Institutional overlap and the survival of intergovernmental organisations

Abstract

A burgeoning literature emphasises the role of favourable external conditions and institutional design as key drivers of the survival of intergovernmental organisations. This article focuses on institutional overlap as a hitherto overlooked determinant of IGO survival. Studies on institutional complexity posit that overlap – the extent to which organisations perform similar tasks to address similar problems for some common members – may decrease the survivability of IGOs due to rule conflicts, institutional paralysis, and competitive pressures for limited resources. Contrary to this perspective, this article argues that greater overlap increases the likelihood of IGO survival for two reasons. First, similar IGOs will survive as member states use them to pursue forum-shopping strategies. Second, overlap enables IGO secretariats to draw on networks of support from other IGOs, rendering organisations that densely overlap with others more resilient. To test these propositions, the article combines data on institutional overlap in global governance and the survival of the 534 IGOs contained in the Correlates of War IGO dataset. Non-linear regression analysis finds that where organisations overlap more with existing organisations in terms of governance tasks and issue areas, the likelihood of IGO survival increases. Further analysis suggests that this result is driven by the forum-shopping strategies of powerful member states. Given that contemporary world politics is governed by an increasingly dense network of IGOs, these results hold important implications for the study of IGO survivability and the evolution of global governance more broadly.

Keywords:

• Intergovernmental organisations
• global governance
• institutional overlap
• IGO death
• regime complexity

The last decades have seen rapid growth in the number and scope of international agreements and organisations governing different areas of world politics (Koremenos 2016; Pevehouse et al. 2021). Issues, such as climate change, global health, and intellectual property rights, once governed by relatively disconnected international rule sets and organisations, are today subject to overlapping agreements and organisations that intersect and interact with one another in multiple ways (Alter and Meunier 2009; Alter and Raustiala 2018; Raustiala and Victor 2004). Intergovernmental organisations (IGOs) are at the core of this networked governance architecture (Kahler 2021) in which their creation, design, and life are fundamentally shaped by how they interact with other organisations (Alter and Raustiala 2018; Eilstrup-Sangiovanni and Westerwinter 2022).

Despite the growing institutional complexity of global governance, little is known about how this environment of institutional overlap affects the life and especially the survival of IGOs.¹ Extant research on IGO decline and death emphasises exogenous political and economic shocks, IGO age, membership, scope, centralisation, as well as secretariat size and autonomy as drivers of organisational vitality and survival (Debre and Dijkstra 2021; Eilstrup-Sangiovanni 2020; Gray 2018). While these explanations focus on IGO features and events that impact the international system as a whole, less attention has been given to the institutional context in which organisations operate. The features of this institutional context and the embeddedness of individual organisations within it, I argue, are an important part of the puzzle of IGO survival.

Much of the growing literature on institutional complexity in global governance emphasises the negative consequences of institutional overlap for individual IGOs. The overlap creates rule conflict between IGOs that can reduce their effectiveness, which ultimately undermines their relevance as problem-solving entities (Panke and Stapel 2023). Moreover, overlap generates competitive pressure for limited resources that can hollow out their problem-solving capacity, leading to stagnation in the growth of organisational populations and triggering niche-finding strategies (Abbott et al. 2016; Morin 2020; Schemeil 2013). From this perspective, overlapping organisations challenge the viability of IGOs because they increase competition for scarce resources which, in turn, may drive individual organisations to change and pursue less crowded institutional niches or push them towards the brink of organisational collapse.

Contrary to this perspective, I argue that institutional overlap may benefit IGO survival. Extending the scope of related literature that considers overlap in terms of common members and policy areas, I focus on functional overlap – the extent to which organisations perform similar governance tasks in similar policy areas – and analytically examine this dimension of overlap separately from membership overlap. In this disaggregated understanding of overlap, two organisations may functionally overlap even if they serve disjunct groups of member states. Importantly, though, institutional dynamics are different when functionally overlapping organisations have common member states. Depending on the presence of membership overlap, I posit two distinct mechanisms underlying the survival-enhancing effect of functional overlap, which have different testable implications.

Where functionally overlapping IGOs have common members, there is scope for cross-institutional politics by these member states. Through a practice known as ‘forum-shopping’, states leverage their membership in organisations with similar mandates and similar functions to assert their preferences (Alter and Meunier 2009; Davis 2009; Henneberg and Plank 2020; Hofmann 2019). States thus have incentives to invest in the viability of these overlapping organisations. However, where functionally overlapping IGOs have no common members, inter-institutional influences are indirect. Here the role of IGO secretariats is more important. Overlap in the issue areas and governance tasks that IGOs address enables organisations to identify other organisations with whom to exchange information, co-finance activities, and form partnerships (Betts 2013; Biermann 2008; Biermann and Koops 2017; Clark 2021; Littoz-Monnet 2014). It also facilitates the identification of role models and lessons learned on which new organisations can draw to set themselves up for success and long life. While both mechanisms predict a positive relationship between functional overlap and IGO survival, they can be empirically disentangled by disaggregating the institutional context into IGOs with common members and IGOs without common members.

I test my argument using original measures of institutional overlap in global governance that distinguish overlaps in terms of member states, issue areas, and governance functions (Reinsberg and Westerwinter 2023), as well as data on the survival of the 534 IGOs from the Correlates of War IGO dataset (Pevehouse et al. 2021). Using non-linear regression analysis, I find that functional overlap increases the likelihood of IGO survival. In addition, I find that this relationship only holds with respect to prior IGOs with which an organisation has any common member states, and not with the prior IGOs with which an organisation has no member states in common. Functional overlap with common members is most strongly correlated with IGO survival for organisations jointly dominated by a majority of the most powerful states in world politics. Taken together, these results lend support to the member-driven mechanism of IGO survival, whereby functionally overlapping IGOs survive because they allow powerful states to reap forum-shopping benefits.

My article makes three contributions to research on IGO survival and institutional overlap in global governance. First, I complement an expanding literature on IGO overlap (Haftel and Lenz 2022; Panke and Stapel 2018; Reinsberg and Westerwinter 2023), considering the consequences of overlap rather than its determinants. I provide a counterpoint to arguments that expect that institutional overlap undermines IGO survival as implied by studies on institutional complexity as well as organisational ecology. I derive alternative expectations drawing on theoretical accounts of cross-institutional politics (Clark 2022; Drezner 2009; Hofmann 2019) and inter-organisational relations from a network perspective (Cao 2009; Hafner-Burton and Montgomery 2006; Ingram and Torfason 2010), which both have not been used to study IGO survival.

Second, survival is an important aspect of the life cycle of individual IGOs that has recently attracted more systematic attention from students of international cooperation and global governance (Debre and Dijkstra 2021; Dijkstra and Debre 2022; Eilstrup-Sangiovanni 2020; Gray 2018; Shanks et al. 1996). Understanding the conditions under which IGOs survive and die is an important complement to the vast number of studies that examine IGO creation. With respect to studies on IGO survival, my argument complements theories that centre on IGO features and exogenous shocks to the international system as main drivers of IGO death and helps researchers to move towards a more contextual understanding of organisational survival. I also contribute to the theoretical discussion about the consequences of institutional overlap in global governance by linking it to IGO survival (Alter and Raustiala 2018; Eilstrup-Sangiovanni and Westerwinter 2022).

Third, I join the quantitative literature measuring overlap in terms of member states and issue areas (Haftel and Hofmann 2019; Haftel and Lenz 2022; Panke and Stapel 2018) but extend its scope to include governance tasks. My continuous measure of functional overlap captures the degree to which two organisations perform similar tasks to address similar problems. Using this original measure, as well as existing measures on membership overlap, my theoretical argument thereby integrates the main dimensions that theoretical discussions of regime complexity and institutional overlap have highlighted as causally important (Busch 2007; Henning and Pratt 2022; Hofmann 2009, 2011; Lipscy 2017; Raustiala and Victor 2004; Urpelainen and Van de Graaf 2015). This addresses a limitation of overlap research based on small numbers of cases in particular issue areas (Henning 2017; Kelley 2009; Keohane and Victor 2011). At the same time, by using the COW IGO dataset – the most widely used dataset for comparative IGO research – my effort is incremental and thus should help advance related research in the future.

The remainder of the article proceeds as follows. In the next section, I review the literature on the determinants of IGO survival and develop competing theoretical perspectives on the relationship between institutional overlap and IGOs survival. I then introduce the research design, including my measure of institutional overlap among IGOs and the data I use. The fourth section presents the results of my empirical analysis. The final section concludes by discussing the broader implications of my argument and identifying areas for future research.

The determinants of IGO survival

Survival – as an important aspect of the life cycle of IGOs – has recently begun to attract more systematic attention from students of international cooperation and global governance (Debre and Dijkstra 2021; Dijkstra and Debre 2022; Eilstrup-Sangiovanni 2020; Gray 2018; Shanks et al. 1996). Survival can mean that an organisation either is operational at a given point in time or has continued to be in operation from its point of establishment to a given point in time. In contrast, an organisation is dead if it is formally disbanded or no longer fulfills the criteria of an intergovernmental organisation. IGO death, broadly construed, is surprisingly common: over one-third of all IGOs established between 1815 and 2014 have ceased to exist (Eilstrup-Sangiovanni 2020). Not all IGO deaths imply outright failures though, given that some organisations have merged with others. In other cases, these IGOs have been superseded by successor organisations, often adopting similar designs.²

Working off the well-established COW IGO dataset, quantitative research has found that favourable external conditions and institutional characteristics of IGOs can explain their longevity. For example, Eilstrup-Sangiovanni and Westerwinter (2022) find that IGOs are more likely to survive in the absence of major exogenous shocks like inter-state wars. While juvenile IGOs are at higher risk of extinction, those IGOs with large memberships and centralised structures tend to survive longer. These results confirm earlier research showing that having a large and heterogeneous membership is associated with greater organisational survivability (Eilstrup-Sangiovanni 2020). A diverse set of members makes IGOs difficult to create, but also difficult to disband, in part due to the utility of these organisations for member states in solving their collective action problems. Focusing on institutional characteristics, Debre and Dijkstra (2021) find that large bureaucracies survive better, even though some exceptions prove the rule. For example, a prominent unlikely case of IGO death is the International Refugee Organisation, whose fate can be linked to exceptionally strong shifts in the collective preferences of member states on refugee policy (Dijkstra and Debre 2022). Surprisingly, member state withdrawal has limited impacts on IGO survival (Borzyskowski and Vabulas 2022).

Institutional overlap and the ‘survival of the fittest’

While this incipient research has greatly advanced our understanding of the life and death of IGOs, a surprising omission concerns the role of institutional context. Considering institutional context is important because IGOs do not exist in a vacuum but in an increasingly dense architecture of global governance institutions (Gehring and Faude 2014; Henning 2019; Raustiala and Victor 2004). Hence, there is an urgent need to understand how institutional overlap affects IGO survival.

The notion of institutional overlap is linked to the growing complexity of global governance, with its increasing number of organisations that perform similar tasks to address similar problems for similar sets of member states (Alter and Raustiala 2018; Eilstrup-Sangiovanni and Westerwinter 2022; Raustiala and Victor 2004). The concept of institutional overlap used here is deliberately broad, encapsulating the core dimensions that theoretical discussions of regime complexity and institutional overlap have highlighted as causally important; namely, overlap of issue areas, memberships, and governance tasks (Busch 2007; Henning and Pratt 2022; Hofmann 2009; Lipscy 2017; Raustiala and Victor 2004; Urpelainen and Van de Graaf 2015). Surprisingly, though, existing work has understood overlap in terms of common membership and policy areas. This definitional choice omits the critical dimension of governance tasks. Moreover, this overlap definition is too narrow because it presumes common membership, even though organisations may interact in other ways than through shared members. For example, during the COVID-19 pandemic, the African Union Commission and the Islamic Development Bank, which have a minimal overlap of member states, signed a Memorandum of Understanding on 17 February 2023 to enhance their collaboration on digitisation (IsDB 2023). Examples like these indicate the existence of diffusion processes of IGO policies that do not require leadership from joint member states (Reinsberg and Westerwinter 2023).

Some studies on institutional overlap take an ecological approach, which focuses on features of an organisational type to predict its viability (Eilstrup-Sangiovanni 2020). Ecological studies view organisations as competing entities in a resource-scarce environment. When organisations struggle to adapt to their changing environment, they may face decline, giving rise to a selection process known as ‘survival of the fittest’. At the macro-level, these processes imply the prediction that when the density of an organisational type increases, its rate of reproduction declines (Abbott et al. 2016; Freeman and Hannan 1977; Lipscy 2017). With some further adjustments, these insights are relevant for IGO survival. To be sure, IGO birth rates have been declining for the past decades, given that IGO-based cooperation has become ever denser. An ever-more densely populated IGO system could eventually undermine the viability of individual IGOs, as many IGOs compete for scarce resources from the same (limited) set of member states to solve similar types of problems in similar issue areas (Haftel and Lenz 2022). Compared to other institutional types, IGOs are less likely to be able to adapt quickly to a changing environment (Abbott and Faude 2021). Even where IGOs (eventually) manage to adapt, such change processes are fraught with uncertainty and may push IGOs into niches for which states have limited demand (Hannan and Freeman 1984). Where new IGOs challenge existing IGOs, the new IGOs are potentially more adept than the legacy IGOs at addressing the imminent cooperation problems for which they were established. Due to this logic of competition, coupled with the institutional rigidity of IGOs, institutional overlap would be expected to undermine the survival prospects of IGOs.

My argument addresses two limitations of ecological approaches. First, these approaches presume that inter-organisational interactions are necessarily competitive, although these interactions may also be cooperative, for instance when it comes to joint efforts to raise awareness for policy problems. Second, these approaches neglect the ways in which member states can exploit ‘less fit’ IGOs to advance their own ends. Even where substantial overlap exists, it takes mutual agreement to kill them off formally. And as long as a couple of countries find them useful, they will not agree to disband ‘less fit’ IGOs. Hence, especially in the presence of functional overlap, it is more likely that one particular institutional arrangement will end up suiting some countries and not others.

Institutional overlap and IGO survival revisited

Different combinations of these overlap dimensions have different implications for the ways in which organisations interact and thus whether they remain alive (Table 1). Where two IGOs perform different functions and serve disjunct sets of members, they will not interact and hence their survival is independent of each other. An example of this constellation is the East African Development Bank (EADB) and the Institute of International Education (IIE). A similar dynamic can be expected for two IGOs that have some member states in common but that have different purposes, for example, the International Monetary Fund (IMF) and the Global Environment Facility (GEF). Where two IGOs perform similar tasks to solve similar problems but for disjunct groups of members, they can interact only indirectly, involving no flow of resources between member states. This implies a greater role for entrepreneurial action by their secretariats (Hall 2016). For example, the East African Development Bank and the European Investment Bank (EIB) are two regional development banks with non-overlapping membership. Finally, two IGOs may perform similar functions for similar members, as in the case of the World Trade Organisation (WTO) and the World Intellectual Property Organisation (WIPO). This constellation has the strongest implications for IGO survival.

Table 1. Systematisation of different combinations of overlap in global governance.

	No membership overlap	Membership overlap
No functional overlap	Completely unrelated IGOs (EADB and IIE) → No implications for survival	IGOs with shared members but distinct functions (GEF and IMF) → No implications for survival
Functional overlap	IGOs with similar functions but no shared members (EADB and EIB) → Implications for survival through indirect mechanisms	IGOs with similar functions and common members (WIPO and WTO) → Strong implications for survival through direct interaction

Notes: Functional overlap refers to overlap in both governance tasks and issue areas.

Building on the premise that member state incentives – rather than the fitness of an IGO – matter for its survival, I expect that institutional overlap will enhance IGO survival. Where two organisations overlap in terms of both memberships and issue areas, there is scope for cross-institutional politics whereby states leverage their membership in organisations with similar mandates and similar functions to assert their preferences. This practice is known as ‘forum-shopping’ (Alter and Meunier 2009; Davis 2009; Drezner 2009; Henneberg and Plank 2020; Hofmann 2019). While forum-shopping has been criticised for creating rule conflict and inefficient duplication of efforts (Drezner 2009), it seeks to advance the political goals of those states capable of doing it. For example, Haftel and Hofmann (2019) examine why regional trade organisations often trespass into the security domain – also known as judicial overreach (Kucik et al. 2022). They find that strategic inter-state rivalries are an important driver of organisational overlap through scope expansion. Henneberg and Plank (2020) seek to explain the persistence of overlapping regional security organisations in Africa. They find that Nigerian resistance towards external intervention and hegemonic interests explain the creation of the Multinational Joint Task Force, adding to the Economic Community of West African States and the African Union as important existing security providers in Africa.³ In the trade domain, Davis (2009) accounts for the proliferation of preferential trade agreements, showing how forum-shopping strategies allow states to choose the set of rules that favour their preferred outcome. The common thread in these studies is that they take a power-based approach to explain forum-shopping. Given the political benefits of IGO overlaps for states, these states thus have incentives to invest in the viability of these overlapping organisations. To be sure, some scholars argue that this ‘institutional thickening’ will ultimately erode the prospects for cooperation, for instance by creating rule conflict (Drezner 2009). However, this argument overlooks the benefits of institutional layering strategies. For example, using the case of the trade regime, Faude (2020) shows that preferential trade agreements help states accommodate heterogeneous preferences and distributive conflicts, thereby furthering the expansion of the membership and the regulatory scope of the GATT/WTO system. If this argument holds, I should observe a positive relationship between functional overlap and IGO survival. In addition, this relationship should be driven by the functional overlap between organisations with common members.

Where functionally overlapping IGOs have no common members, inter-institutional influences are indirect. IGOs do not have any direct ties that can serve as conduits for the flow of information and other valuable resources through shared members (Böhmelt and Spilker 2016). Without these member-directed resource flows, the role of IGO secretariats is arguably more important. Overlap in the issue areas and governance tasks that IGOs address enables organisations to identify other organisations with whom to exchange information, co-finance activities, and form partnerships (Betts 2013; Biermann 2008; Biermann and Koops 2017; Clark 2021). For example, the Partnership of International Organisations for Effective International Rulemaking (OECD-IOP) is a voluntary forum of over 50 IGOs to ‘foster collective action among [IGO] secretariats […] to promote greater quality, effectiveness, and impact of international rules’. Convened by the OECD, the initiative seeks to ‘build greater confidence of domestic regulators and legislators in international rules and support greater use of good quality international instruments in national legislation’, through regular meetings and working groups in which organisations can exchange experience and develop common approaches to solve regulatory problems (OECD 2023). Furthermore, functional overlap (even without common members) facilitates the identification of role models and lessons learned on which new organisations can draw to set themselves up for success and long life (Sommerer and Tallberg 2019). This does not even require strong organisational capacities: As argued by sociological literature on institutional isomorphism, an organisation might copy the behaviours of its peers, thereby increasing its attractiveness in the field through ‘naïve learning’ (Beckert 2010; DiMaggio and Powell 1983; Dingwerth and Pattberg 2009). Using the example of the OECD-IOP, the activities of the forum so far helped foster exchange with academic experts and led to a brochure on how to enhance the quality of international rulemaking, which holds the potential to make participating IGOs more valuable addresses of regulatory endeavours. Overall, this argument draws on research highlighting the benefits of dense institutional networks in global governance (Cao 2009; Greenhill and Lupu 2017; Hafner-Burton et al. 2009; Ingram and Torfason 2010). In sum, functional overlap without common members may increase the probability of IGO survival because greater overlap implies the potential for more connections with incumbent organisations, which provides access to more resources that ease the start-up phase and make early failures less likely. If this argument holds, I should observe a positive correlation between institutional overlap and IGO survival, specifically with the set of IGOs without common members.

Summarising my theoretical discussion, I formulate the following (mutually non-exclusive) hypotheses.

Hypothesis 1: Institutional overlap will enhance IGO survival.

Hypothesis 2a [forum-shopping benefits]: The positive relationship between functional overlap and IGO survival will hold particularly between organisations with common members.

Hypothesis 2b [organisational networks]: The positive relationship between functional overlap and IGO survival will hold particularly between organisations without common members.

Data and methods

To test my argument, I examined 534 IGOs in 1815–2016 from the COW IGO dataset (Pevehouse et al. 2021).⁴ The COW IGO dataset is the most commonly used dataset in world politics (Borzyskowski and Vabulas 2022; Debre and Dijkstra 2021; Eilstrup-Sangiovanni 2020; Reinsberg and Westerwinter 2021). I arrange the data in two ways, each providing complementary insights for my inquiry. First, I use an IGO panel dataset in which I observe each organisation annually from its year of inception until the year of its death. If an organisation dies, it leaves the sample; if it survives, it remains until the end of the sample period. This dataset allows for dynamic analysis of how the evolving overlap of an organisation with its environment affects its survival. Second, I use a cross-sectional dataset in which each organisation is observed only once. With key covariates measured at the point of inception, this dataset allows for an analysis of how the institutional context and organisational characteristics at the design stage affect whether an organisation survives.

Both empirical setups complement each other and none is inherently better than the other. The cross-sectional design offers the most faithful test of theories of rational design, which posit that IGO designers can foresee contingencies and adjust IGO design already at the outset. The repeated cross-sectional analysis allows for tests of organisational theories that endow organisations with an agency that enables them to interact with their institutional environment and thus manage overlap. From an empirical perspective, a key challenge is that many IGO covariates are not observed repeatedly over time. Therefore, the cross-sectional analysis likely yields more conservative estimates while allowing for testing a wider range of empirical implications due to its more limited data requirements.

IGO survival

My key outcome is an indicator variable for whether an IGO has continued to meet the definitional criteria of an international organisation. These three criteria require that IGOs have at least three member states, hold regular meetings at least every ten years, and have a permanent secretariat. While IGOs may fail to meet any of these criteria in any given year, IGO death is practically irreversible, especially if it involves the dismantlement of organisational capacities. For a time-invariant measure of IGO survival, it is, therefore, appropriate to require that the organisation fulfills all three criteria in each year of its existence. I do not consider cases in which IGOs continued living under a new name as IGO deaths (Borzyskowski and Vabulas 2022). In robustness tests, I infer IGO survival from the COW IGO variable indicating an organisation is not considered ‘terminated’.⁵

Among the 534 IGOs in my sample, 395 IGOs (74%) have survived until 2014. The survivability of IGOs has differed across certain time periods. Figure 1 plots the number of IGOs that survived against the number of IGOs that died separately for different cohorts of IGOs that were established in different time periods. I find that the likelihood of survival has increased over time. While immediately after World War II about 60 IGOs were created – half of which survived and half of which eventually died – around the end of the Cold War about 90 IGOs were created – but here 70 IGOs survived and fewer than 20 IGOs died. Of course, these numbers are not adjusted for the fact that older IGOs have a higher chance of dying only because they have been around for longer. Nonetheless, I do not see major disruptions to the general pattern that a higher share of newly created IGOs does survive, perhaps except for the period after the Cold War.

Figure 1. Survival of IGOs over time.

IGO survival rates also differ across issue areas. Table 2 shows the percentage of IGOs that did not die, alongside the total number of IGOs, across several issue areas. In line with previous descriptive findings in the literature, I find that some policy areas, notably security and trade, have lower survival rates, whereas areas like development, environment, and finance are more benign for IGO survival.

Table 2. Death rates across issue areas (1815–2014).

Issue area	IGOs that survived (%)	Number of IGOs in issue area
Security	75.8	33
Environment	82.6	109
Finance	85.5	62
Trade	73.4	158
Development	84.2	196

Institutional overlap

Institutional overlap refers to the degree to which two institutions share specific features, such as common members, governance tasks, and issue areas. It is naturally defined between pairs of organisations, which is why I need to aggregate dyadic overlap measures for monadic analysis. The literature distinguishes broadly between membership overlap (Sommerer and Tallberg 2019) and functional overlap, which implies that two organisations perform similar functions within similar issue areas (Rosendal 2001; Urpelainen and Van de Graaf 2015; Young 1996).

Functional overlap requires me to calculate overlap along two additional dimensions. First, issue overlap is computed as the cosine similarity of the profiles of the issue areas in which two IGOs are active. I distinguish nine major issue areas in which IGOs can operate: security, environment, health, human rights, development, trade and commerce, finance, social affairs, and technical affairs, similar to existing databases (Hooghe et al. 2019; Koremenos 2016; Tallberg et al. 2014). An IGO can be active in the governance of more than one issue area. Second, governance task overlap is the cosine similarity of the profiles of non-mutually exclusive governance tasks of two IGOs. A governance task profile is a tuple of eight dummies, capturing whether the organisation has a mandate, respectively for (1) information-gathering, (2) agenda-setting, (3) service provision, (4) funding, (5) capacity-building, (6) standard-setting, (7) policy implementation, and (8) monitoring (Abbott and Snidal 2009; Avant et al. 2010; Westerwinter 2021). To create a monadic measure of functional overlap, I multiply governance task overlap and issue area overlap between a given IGO and each other IGO that was still alive in a given year, and average across these pairwise similarities for each IGO. For a time-invariant measure of overlap, I proceed in the same way but only consider the average overlap of an IGO with its predecessors at its year of establishment.

To gain purchase on underlying mechanisms, I further disaggregate functional overlap into two measures. On the one hand, I compute the average functional overlap between all IGO pairs that have at least some common member states. The intuition for this measure is that functional overlap might only matter for governance outcomes if the organisations serve overlapping sets of member states (Urpelainen and van de Graaf 2015). Consequently, I can use this measure to identify mechanisms primarily driven by member states, such as forum-shopping considerations. On the other hand, I compute the average functional overlap between all IGO pairs that have no common member states. This measure allows me to capture mechanisms unrelated to forum-shopping strategies of states, such as direct interactions between IGO staff and demonstration effects.

Figure 2 depicts the functional overlaps at the inception of organisations established in specific time periods. Functional overlap is relatively stable over time with respect to prior IGOs with common members and prior IGOs without common members. The figure would look similar if I were to use the dynamic measure of functional overlap. In the Online Appendix, I show that membership overlap at inception has consistently declined over time (Figure A1) – a likely consequence of the expanding state system and the increasing segmentation of global governance institutions. I also explore the relationship between different dimensions of overlap, finding no correlation between functional overlap and membership overlap (Figure A2).

Figure 2. Evolution of functional overlap over time.

Notes: Dark gray bars show the functional overlap with prior IGOs with which an organisation shares any member states. Light gray bars show functional overlap with prior IGOs with which an organisation has no common member states. Vertical lines in bars show the median values of functional overlap, whiskers indicate the interquartile range, and dots indicate outliers.

Control variables

I draw on the literature on IGO survival to capture alternative explanations for IGO death and survival (Dijkstra 2019; Eilstrup-Sangiovanni 2020; Gray 2018). I begin with a basic set of control variables, sequentially including time effects, regional effects, and issue-area dummies. Time period dummies account for the fact that older IGO cohorts have had more time to live and therefore longer exposure to risk factors of IGO death. Regional dummies account for differences in average survival rates across world regions and particularly global organisations. Issue-area dummies control for factors that vary across issue areas, such as the severity of the cooperation problem, which would create demand for organisations and ensure their survival.

I also add control variables from alternative theoretical accounts. In line with rational institutionalism, issue area dummies control for specific problem structures and differences in levels of governance demand that establish varying incentives for sustaining institutionalised cooperation (Koremenos 2016). I also include the (logged) number of member states, following previous research showing that IGOs with more member states are more likely to survive given that IGOs can tap into a greater potential resource base when they have more members (Eilstrup-Sangiovanni 2021). The arrival of new member states might also unleash survival-relevant institutional dynamics by upsetting the original institutional bargain among founding states (Gray et al. 2017). Furthermore, I measure whether the organisation has an independent secretariat (Westerwinter 2021). Proxying for capacities for autonomous behaviour, I argue that secretariats are positively related to IGO survival (Bauer and Ege 2016; Biermann and Siebenhüner 2009; Debre and Dijkstra 2021; Reinalda 2020).

In the IGO panel analysis, I also include a cubic time polynomial to approximate the survival hazard.⁶ The hazard captures the effect of time elapsed since birth on the survival of an organisation and differs from cohort effects captured by the period dummies. While measures of secretariat design, issue areas, and governance tasks are time-invariant, the number of member states and functional overlap are indeed time-varying.⁷ Given the substantive interest in many time-invariant variables, I estimate probit models in my main analysis. In robustness checks, I also use linear fixed-effects estimation. To account for inter-temporal dependence, I cluster standard errors on organisations.

In the cross-sectional analysis, covariates are measured at the point of creation of the organisation. I estimate ordinary probit models and compute robust standard errors to mitigate potential heteroskedasticity. The Online Appendix includes further information on variable definitions and data sources alongside summary statistics for all variables used in both the panel analysis (Table A1) and the cross-sectional analysis (Table A2).

Results

The empirical section begins with the panel analysis, where I find a significantly positive relationship between contemporary overlap and IGO survival. This result is robust across different model specifications. I then present the findings from the cross-sectional analysis, which yields a positive relationship between overlap at inception and IGO survival.

Dynamic analysis over the IGO lifetime

Table 3 presents the baseline findings from the panel analysis, finding a consistently positive relationship between (contemporary) overlap and IGO survival. This result holds in the baseline model – with only region dummies, period dummies, and hazard polynomials – and when adding issue area dummies. It also holds when controlling for the number of member states, which suggests that the result cannot be explained by membership expansion. The final model further suggests that independent secretariats boost IGO survival, but this mechanism operates independently of overlap.

Table 3. Functional overlap and IGO survival.

	(1)	(2)	(3)	(4)
IGO survival
Overlap	3.021*** (0.461)	5.494*** (0.979)	5.514*** (1.067)	5.181*** (1.073)
Security		0.298** (0.148)	0.356** (0.160)	0.315* (0.179)
Environment		0.108 (0.092)	0.215** (0.100)	0.179* (0.102)
Health		−0.003 (0.130)	−0.043 (0.136)	−0.056 (0.133)
Human rights		0.127 (0.159)	0.043 (0.163)	0.072 (0.177)
Trade and commerce		−0.261*** (0.087)	−0.264*** (0.091)	−0.298*** (0.091)
Finance		0.397*** (0.144)	0.423*** (0.153)	0.356** (0.150)
Development		−0.080 (0.103)	−0.066 (0.111)	−0.093 (0.111)
Social affairs		−0.347*** (0.122)	−0.369*** (0.130)	−0.311** (0.133)
Technical issues		−0.153* (0.088)	−0.183* (0.096)	−0.200** (0.095)
Number of member states			0.243*** (0.048)	0.214*** (0.049)
Independent secretariat				0.319*** (0.095)
Period dummies	Yes	Yes	Yes	Yes
Region dummies	Yes	Yes	Yes	Yes
Observations	18,709	18,709	17,883	17,883
Organisations	495	495	495	495
Pseudo-R^2	0.059	0.080	0.099	0.107

Probit estimation with IGO-clustered standard errors in parentheses. Third-order time polynomials modelling the hazard are included but omitted from the results presentation. Significance levels: *p < .1, **p < .05, ***p < .01.

I begin to probe the underlying mechanisms by disaggregating functional overlap. I consider different subsets of IGOs in the institutional context, depending on whether or not they share common members with a given organisation. Table 4 finds that functional overlap with those IGOs with which a given organisation shares any common members is significantly positively related to its survival. In contrast, overlap with IGOs with non-overlapping membership is not significantly related to survival. These findings indicate that the forum-shopping behaviour of member states may underlie the survival-enhancing effect of overlap.

Table 4. Functional overlap with different organisations and IGO survival.

	(1)	(2)	(3)	(4)
IGO survival
Overlap with common members	2.562*** (0.683)	3.351*** (0.970)	2.984*** (0.934)	2.671*** (0.925)
Overlap without common members	0.272 (0.584)	0.600 (0.696)	0.592 (0.724)	0.333 (0.722)
Number of member states			0.224*** (0.056)	0.195*** (0.057)
Independent secretariat				0.431*** (0.129)
Period dummies	Yes	Yes	Yes	Yes
Region dummies	Yes	Yes	Yes	Yes
Issue area dummies	No	Yes	Yes	Yes
Observations	14,723	14,723	14,723	14,723
Organisations	446	446	446	446
Pseudo-R^2	0.050	0.077	0.094	0.107

Probit estimation with IGO-clustered standard errors in parentheses. Third-order time polynomials modelling the hazard are included but omitted from the results presentation. Significance levels: *p < .1, **p < .05, ***p < .01.

I probe the robustness of these results in the Online Appendix. First, I include additional control variables. Specifically, I probe whether US membership in an IGO affects its survival. I also control for initial overlap conditions, notably the average number of shared G7 states and the average combined overlap with predecessor IGOs. To rule out scaling effects due to the size of the institutional environment of a given organisation, I measured the number of IGOs that were still alive and those that were dead in a given year. Finally, I measure three characteristics of the IGO membership: preference heterogeneity, democratic credentials, and total output. I include the average of each variable as well as its standard deviation. While none of these variables qualitatively changes the overlap coefficient, I also find that organisations with more powerful members are less likely to survive but greater diversity in terms of economic power enhances survival (Table A3). Second, I include fixed effects to control for unobserved IGO heterogeneity and estimate linear regressions. I corroborate a positively significant relationship between overlap and survival, especially with respect to the organisations with common members (Table A4).

Due to limited data, I cannot systematically test an important rival hypothesis. IGOs might be better able to survive if they expand their policy scope over time. If this was the case, I should observe the relationship between functional overlap and IGO survival to differ depending on whether an organisation expanded its policy scope. Visual inspection of the relationship of interest across these two-subsets does not reveal any difference (Figure A3). These patterns suggest that scope expansion is unlikely to confound my core result.

Analysis of overlap at IGO inception

I complement the dynamic analysis with a cross-sectional analysis measuring overlap at the point of IGO inception. Table 5 presents the baseline results, with sequentially expanding sets of controls in each column. The relationship between functional overlap and IGO survival is consistently positive. For example, based on the third model, moving functional overlap from its minimum to its maximum in that sample, the likelihood of IGO survival increases from 57.5% (95%-CI: 41.4–73.5%) to 89.7% (95%-CI: 80.8 – 98.6%). Together with the earlier results from the panel analysis, these results provide support for hypothesis 1, suggesting that functional overlap insulates IGOs against the risk of death.

Table 5. Functional overlap and IGO survival at the time of IGO inception.

	(1)	(2)	(3)	(4)
IGO survival
Overlap	3.273*** (0.853)	4.353*** (1.611)	4.041** (1.677)	3.282* (1.743)
Security		0.294 (0.258)	0.278 (0.265)	0.187 (0.270)
Environment		0.316* (0.174)	0.390** (0.174)	0.349** (0.176)
Health		−0.185 (0.241)	−0.122 (0.260)	−0.134 (0.266)
Human rights		0.109 (0.343)	−0.088 (0.339)	−0.142 (0.348)
Trade and commerce		−0.312** (0.158)	−0.256 (0.159)	−0.308* (0.161)
Finance		0.653*** (0.233)	0.611*** (0.228)	0.530** (0.231)
Development		0.283* (0.165)	0.261 (0.167)	0.240 (0.169)
Social affairs		−0.024 (0.237)	−0.030 (0.242)	0.056 (0.255)
Technical issues		−0.137 (0.173)	−0.160 (0.181)	−0.161 (0.183)
Number of member states			0.284*** (0.062)	0.279*** (0.062)
Independent secretariat				0.649*** (0.179)
Period dummies	Yes	Yes	Yes	Yes
Region dummies	Yes	Yes	Yes	Yes
Observations	506	506	506	506
Pseudo-R^2	0.123	0.160	0.197	0.219

Probit estimation with robust standard errors in parentheses. Significance levels: *p < .1, **p < .05, ***p < .01.

Mirroring my earlier findings, I find that functional overlap is not the only determinant of IGO survival. IGOs in different issue areas have different survival rates. Moreover, in line with previous findings in the literature, I find that an organisation with more member states is more likely to survive (Eilstrup-Sangiovanni 2020). Finally, having an independent secretariat insulates organisations against possible demise (Dijkstra and Debre 2022). Strikingly, my results with respect to overlap are qualitatively unaffected by the inclusion of these alternative explanations, although their statistical significance decreases somewhat.

Table 6 shows the results for models with two overlap measures considering whether or not an IGO has any common members with other IGOs. Functional overlap is significantly related to the survival of an organisation only when considering its functional overlap with the set of prior IGOs with which it shares any member states. Substantively, using the third model, a change in functional overlap with common member states from the sample minimum to the sample maximum increases the probability of IGO survival from 58.4% (95%-CI: 39.7–77.1%) to 90.6% (95%-CI: 80.5–99.9%). In contrast, the relationship between overlap and survival is insignificant when considering overlap with the set of prior IGOs without any common member states. These findings suggest that the survival-enhancing effect of functional overlap is linked to the forum-shopping strategies of member states – thus lending support to hypothesis 2a but finding no support for hypothesis 2b.

Table 6. Functional overlap with different legacy organisations and IGO survival.

	(1)	(2)	(3)	(4)
IGO survival
Overlap with common members	3.867*** (1.352)	4.157** (1.826)	3.859** (1.910)	3.277* (1.913)
Overlap without common members	−1.695 (1.284)	−2.024 (1.400)	−1.915 (1.516)	−2.301 (1.431)
Number of member states			0.266*** (0.074)	0.264*** (0.074)
Independent secretariat				0.670*** (0.199)
Period dummies	Yes	Yes	Yes	Yes
Region dummies	Yes	Yes	Yes	Yes
Issue area dummies	No	Yes	Yes	Yes
Observations	438	438	438	438
Pseudo-R^2	0.113	0.157	0.184	0.206

Probit estimation with robust standard errors in parentheses. Significance levels: *p < .1, **p < .05, ***p < .01.

If forum-shopping strategies enhance the survival of IGOs that serve similar groups of member states with similar functions, I should observe this mechanism for the set of IGOs controlled by the most powerful states. This observation should hold for two reasons. First, these states enjoy the highest returns to the use of power, which provides an incentive for them to exert power, including through institution-shaping strategies (Eilstrup-Sangiovanni 2015; Gruber 2000; Stone 2011). Second, powerful states also have the capability to navigate multiple forums. They have well-staffed diplomatic missions and hence can more easily assert their preferences in these forums (Pouliot 2016).

To test this expectation, I split the sample at the median of the number of shared members that belong to the nine most powerful states in the institutional context of an organisation. These nine most powerful states are the G7 countries, China, and Russia. Cases below the median draw on IGOs in which few of these nine states participate, whereas cases above the median draw on IGOs in which many of these nine states participate. Table 7 shows the results. I find a consistently positive relationship between functional overlap and IGO survival in the field of organisations that tend to be controlled by powerful states. Conversely, there is no relationship between functional overlap and IGO survival for the organisations not controlled by an above-median number of powerful states.⁸

Table 7. Functional overlap and IGO survival in different sub-samples.

	Many common powerful members		Few common powerful members
	(1)	(2)	(3)	(4)
IGO survival
Overlap with common members	4.244** (1.791)	3.916** (1.741)	−0.967 (3.525)	−1.395 (3.556)
Overlap without common members	−2.138 (1.478)	−2.560* (1.477)	4.159 (4.021)	4.421 (3.993)
Number of member states		0.292*** (0.099)		0.203* (0.109)
Independent secretariat		0.778*** (0.261)		0.564* (0.304)
Period dummies	Yes	Yes	Yes	Yes
Region dummies	Yes	Yes	Yes	Yes
Issue area dummies	No	No	No	No
Observations	206	206	220	220
Pseudo-R^2	0.172	0.248	0.096	0.122

Probit estimation with robust standard errors in parentheses. The sample split is performed at the median of the number of shared powerful members (which include the G7 states, China, and Russia). Significance levels: *p < .1, **p < .05, ***p < .01.

In additional tests, I want to exclude the possibility that the higher likelihood of IGO survival comes at the expense of lower quality of life, better known under the label ‘zombies’ (Gray 2018). Data to test whether overlap affects the odds of becoming a zombie are available for a small subset of regional economic organisations.⁹ While zombyism is time-varying in principle, I consider whether an organisation has been a zombie in any year of its existence. In my sample, about 35 out of 43 regional organisations (81%) are classified as zombies.

In the Online Appendix, I examine whether overlap predicts zombyism, respectively for IGOs jointly controlled by an above-median number of powerful states and IGOs for which that is not true. Functional overlap can turn organisations into zombies when they are controlled by less powerful states while the opposite is true when more powerful states hold the strings (Table A5). These results seem to indicate that institutional overlap in environments of poor capacity – but a desire by certain states to keep ‘their’ IGO alive for whatever boutique reason – can undermine the quality of life of such organisations, even if outright death is avoided.¹⁰

In addition, I probe how overlap affects other IGO outcomes. A different form of survival not included in my outcome is replacement by a successor organisation. IGOs may also be integrated into other organisations. These alternative outcomes are forms of institutional evolution that do not necessarily reflect IGO failure, but they are qualitatively different from IGO survival as examined here. Thus, I would not expect my theoretical mechanism to be at work for these outcomes. Indeed, I do not find any consistent relationship between functional overlap and these alternative outcomes (Table A6).

By way of probing the robustness of my findings, I consider an alternative indicator of IGO survival, directly available from the COW IGO dataset. When an IGO dies, it leaves the COW IGO dataset; hence, I define survival as the absence of any death in the sample period for a given organisation. Albeit widely used, this indicator does not capture IGOs that fade away and is therefore less suitable for my analysis. Yet, using this alternative indicator, I corroborate the positively significant relationship between functional overlap and IGO survival (Table A7). Furthermore, I find that this relationship only holds for organisations with common members, but not for organisations without common members (Table A8) – highlighting again that the mechanism underpinning this relationship must be related to member-state behaviours.

Building on my earlier descriptive result that survival rates differ across issue areas, I also probe the stability of my findings using a leave-one-out sensitivity check. Here I replicate the analysis while dropping one issue area at a time and gauging the coefficient of interest. In all leave-one-out iterations, I find a significantly positive relationship between functional overlap and IGO survival (Figure A4).

Rather than measuring whether IGOs die, I can also measure the number of years an eventually dead IGO manages to survive. As this is only defined for the subset of dead IGOs, I need to estimate a two-step model in which the first stage is IGO death and the second stage is lifetime until death. I read off the lifetime in years directly from the COW IGO data by taking the difference between the end year and the start year of a dead organisation. Unfortunately, it is impossible to separate the two processes: When an IGO does not accumulate additional life years, it is dead. Identification therefore relies on assumptions about the non-linearity of the Heckman estimator (Sartori 2003). In addition to confirming that functional overlap insulates IGOs against death, I find that overlap is positively related to the IGO lifetime until death (Table A9). This is useful information because it confirms my findings for the probability of IGO survival and provides an estimate of the additional lifetime generated from overlap. Specifically, an increase of overlap from its minimum to its maximum increases the lifetime of IGOs by about 19.27 years. I also find similar results when distinguishing overlap with prior IGOs with or without common members (Table A10).

In sum, I found that functional overlap is positively related to IGO survival. I suspected that underlying this result is the investment of states in multiple IGOs that perform similar functions in similar issue areas and that help them assert their interests. Consistent with this argument, I showed that only functional overlap among organisations with common members is relevant for IGO survival, especially so for the subset of IGOs that involve an above-median number of powerful states.

Conclusion

I examined how institutional overlap affects the survival of intergovernmental organisations. Using data from the COW IGO dataset and a measure of functional overlap that indicates the extent to which an organisation performs similar tasks in a similar issue area to its predecessors, I found that functional overlap is significantly positively related to IGO survival. This result held both in a dynamic setup where overlap can evolve over time as new organisations enter the institutional environment, as well as in a static setup that examines how overlap at the point of inception of an organisation affects its subsequent survival.

By considering functional overlap separately for organisations with common members and organisations without common members, I began to probe two complementary mechanisms underpinning the survival-enhancing effect of overlap. My results were most consistent with the interpretation that states have incentives to construct multiple organisations that perform similar functions in the same issue area because they help them assert their interests. Three observations are consistent with this argument. First, I found that functional overlap promotes the survival of organisation only with the existing organisations with which it has some common member states but not with those organisation that serves different groups of member states. Second, the relationship of interest is driven by those organisations jointly controlled by the most powerful states, which arguably have the willingness and the capacity for effective venue shopping. Third, further analysis on a sub-sample of regional organisations suggests that overlapping organisations jointly controlled by the most powerful states are effective, suggesting that their forum-shopping activities do not create ‘zombie organisations’ but that this risk indeed increases where states with less capacity try to uphold duplicate institutions to advance their boutique interests.

Before discussing further implications, I note the limitations of my study. First, my statistical tests are not causal, and they ultimately cannot prove underlying mechanisms. Nonetheless, the empirical patterns are consistent with the theoretical argument that overlap increases survival because of state-driven forum-shopping strategies. Second, my statistical tests did not support the notion that autonomous IGO interactions bolster IGO survival. However, that does not exclude the possibility that such interactions matter, but potentially that they cannot be reliably measured. Moreover, due to severe data limitations, I could not account systematically for the possibility that overlapping organisations informally expand their mandates to boost their survival (Littoz-Monnet 2021). Nor did I consider that states seise opportunities for ‘emergent flexibility’ to reinterpret organisational rules to respond to new problems (Búzás and Graham 2020). Yet, I considered overlap both during the lifetime of IGOs, as well as overlap at birth. Future research could look at additional conditions for survival. An obvious starting point is shifts in bargaining power (Huikuri 2023), but also endogenous processes of institutional change (Genschel 2002; Gray 2020; Heldt et al. 2022). Another area for future research concerns the interaction of IGOs with other types of international organisations, specifically informal ones (Roger and Rowan 2023; Vabulas and Snidal 2013). There seems to be no indication that informal IGOs are associated with the death of formal ones; they might even enhance their productivity (Roger 2022).

My findings complement a burgeoning literature on the determinants of IGO survival. While highlighting the independent role of institutional context, I also confirmed that other correlates of IGO survival continue to matter, such as the existence of an independent secretariat, a large body of member states, and issue characteristics. Strikingly, my results are at odds with arguments that overlap decreases IGO survival because organisations compete for the same limited resources. I instead find that overlap bolsters survival – not primarily due to secretariats fostering partnerships with existing organisations but the preferences of powerful states to keep these overlapping institutions alive that serve their interests well. From a broader perspective, my results, therefore, argue for ‘bringing states back in’ in studies of organisational outcomes (Zaccaria 2022). From a normative perspective, while critics (rightfully) point to the efficiency losses due to the duplication of efforts by overlapping organisations, there is little hope for remedial action if this institutional architecture is useful for (powerful) states.

Acknowledgements

Christian Andres, Ruslan Aybazov, Lorenzo Furlani, Tino Good, Stefano Jud, Rosie Keller, Laura Leibundgut, Giulia Parini, Dominik Schneeberger, Johannes Schultz, and Keto Schumacher provided excellent research assistance. Previous versions of this paper benefitted from the generous comments of Hylke Dijkstra, Benjamin Faude, Julia Gray, Diana Panke, Mette Eilstrup-Sangiovanni, Ueli Staeger, Martin Steinwand, and participants in the IO life cycle and ERC NestIOr workshops.

Disclosure statement

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

Data availability statement

Replication material for this article is freely available on Harvard dataverse (https://doi.org/10.7910/DVN/13920X).

Additional information

Funding

Funding from the Swiss Network for International Studies (SNIS) and the University of St. Gallen is gratefully acknowledged. Funding from UK Research & Innovation (MR/V022148/1 – Future Leaders Fellowship) is gratefully acknowledged.

Notes on contributors

Bernhard Reinsberg

Bernhard Reinsberg is a Reader in Politics & International Relations at the University of Glasgow and a Research Associate in Political Economy at the Centre for Business Research at the University of Cambridge. His research is on the policies and politics of international organisations. [[email protected]]

Notes

1 I focus on IGOs – defined as international institutions established by intergovernmental treaty between at least three member states that have a permanent organisational structure and that hold regular meetings among member states (Boehmer et al. 2004; Pevehouse et al. 2021; Rittberger et al. 2019) – although my mechanisms could likely be extended to informal IGOs and multi-stakeholder global governance initiatives.

2 As cases like the League of Nations demonstrate, the decision whether to code an organisation as dead – rather than as replaced – is not straightforward and researchers differ in their assessment (Dijkstra and Debre 2022; Eilstrup-Sangiovanni 2020; Pevehouse et al. 2021).

3 Recent work on European security institutions even more explicitly discusses the idea of an ‘institutional reserve’ whereby member states deliberately create a collective institution and an overlapping patchwork of sub-regional organisations: PESCO remained dormant as long as an alternative set of fragmented institutions were politically more attractive but then got reinvigorated as Brexit created a need for stronger collective defense cooperation (Biermann 2019).

4 I use version COW IGO 3.0 covering 534 IGOs created between 1815 and 2014.

5 According to the COW IGO 3.0 codebook, an IGO is considered ‘terminated when the following words were used to describe the context of the organisation: Replaced; Succeeded; Superseded; Integrated; Merged; Dies’.

6 Probit models with cubic time polynomials are equivalent to survival models (Shiran and Shea 2022). For reasons of consistency, probit models are used in both types of analyses presented.

7 Overlap is time-varying only because the environment changes.

8 I obtain similar results when I use only the shared number of G7 states (Table A11).

9 I thank Julia Gray for making available her data for this analysis.

10 The results are similar when using the shared number of G7 states to perform the sample split (Table A12).