terms between IO authority and protest. M2 replicates this analysis and includes control variables. Given how we measure IO authority and the temporal sequence suggested by the reactive argument, both models lag explanatory variables by three years. In line with the research on politicization, we expect mass mobilization and large-scale protests to occur when the consequences of authority growth become apparent to a broader segment of the population (Schmidtke, 2016). Since the entry into force of an authority growth requires some time to manifest in concrete policy changes, mass mobilization and the organization of large-scale protests take time, as does the reaction from IO representatives, who take time to both prepare and agree on a joint response and to publish the subsequent annual report or communiqué. M3 turns to proactive self-legitimation, estimating the direct association between IO authority and self-legitimation intensity. M4 replicates this analysis and includes control variables. Both models use authority variables without a temporal lag to capture the idea that authority growth and proactive self-legitimation unfold in a close temporal sequence.Footnote 18 All models include the unlagged type of documents control variable. Models 2 and 4 use a one-year lag for the external shock control variables to avoid simultaneity.

Table 3 Changes in IOs’ self-legitimation intensity

We first turn to the reactive self-legitimacy argument (H1). Empirical support for this hypothesis would require positive and statistically significant coefficients for the two-way interaction terms between protest and IO authority and a positive and statistically significant coefficient for protest. Models 1 and 2 do not support these expectations. Neither the interactions nor protest has a consistently positive and statistically significant coefficient. Instead, we observe negative signs for protest in M2, for the interaction between protest and delegation in M1 and M2, and a statistically significant negative coefficient for the interaction between pooling and protest in both models. The latter finding contradicts the reactive self-legitimation argument and could indicate that representatives of IOs with high levels of pooled authority tend to use different strategies to manage politicization (Schimmelfennig, 2020). Instead of intensifying self-legitimation to convince critics, IOs with high pooling levels shy away from the limelight when the IO becomes politicized, maybe to shield their organization from popular dissensus constraints. Overall, these results suggest that the findings from studies of global IOs, which support the reactive legitimacy-protection perspective (Dingwerth et al., 2020; Rauh & Zürn, 2020), do not transfer to regional IOs.

By contrast, models 3 and 4 support the proactive self-legitimation argument (H2). All four authority variables have the expected positive sign, the coefficients for pooling are highly statistically significant across both models, and the coefficient for membership scope is statistically significant in M3. This shows that pooled authority is clearly and consistently associated with self-legitimation intensity. IOs that shift from unanimity or consensus to majority voting tend to intensify discursive self-legitimation. The association between membership scope and self-legitimation is less consistent as the respective coefficient is not statistically significant in M4. However, the statistically significant coefficient in M3 provides evidence of the importance of membership scope. By contrast, we do not find statistically significant coefficients for delegation and policy scope. These results imply a direct link between changing IO authority and discursive self-legitimation, but this association is mainly related to the “inter-state component” of international organizations. This may suggest that proactive self-legitimation primarily aims to mobilize support from actors concerned with the sovereignty costs of pooled authority. Scholars agree that pooling has exceptionally high sovereignty costs for member states because it deprives them of the ability to veto unwanted decisions (Lake, 2007, p. 232; Rixen & Zangl, 2013, p. 368). IO representatives may therefore reason that pooling authority presents particularly high legitimation requirements, which they meet by intensifying their self-legitimation discourse.

The AU illustrates this interpretation of the data. In 1963, the organization was created as the Organization of African Unity to promote decolonization and solidarity among members. As newly independent nations, most member states insisted on “state sovereignty, territorial inviolability, and non-interference” (Rechner, 2006, p. 547). Hence, the level of pooled authority was minimal until 1993, when the Abuja Treaty established the African Economic Community and formalized majority voting rules in several policy areas (Hooghe et al., 2017, pp. 167–188). This reform almost tripled the level of pooled authority. The organization also welcomed South Africa as a new and powerful member only a few months later. Intensive self-legitimation accompanied this considerable growth in authority, driving the upward trend shown in Fig. 2.

The respective statements reveal that self-legitimation (a) primarily addressed future supporters, including African citizens, member states, and other IOs, by (b) highlighting how the Abuja reform would contribute to the organization’s success in the future, indicating proactive attempts by AU representatives to expand the IO’s legitimacy at the beginning of a new stage of its evolution. In his introductory note to the 60th Council of Ministers, Secretary-General Salim Ahmed Salim applauded, for instance, that “with the entry into force of the [Abuja] Treaty, Africa is now geared towards its actual implementation,” that the “Treaty is part of great determination on the part of Africa, to re-seize the initiative and foster unity as a means of accelerating economic development,” and that “henceforth, our countries will need to work even more closely to promote cooperation among themselves” (Organization of African Unity, 1994, pp. 7–8). Similarly, he expressed the hope that South Africa would bring “its technological know-how and economic potential in our collective quest for the development of our Continent [and] to see South Africa strengthen the hand of Africa in the international arena” (Organization of African Unity, 1994, p. 5). Ahmed Salim emphasized that African citizens were critical to the success of the reformed organization when he concluded that:

Ultimately, the people of Africa will find the Community relevant, only if they are involved and made to have a stake in it [and that] our countries should move more resolutely, both in broadening the areas of cooperation between them and in sensitizing the people of Africa on the African Economic Community. Taking the Community to the people of this Continent, and placing them at the centre of its development must now be the primary objective and the theme of the future (Organization of African Unity, 1994, p. 9).

During the treaty signing ceremony in 1991, the heads of state and government had already indicated that support from the international community was essential to them. They called “upon all member states of the respective regional communities and the other relevant Inter-governmental Organizations to lend full support and cooperation” and urged “the International Community, the UN System, UNDP and other multilateral funding Agencies and institutions to lend their full technical and financial support to the establishment of the African Economic Community” (Organization of African Unity, 1991, p. 30).

Overall, the analysis shows that pooled authority is important for discursive self-legitimation. In contrast to the reactive self-legitimation argument, which attributes intensive self-legitimation to the public politicization of growing IO authority, this association seems to operate independently of popular dissensus. Hence, discursive self-legitimation should not be interpreted exclusively as a defensive response to public demands but also as a proactive attempt to expand legitimacy by convincing new audiences that may be concerned about the sovereignty costs of pooling.

We also estimate a series of alternative models to demonstrate the robustness of our results (see Appendix A4). First, we have controlled for the potential effects of our fixed-effects specification by estimating a multilevel regression model with a random IO intercept, which allows for both cross-sectional and temporal variation. The results provide equally strong support for the finding that pooling is systematically associated with self-legitimation intensity. The coefficients for pooling and its interaction with protest have the same signs and significance as in Table 3. Furthermore, this model offers additional evidence in support of the proactive model as we find positive, statistically significant coefficients for policy scope (M3) and membership scope (M4).

Second, we have replicated the analysis with three alternative versions of our dependent variable. The first two robustness checks use self-legitimation intensity without superimposing the two-year rolling mean and with a three-year rolling mean, respectively. We also run a logistically transformed dependent-variable model because estimations with truncated dependent variables can lead to prediction errors (Thomas, 1997). Our findings on pooling and its interaction with protest are robust to all three operationalizations of self-legitimation intensity. Yet the results for membership scope are sensitive to these specifications. While the respective coefficients are not statistically significant without the rolling mean or with a logistically transformed two-year rolling mean, they are statistically significant across M1 through M3 when we use the three-year rolling mean.

Third, we have estimated the models using a set of alternative lag structures. For one, we examine the implication of our proactive self-legitimation argument that IO representatives might intensify self-legitimation before an authority growth enters into force. To this end, we have estimated the proactive models (M3 and M4) with a one-year lead for the authority variables. The results are almost identical, except for lower statistical significance for membership scope in M3. Because it suggests that IO representatives adapt self-legitimation even before a formal authority growth, this specification provides additional evidence in support of the proactive self-legitimation argument. To account for potentially shorter latency, we calculated the reactive self-legitimation models (M1 and M2) using (1) a two-year lag for the authority variables and protest, and (2) a three-year lag for the authority variables with a two-year lag for protest. Both models support our main findings for pooling and its interaction with protest, as the results are almost identical. In addition, these models provide limited evidence in support of the reactive self-legitimation argument, showing a positive, statistically significant coefficient for the interaction of policy scope and protest in M2. However, the statistical significance is at the 5 percent level and inconsistent across M1 and M2.

Finally, we have estimated the models with a broader operationalization of politicization, using the media salience of an IO.Footnote 19 While protests are arguably the most severe form of public delegitimation, we may consider politicization to be a broader phenomenon that includes other forms of discursive and non-discursive delegitimation efforts. Building on the assumption that public delegitimation will generally enhance the visibility, or salience, of an IO in the media, this measure builds on keyword searches in the Major World Newspapers corpus of LexisNexis but uses the logarithmized count of all articles that include the IO name or acronym. The results are similar to those reported in Table 3 and underscore our findings on pooling and proactive self-legitimation. However, the interaction of delegation and salience has a positive coefficient under this specification and is statistically significant at the 5 percent level in M1, thus providing some evidence in support of the reactive self-legitimation argument.

In sum, these alternative models show that the direct association between pooled authority and self-legitimation intensity is robust across different model specifications, variable operationalizations, and lag structures, lending strong support to the proactive self-legitimation argument. In contrast, evidence supporting the reactive self-legitimation argument is limited and dependent on particular estimation decisions. In some cases, it is even contradictory.

5 Conclusion

This article has examined the intensity of IOs’ discursive self-legitimation. Drawing on a new dataset that captures the legitimacy claims of 28 regional IOs from 1980 to 2019, we have shown that self-legitimation intensity has increased over time. After a slight decline during the 1980s, we can observe self-legitimation growth during the 1990s and 2000s and relative continuity during the 2010s. We set out to explain this variation of self-legitimation intensity by developing a novel theoretical argument that draws on organizational sociology and management studies and highlights the life stages of organizations. In contrast to established work that conceives of self-legitimation as a defensive reaction to the public politicization of an IO’s authority, we propose that self-legitimation follows IO representatives’ attempts to proactively enhance the legitimacy of an expanding organization. It follows that while growth in authority should activate IO representatives to engage in the proactive self-legitimation of their expanding organization, authority decline leads IO representatives to decrease self-legitimation.

Our empirical analysis supports this relationship between IO authority and self-legitimation intensity. Specifically, IOs that are shifting from unanimous decision-making towards majority voting tend to intensify their discursive self-legitimation. These results suggest that IOs with different levels of authority approach self-legitimation differently, irrespective of the degree of public politicization. Consequently, IO self-legitimation is not primarily a reaction to public delegitimation. Instead of defending the legitimacy of an IO against criticism, representatives appear to engage in proactive self-legitimation in reaction to the most sovereignty-constraining form of authority expansion: a departure from consensus or unanimous decision-making.

These findings have important implications for research on the legitimacy and legitimation of IOs. Most fundamentally, our results demonstrate that growing IOs are more active legitimizers and thus have better chances of convincing audiences of their legitimacy, becoming more focal in their issue area, and ultimately attracting more support for their work. This dynamic could imply a mutually reinforcing relationship between authority growth and self-legitimation in which organizational growth is associated with more intensive legitimation, thus generating higher levels of support and potentially another growth stage. This perspective contrasts with much of the literature on IO legitimation, which describes self-legitimation as a reaction to public criticism, declining support, and legitimacy crises (Gronau & Schmidtke, 2016; Rauh & Zürn, 2020; Tallberg & Zürn, 2019). In this conventional view, IO self-legitimation is a defensive activity that IO representatives use to stabilize legitimacy, avoid decline, and ultimately prevent the death of their organization.

This article also has broader implications for the emerging research agenda on the life stages of IOs (Gray, 2018; Hirschmann, 2021; Young, 1982). While much of the theoretical and empirical International Relations literature perceives IOs as a highly rigid form of international cooperation (Abbott & Faude, 2021; Vabulas & Snidal, 2021) and thus lacks a systematic approach to understanding IOs beyond their moments of creation and on through the course of their existence, recent work argues that IOs experience ebbs and flows throughout their lifetime that result in fundamentally different behaviors, output, and performance (Gray, 2020). Our findings contribute to this growing research agenda by uncovering the relationship between IO life stages and legitimation. In line with the basic arguments of an emerging theory of IO life stages, our analysis shows that distinct stages in the life cycles of IOs are associated with different approaches to legitimation. While we cannot generalize our results beyond legitimation, this insight suggests that IO behavior does indeed vary with the various life stages.