The Death of Major International Organizations: When Institutional Stickiness is not Enough

Major international organizations (IOs) are heavily contested, but they are rarely dissolved. Scholars have focused on their longevity, making institutional arguments about replacement costs and institutional assets as well as IO agency to adapt and resist challenges. This article analyzes the limits of institutional stickiness by focusing on outlier cases. While major IOs are dissolved at considerably lower rates than minor IOs, the article nevertheless identiﬁes twenty-one cases where major IOs have died since 1815. These are tough cases as they do not conform to our institutionalist expectations. To better understand these rare but important events, the article provides case illustrations from the League of Nations and International Refugee Organization, which were dissolved due to their perceived underperformance and a disappearing demand for cooperation. These cases show the limits of the institutional theories of IO stickiness: sometimes member states ﬁnd high replacement costs justiﬁed or consider assets as sunk costs, and IOs may lack agency to strategically respond. This article reﬁnes theories of institutional stickiness and contributes to the institutional theory of the life and death of IOs. Les principales organisations internationales (OI) sont fortement contestées, mais rarement dissoutes. Pour expliquer leur longévité, les chercheurs ont avancé des arguments institutionnels concernant les coûts de remplacement et les actifs de l’institution, mais aussi la capacité des OI à s’adapter et à résister aux déﬁs. Cet article analyse les limites de la persistance des institutions en se concentrant sur des cas particuliers. Tandis que les principales OI sont dissoutes bien moins fréquemment que des OI moins importantes, cet article identiﬁe néanmoins 21 cas de disparition d’OI principales depuis 1815. Ces derniers sont particulièrement difﬁciles, car ils ne correspondent pas à nos attentes en termes d’institutions. Aﬁn de mieux


Introduction
Against the backdrop of the crisis of liberal international order, scholars have started to study the demise of international organizations (IOs).Challenged by illiberal powers, populist governments, and above all the Trump administration, IOs face gridlock, contestation, politicization, loss of legitimacy, and even state withdrawal (e.g., Zürn, Binder, and Ecker-Ehrhardt 2012 ;Hale, Held, and Young 2013 ;von Borzyskowski and Vabulas 2019 ;Lake, Martin, and Risse 2021 ;Heinkelmann-Wild and Jankauskas 2022 ).The ultimate outcome for IOs may be dissolution or "death," which recent studies show is a regular occurrence.Eilstrup-Sangiovanni (2020) informs us, for instance, that 39 percent of the IOs created since 1815 have ceased to exist (cf.Pevehouse et al. 2020 ).Gray (2018) equally finds that only half of international economic organizations are functioning, whereas the other half are either dead or "zombies"-organizations that continue to operate without any progress toward their mandates.
Research on IO dissolution is important, but these large-N studies hardly distinguish between major IOs and minor IOs. 1 This differentiation is significant for three reasons.First, IOs "are organized in radically different ways" ( Koremenos, Lipson, and Snidal 2001 , 761).Thus, the dissolution of the League of Nations was not the same as the dissolution of the International Wool Study Group both in terms of impact to global governance and in terms of the difficulty of dismantling a large organization.Second, major IOs are often considered "sticky" from a theoretical perspective, due to their high replacement costs, institutional assets, and secretariat officials taking on new tasks and fighting for survival (e.g.Keohane 1984 ;Strange 1998 ;Ikenberry 1999 ; Barnett and Finnemore 2004 ;Chorev 2012 ;Jupille, Mattli, and Snidal 2013 ).Third, Debre and Dijkstra (2021) empirically find that institutionalized IOs and those with large secretariats ( > 50 staff) are more robust.This is consistent with the observation of Hooghe et al. (2017 , 17) that only 2/76 major IOs in their dataset have died.
To better understand survival and the limits to institutional stickiness-the ability to stick around even when significantly challenged-this article studies death among major IOs.We consider major IOs that have died as outlier cases or "tough cases" as they do not conform to our institutionalist expectations.By studying these cases in greater detail, we can improve our theories and hypotheses.Building on recent quantitative IO research (e.g., Hooghe et al. 2017 ;Debre and Dijkstra 2021 ;Zürn, Tokhi, and Binder 2021 ), major IOs can be defined as IOs with a large membership, or high institutionalization, or substantial secretariat staff.The article specifies four mechanisms of institutional stickiness that can explain the survival of major IOs even if the demand for cooperation weakens or IOs are no longer perceived to sufficiently perform: member states may repurpose IOs due to their valuable institutional assets or tolerate underperforming IOs due to high replacement costs and IO secretariats and like-minded states may pursue adaptation or resistance strategies to protect their institutions and private interests.While a combination of these mechanisms often ensures the longevity of major IOs, they also clarify when institutional stickiness is not enough: whether member states consider institutional assets as sunk costs or find replacement costs justifiable and when IOs lack the agency to adapt to a changing environment or to successfully avert contestation.
As this is the first article that systematically seeks to study the death of major IOs, it starts by carefully examining the available large-N data to uncover which major IOs have died.It cross-references such existing data with the available primar y and secondar y sources and recodes data where necessary.In total, it finds twenty-one major IOs that have died and shows that major IOs indeed die at a considerably lower rate compared to minor IOs: 10 percent of major IOs compared to 33 percent of minor IOs (with an overall death rate of 140/534 IOs [26 percent]).This finding thus qualifies the higher numbers of IO death mentioned in previous studies, provides further weight for the argument to differentiate between major and minor IOs in the study of IO death, and supports the overall claim that major IOs are sticky.In order to further examine the limits of institutional stickiness (i.e., the twenty-one outlier cases), this article provides two case illustrations of major IOs that were dissolved: the League of Nations and the International Refugee Organization (IRO) due to their perceived underperformance and a disappearing rationale for cooperation, respectively.Both were major 1 Many empirical studies on IO dissolution rely on Correlates of War Intergovernmental Organizations dataset, which includes 534 IOs ( Pevehouse et al. 2020 ), a number of observations that allow for statistical analysis, and that goes back to 1815 allowing for long-term analysis.
IOs in their days and institutionalist theory has a tough time explaining their dissolution.The case illustrations show that while the institutional logics were actually present to some extent in the League and the IRO, other factors weighed more heavily: member states considered high replacement costs justified to dissolve the League, and create a new IO in the United Nations (UN), and they qualified IRO assets as sunk costs.In both cases, IO secretariat officials did not have the ability to strategically respond.
By studying and providing an overview of outlier cases, this article thus contributes to institutional theory and the understanding of the concept of IO death.First, this article summarizes and brings together different institutional logics of longevity, which are now spread throughout the literature, into a coherent argument.Second, it shows that these institutional theories of IOs so far privilege institutional stability rather than considering rare but important instances of death of major IOs.Third, this article also complements recent large-N studies on the death of IOs (e.g., Gray 2018 ; Eilstrup-Sangiovanni 2020 ; Debre and Dijkstra 2021 ) with qualitative case illustrations, thereby providing a richer understanding of the dissolution of major IOs.Finally, scholars have studied the death of major IOs such as the League as individual cases (e.g., Holborn 1956 ;Scott 1973 ;Mat ějka 1997 ;Henig 2010 ), but they have rarely engaged in theory-informed comparative analysis.Overall, this article contributes to our understanding of the death and survival of IOs.
The article starts by discussing the institutional logics of the longevity of major IOs.It continues with the available data on dead IOs and identifies twenty-one major outlier IOs that have died.The article then provides case illustrations of two outlier IOs and revisits institutional theory on this basis.The conclusion reflects on the implications of the findings for the liberal international order.

Institutional Theory and the Stickiness of Major IOs
Ever since the first IO was established following the Treaty of Vienna in 1815, IOs have faced many serious crises.These included world wars and other conflicts in which member states found themselves at opposite ends, hegemonic transitions during which existing institutions were questioned, financial crises and economic nationalism, and changing cooperation problems.A variety of international relations (IR) theories have predicted that IOs cannot withstand such pressures and exogenous challenges do indeed help to explain the death of IOs (see Eilstrup-Sangiovanni 2021 for a review).Yet, institutional scholars have traditionally maintained that major IOs are resilient-like other sticky institutions that tend to persist ( Ikenberry 1999 , 45-46;Pierson 2000 , 490-93).When major IOs occasionally do die, these are rare but nonetheless important events.This section reviews the institutional logics of longevity as a starting point for the survival of major IOs.Realist and notably constructivist theories also provide reasons for IO stability (see Cottrell 2016 , 24-26), but the focus of this article is on institutional stickiness.
Classic rationalist explanations consider IOs in terms of "demand and supply" ( Keohane 1982 ).IOs are established because (1) there exists a demand for cooperation to address cross-border policy problems or to facilitate cooperation within a community of states and (2) states anticipate that IOs can supply or at least facilitate such cooperation ( Moravcsik 1993 ; Abbott and Snidal 1998 ;Hooghe, Lenz, and Marks 2019 ;Rittberger et al. 2019 ).The flipside is that when demand and supply no longer meet, IOs are in trouble.If the rationale for cooperation among a group of states disappears or IOs fail to achieve the goals for which they were created, the rationalist argument goes, they will be dissolved or fall into desuetude.2Institutional theory provides a critique against such rationalist demand and supply dynamics: institutions persist because there is no perfect competition, there are feedback loops and lock-in effects, and many actors privately benefit from existing institutions.In the literature, there are therefore multiple distinct but mutually reinforcing logics of institutional stickiness that help to explain why major IOs tend to survive.Since these logics are not always fully spelled out and rarely compared, it is useful to first summarize them.We distinguish between demand-and supply-side problems and between institutional logics that focus on calculations by the member states and those that focus on the agency of IOs themselves ( table 1 ). 3R theories uniformly point out that the demand for cooperation between states fluctuates over time.This puts pressure on IOs.Scholars have, however, shown that IOs may outlive their original purpose because of the investment states have made in terms of institutional assets .This is a state-centric institutional logic that explains IO survival even when demand runs out.IO institutional assets remain valuable and states can use them for other cooperation purposes.Wallander (2000) notably explains the North Atlantic Treaty Organization's (NATO) persistence after the Cold War with reference to institutional assets (such as its established norms and practices but also its command structure).These institutional assets remained valuable in the post-Cold War period, and NATO was repurposed to also engage in collective security and crisis management operations in addition to its collective defense mandate.NATO is not the only example.IOs set up to administer post-Second World War (WWII) reconstruction in Europe-such as the Organization for European Economic Cooperation (OEEC) and the International Bank for Reconstruction and Development (World Bank)-were also repurposed once Western Europe was reconstructed ( Hahn 1962 ;Marshall 2008 ;Leimgruber and Schmelzer 2017 ).Major IOs with considerable institutional assets may therefore survive even if the original rationale for cooperation disappears.
In addition to member states' interest in preserving institutional assets, major IOs also have some agency of their own (e.g., Barnett and Finnemore 2004 ;Hawkins et al. 2006 ), which they can use to adapt IOs to a changing external environment, for instance, by expanding the mandate and scope.This is thus an IO-centric institutional logic why IOs survive even when the original rationale for cooperation runs out.Weinlich (2014) details how the UN Secretaries-General and the Secretariat have been central in developing peacekeeping as a core task not originally envisioned in the UN Charter.Hall (2016) similarly shows how the High Commissioner for Refugees, International Organization for Migration, and UN Development Programme have moved beyond their mandates to engage with climate change as an emerging international problem.IO leaders and officials are not alone in their attempts to expand and change their mandates.They typically work with like-minded states that equally have a stake in the organization ( Dijkstra 2017 ).A key example is once again NATO.After the Cold War, US senator Richard Lugar famously noted that NATO should go "out-of-area or out-of-business," implying that the NATO mandate should no longer be limited to the geographical North Atlantic area.This became a mantra that the United States then pursued in conjunction with NATO officials (e.g., Rosenfeld 1993 ;Tuohy 1993 ), which ultimately resulted in military operations from the Western Balkans to Afghanistan.NATO even has recently adopted a China strategy, which is a long way from the original aims of the 1949 Washington Treaty (e.g., Schuette 2021 ).IOs can thus affect to some extent the demand for their services and by helping the organizations expand into new areas in support of survival.
In addition to changing demands, the supply by IOs may also be considered insufficient, which may result in dissolution and replacement.Yet, despite reoccurring discussions over the varying performance of IOs ( Gutner and Thompson 2010 ;Lall 2017 ;Sommerer et al. 2021 ), a member states-centric institutionalist logic argues that such market dynamics rarely apply, because major IOs cannot be easily replaced by other institutions-other IOs, informal institutions, or other forms of governance.The negotiation of replacement institutions is often complex and involves considerable uncertainty about future payoffs.The replacement costs are simply too high.In this regard, Jupille, Mattli, and Snidal (2013 , 7) show that states, even when unhappy with certain IOs, rarely create new ones and prefer to "stick with the institutional 'devil they know' as long as the status quo produces results above some minimum threshold."Keohane (1984 , 102) tells us that "[t]he high costs of regime-building help existing regimes to persist." Where states may object to high replacement costs, IOs also work actively to resist dissolution and to fend of existential challenges.This is the final, IO-centric logic, explaining why IOs may persist.Kaufman (1976) notes that officials in public agencies "are not helpless, passive pawns in the game of politics as it affects their lives; they are active, energetic, persistent participants.The motives […] to preserve the organisations to which they belong are very strong" ( Kaufman 1976 , 9; see also Strange 1998 ).IO officials with the help of liked-minded states are actively shielding their organizations in case of contestation ( Chorev 2012 ;Hirschmann 2021 ;Schuette 2021 ).They can use a range of behavioral and discursive strategies and IOs increasingly engage in public relations offensives and other forms of self-legitimation.Secretariat officials in major IOs can thus affect to some extent the demand and supply dimension and support the survival of their organizations.
The academic literature thus includes multiple logics of institutional stickiness, and these four explain in different ways the persistence of IOs.Importantly, they are often mutually reinforcing.Indeed, the work of Barnett and Finnemore (2004) on institutional pathologies shows, at once, that IOs are often driven by their bureaucratic considerations and that member states have a difficult time rectifying underperformance due to the high cost of institutional replacement.Furthermore, while Wallander (2000) rightly points at the institutional assets of NATO, the survival of the Alliance after the Cold War was also ensured through close cooperation between the United States and the services of the Secretary-General to expand the mandate.The reason why IOs persist is therefore not a question of assets or replacement costs or adaptation or resistance strategies, but oftentimes a combination of these different logics at the same time.
These institutional logics of the longevity of IOs, both individually and in combination, particularly apply to major IOs.Most institutions have a degree of stickiness, but these institutional logics clarify that there is a significant difference between major and minor IOs.After all, replacing minor IOs is exponentially less complex, minor IOs have less assets, and they typically do not employ many staff members who can fight for survival.Major and minor IOs are thus like apples and oranges when it comes to institutional stickiness.Indeed, as Debre and Dijkstra (2021) show, large IOs are significantly less likely to die compared to smaller institutions, a finding that is also confirmed in the next section: only 10 percent of major IOs have died compared to 33 percent of minor IOs.
Beyond this overall expectation of longevity, this summary of the different institutional logics is also beneficial in understanding when stickiness for major IOs is not enough.By attaching value to institutional assets , we can distinguish between more general assets that are relevant for different types of policy problems and those that are specific to certain policy problems.The former may be repurposed while the latter type of assets may be considered as sunk costs when the rationale for cooperation disappears.Furthermore, while member states might tolerate some underperformance, there clearly is a threshold at which point they will actively intervene and incur replacement costs resulting in the dissolution, or death, of a major IO and the creation of a new IO.IO agency to adapt the organization or resist the different pressures will also be conditional upon whether IO staff can leverage resources and engage with like-minded actors, but this also clearly requires strategic leadership and organizational direction.
The four institutional logics are thus not deterministic and institutional stickiness has its limits.These logics help to explain the robustness of many major IOs, but we should also pay attention to instances when they do not apply.These four institutional logics can be tested empirically across the population of IOs and can also be used as yardsticks to assess death and survival in individual cases.This overview of the different institutional logics therefore serves as a background for the remainder of the article, which studies the death of major IOs.First, however, the next section will serve to conceptualize major IOs from the total population of 534 IOs in existence since 1815 and empirically identify which have died despite our expectations of institutional stickiness.

Descriptive Statistics: Which Major IOs Have Died?
To understand the limits to the stickiness of major IOs, it is important to first provide a descriptive overview of all major IOs that have died since 1815.Our starting point is The Correlates of War Project Intergovernmental Organizations (COW-IGO v3.0), which includes a total of 534 IOs ( Pevehouse et al., 2020 ).COW-IGO is an important resource and while it is not always accurate at the level of individual IOs (as we will show below), it does provide a good overall picture and an entry point for studying IO death.To distinguish major from minor IOs, the article follows standard definitions of IOs that include membership, plenary meetings, secretariats ( Pevehouse et al. 2020 ), and institutionalization ( Hooghe et al. 2019 ).We consider that major IOs have a large membership (50 percent of existing system member states), which is related to the membership dimension of the IO definition, or high levels of institutionalization (such as an adjudication mechanism), which is related to the institutionalization aspect, or substantial administrative resources (fifty or more staff members), which is related to the secretariat. 4The purpose is to be relatively inclusive with regard to the definition of "major IOs" simply because we want to, as a first step, get an overview.Furthermore, since Hooghe, Lenz, and Marks (2019) find a trade-off between the number of member states and the depth of cooperation, we want to include IOs that can be considered "major" on any of the three criteria.Other scholars are more restrictive in defining major IOs (e.g., Hooghe et al. 2017 , 14-15;Zürn, Tokhi, and Binder 2021 ). 5 We want to avoid a bias against pre-1945 IOs, which carried out important functions in their days, such as mapping the world, sharing statistical data on agriculture, or overseeing quarantine rules to prevent the spread of plague and cholera.In total, our inclusion criteria lead to 153 major IOs since 1815 ( table 2 ). 6he purpose of distinguishing major IOs is that they are more likely robust, and this goes for each of the three inclusion criteria.If the number of member states increases, IOs are more likely to survive (cf.Eilstrup-Sangiovanni 2020 ), because it raises the replacement costs .Establishing IOs with a large membership is more complex than IOs with a smaller membership, simply because more member states are involved in the negotiations.Due to such transaction costs, IOs with a large membership will be much more difficult to replace.IOs with a large membership are also less dependent on the idiosyncrasies of individual member states.Thus, the article includes IOs that at some point of their lives included 50 percent of existing states as members.
While an arbitrary cutoff point, Hooghe, Lenz, and Marks (2019) show that most IOs are either universal or more limited in terms of membership with few IOs positioned in between.The 50 percent membership threshold does vary significantly over time.For IOs during the interbellum, for instance, 50 percent means thirty-three members (out of around sixty-five states), whereas currently it requires almost one hundred members (out of nearly two hundred states).Data are available in COW-IGO v3.0 ( Pevehouse et al. 2020 ).
In total, there are fifty IOs that had at least 50 percent of existing states as members at some point in their existence.We thus only include IOs that are in the ninetieth percentile in terms of their membership.
The second inclusion criterion for major IOs concerns high levels of institutionalization, which is both related to the logics of replacement costs and institutional assets .Establishing high levels of institutionalization by granting IOs considerable authority requires significant negotiations by the member states because it involves sovereignty costs ( Abbott and Snidal 2000 ).At the same time, institutionalization may involve assets, which go with such a grant of authority, such as adjudication courts with track records of case law, headquarters, or internalized norms and practices ( Wallander 2000 ).Once highly institutionalized IOs are in place, chances are smaller that member states will abandon or replace them ( Debre and Dijkstra 2021 ).The article relies on Karreth and Tir (2013) , who build on Boehmer, Gartzke, and Nordstrom (2004) , to differentiate between low, medium, and highly institutionalized IOs.Highly institutionalized IOs "contain mechanisms for mediation, arbitration and adjudication, and/or other means to coerce state decisions" ( Boehmer, Gartzke, and Nordstrom 2004 , 18).In total, there are forty-five IOs with high levels of institutionalization.This equals the membership criteria, since we only include IOs in the ninetieth percentile in terms of institutionalization.
The final inclusion criterion for major IOs concerns substantial administrative resources, which relate directly to the institutional logics of adaptation and resistance strategies .It has long been recognized that secretariat officials are key to the functioning of IOs (e.g., Barnett and Finnemore 2004 ).They are in a position to foster cooperation between member states as well as resist and fight off pressures on IOs ( Chorev 2012 ;Gray 2018 ;Debre and Dijkstra 2021 ).While the indicator of institutionalization concerns delegated authority, administrative resources are about capacities of secretariats to act.The article follows Debre and Dijkstra (2021) , who code IOs with fifty or more staff members as "large."Below fifty staff members, it is unlikely that an IO has a substantial policy directorate that is required to have agency in policy-making ( Debre and Dijkstra 2021 ).The Yearbook of International Organizations provides staff numbers for the last reporting year of an IO.While some IOs have acquired substantial staff resources over time, Debre and Dijkstra (2021) show that very few IOs that start with less than fifty staff members gain more than fifty staff members during their lifespan.In other words, IOs with a small secretariat typically keep a small secretariat.In total, there are 126 IOs with 50 or more staff members.Because this cutoff point includes many more major IOs (seventy-fifth percentile), we include a robustness check with very large administrative resources ( ≥300 staff members), which results in forty-nine major IOs (ninetieth percentile) (see table 2 ).
When adding up the IOs with a large membership, or high institutionalization, or substantial administrative resources, the article thus identifies 153 major IOs in total (29 percent of all 534 IOs).16 of these 153 are coded as "dead" in the COW-IGO v3.0 dataset ( Pevehouse et al. 2020 ).Death means that IOs no longer have (1) three or more member states, (2) a plenary meeting once every ten years, or (3) a secretariat and correspondence address, and did not continue under a different name or as part of a different IO.The COW-IGO includes several major IOs that were legally replaced or integrated, but these are instances of institutional change and development rather than termination.The logics of institutional stickiness that help to explain why the death of major IOs is rare do not apply to such cases of replacement or integration, because the very existence of IOs was not challenged in these cases. 7s a first cut, on the basis of the data from the COW-IGO, the overall picture is clear.At 10 percent, the number of major IOs that have died is considerably lower than the 33 percent of dead minor IOs ( table 2 ).This finding is consistent for each of the three inclusion criteria for major IOs as well.Only 12 percent of the IOs with a large The Death of Major International Organizations membership has died, 13 percent with a high level of institutionalization, and 5 percent of the IOs with substantive administrative resources (only 2 percent for ≥300 staff members).Of the eleven IOs that possess all these characteristics, not a single IO has died.In other words, the institutional logics hold up.At the same time, the COW-IGO data show that still sixteen major IOs have died since 1815 and the reminder of the article focuses on those "outlier" cases to explore the limits of institutional stickiness.
Table 3 provides an overview of all these large dead IOs.Yet, as with any dataset, there are question marks about borderline cases and missing data.The COW-IGO is an important resource, but if we are to make more definite statements about individual observations, we need to have more detail about individual IOs.We have therefore gone carefully through the COW-IGO dataset to identify additional "dead" major IOs.Our strategy was two-fold.
First, we have gone through all the 140 cases of dead IOs with missing data on any of the three indicators to see whether they could also be reasonably coded as "major."We found this to be the case for the International Allied Rhineland High Commission (IARHC) and the IRO.For both IOs, there are no available data on institutionalization and staff in Karreth and Tir (2013) and the Yearbook , respectively.Yet, the secondary literature makes clear that the IARHC had substantial courts and an administrative staff.The IRO similarly had nearly 2,900 international staff in 1949 ( Holborn 1956 , 99), which puts it among the most major IOs on administrative resources.
Second, we have gone through all cases of major IOs that were coded as "replaced/integrated" in COW-IGO to see whether these could equally be coded as "dead."We found this to be the case for the International Institute of Agriculture (IIA) and the League, both of which were replaced by different IOs with a different mandate, membership, and secretariat location after WWII. 8The succeeding institutions differed significantly and were not simple replacements.States established entirely new IOs and then dissolved the IIA and League.The League is also recoded as "dead" in Eilstrup-Sangiovanni (2020) .Finally, we added the Arab Maghreb Union (AMU), which has not met since 2007 but still has an operating secretariat.Because COW-IGO data end in 2014, the AMU is not yet coded as "dead" in COW-IGO as it requires ten years of no plenary meeting.However, that period has now passed.
In total, we get to around 21 (16 + 5) dead major IOs as listed in table 3 .We fully acknowledge that some of these cases are borderline and that the total number of dead major IOs varies depending on how stringent one defines the cutoff points for both major IOs and for death.However, in general, it is safe to say that there are more than a dozen dead major IOs and less than two dozen.By providing transparency and the overview, we shed light on the phenomenon of dead major IOs and the limits to the logics of institutional stickiness.
What is striking about this overview of twenty-one major IOs is that their causes of death are very varied (see table 3 ).Many of these causes are related to some of the major IR theories, but it is difficult for a single theory to explain IO death (cf.Eilstrup-Sangiovanni 2021 ).It is nonetheless fair to say that most major IOs occasionally face serious and existential challenges and that these are well captured in terms of the demand for and/or supply of cooperation.
When it comes to the decreasing demand for cooperation, we can identify IOs where cooperation is clustered around key (hegemonic) member states.In case of a decline of those states, the relevant IOs may be in trouble.This is most notable following the collapse of the Soviet Union, which in turn reduced the demand for the Council for Mutual Economic Aid and Warsaw Treaty Organization ( Mat ějka 1997 ;Mastny and Byrne 2005 ).The Commonwealth Air Transport Council became less relevant over time as the Commonwealth became less cohesive and the International Civil Aviation Organization dealt with many aviation questions as well ( Mackenzie 1993 , 119-21).The AMU, on the other hand, is an example of a conflict between two member states-Algeria and Morocco-which as a secondary effect reduced demand for cooperation ( Zoubir 2012 ; Hernando de Larramendi 2019 ).Apart from demandside challenges around states, we can identify deaths due to reduced demand for policy: Some IOs are dissolved because their mandates have been (partially) fulfilled or because they expired (IRO, OEEC, and IARHC) ( Holborn 1956 ;Hahn 1962 ;McDougall 1978 ;Griffiths 1997 ;Pawley 2007 ;Orchard 2014 ).
Various major IOs have also died due to supply-side problems.Notably are interbellum IOs, which were dissolved after WWII to make way for a more ambitious global order (International Office of Public Hygiene; IIA; International Technical Committee of Legal Experts on Air Questions; League) ( Wilberforce 1947( Wilberforce , 2007 ; ;Clavin 2013 ;Cueto, Brown, and Fee 2019 ).States also considered that some of these massively underperformed (e.g., League) or were sustained by the "wrong" member states (IIA hosted by Fascist Italy; see Herren 2017 ).Two IOs were dissolved as the services they offered were not sufficiently advanced for the post-WWII era (Central Bureau of the International Map of the World, International Commission for the Decennial Revision of the International Lists of Diseases and Causes of Death) ( Pearson et al. 2006 ;Pearson and Heffernan 2015 ).
Beyond the exogenous shock of WWII, there are several different supply-side challenges that have resulted in the death of major IOs.Two IOs, for instance, were dissolved due to competing IOs (Intergovernmental Bureau for Informatics and Western European Union) ( Wessel 2001 ;Bailes and Messervy-Whiting 2011 ;Pohle 2013 ).Two other IOs were privatized (African and Malagasy Coffee Organisation and International Telecommunications Satellite Organization) ( Frieden 1994 ;Feder 2001 ;Katkin 2005 ).The Central American Research Institute for Industry was ultimately dissolved because of mismanagement ( Remiro Brotons 2003 ).Finally, the Organization for the Management and Development of the Kagera River ended after secretariat staff had to evacuate due to the war in Rwanda ( Uganda; Apathy Finally Kills the Kagera Basin Organisation 2001 ).In none of these cases, the demand for cooperation really disappeared.Rather these IOs failed to sufficiently meet the demand through their services.
This section has distinguished major from minor IOs and found that 10 percent of major IOs have died compared to 33 percent of IOs.This provides strong support for institutional stickiness.The section has also identified twenty-one major IOs that have died.This requires us to study how IOs respond to demand-and supply-side challenges and reassess institutional stickiness.The next section zooms in on the case illustrations of the League (lack of supply) and the IRO (lack of demand).2020) , institutionalization by Karreth and Tir (2013) , secretariat staff by Debre and Dijkstra (2021) , and Yearbook of International Organizations ; principal causes of death is own coding based on secondary sources.

Case Illustrations of Major IOs and the Lack of Stickiness
This section explores two examples of major dead IOs.The purpose is to better understand these outliers and the limits to institutional stickiness.What we want to do with these case studies is to trace any presence of the four institutional logics and to see why they did not stick.We thus study why institutional assets were not considered valuable by the member states and why member states were willing to incur replacement costs .We also study why IO secretariats, officials, and their like-minded allies were not able to put up adaptation and resistance strategies .To carefully link the case illustrations to the institutional logics, the article uses the wellknown case of the League as an instance of an IO that was widely perceived as failing to supply collective security during the 1930s.We are therefore wondering, in particular, why member states did not worry about replacement costs and why League officials did not put up resistance strategies (see T able 1 ).The IRO is an instance of disappearing demand for cooperation.As the demand for refugee support in Western Europe declined in the early 1950s, so did the demand for the IRO.Given, however, that the IRO was a major IO, we are wondering why member states did not make use of the institutional assets and why IO staff did not pursue adaptation strategies to deal with the continuing refugee problems elsewhere.
The League and the IRO are therefore ideal types when it comes to lack of demand and lack of supply, which allows us to carefully trace the (lack of) institutional logics.It should also be said that from the population of twenty-one major dead IOs, the League and IRO were among the strongest institutions and, from an institutional perspective, therefore least-likely to die.Since the purpose is to explore the limits of institutional theory rather than to test the theory, this is a valid selection strategy ( Seawright and Gerring 2008 ): we want to shed light on the most-likely cases for institutional theory to understand when institutional logics fall short.
The next section presents details on both cases.For the League, we find that member states considered the high costs of negotiating a new institution with the UN justified and that reduced League staff, encircled in Geneva and scattered around the globe, was unable to fight for their organization.For the IRO, we find that the assets (refugee camps in Europe) were too specific and thus sunk costs and at-tempts to expand the mandate proved futile as the United States sought to close the organization.

League of Nations
The League is undoubtedly the most studied dissolved IO ( Pedersen 2007 ).We use this case as an instance of lack of supply .While the world wars of the twentieth century very much highlighted a need for collective security, the League is widely considered as a failure in that it did not supply such security in the 1930s.The League is often portrayed as an inevitable casualty of WWII, but from an institutional perspective the story is more puzzling: it was kept alive during the war, but it was not revived and indeed dissolved once the war was over.While most contemporary historians point at continuities between the League and the postwar institutions ( Pedersen 2007 ;Clavin 2013 ;Gram-Skjoldager and Ikonomou 2019 ), it is striking that the logics of institutional stickiness do not seem to apply: the founders of the UN established a new organization instead of trying to reform the League and the remaining League officials were sidelined and had little agency to craft a resistance strategy.Several League assets were simply transferred to the UN.This thus raises the questions, from an institutional perspective, why the member states were not worried about replacement costs and why the League officials could not put up a resistance strategy .
Set up as a collective security organization and concerned with "[a]ny war or threat of war" ( The Covenant of the League of Nations 1919 , article 11), the League obviously failed to prevent WWII.Its principal cause of the death is, however, normally traced to the conflictual relations between the member states in the 1930s and the League's failure to play a meaningful role ( Hinsley 1963 , 309-22).The Manchurian conflict between China andJapan (1931-1933) resulted in the withdrawal of Japan from the League ( Burns 1935 ;Scott 1973 , 207-41).Nazi Germany left the same year and Fascist Italy withdrew in 1937 after the League sided with Ethiopia ( Henig 2010 ).The Soviet Union was expelled after its invasion of Finland ( Beck 1981 ).This meant that the Soviet Union harbored considerable hostility toward the organization and, as such, "the act of expulsion became the League's death warrant" ( Beck 1995 , 178).Such a lack of performance and the absence of the two of the winners of WWII-Soviet Union and United States-from the League ultimately trumped any concern about possible replacement costs (see further below).
The agency of the League itself had also been weakened and as a result its officials played only a limited role in the ultimate negotiations of the UN.Secretary-General Joseph Avenol described the period prior to WWII as "demiguerre" (as cited in Beck 1995 , 175).Strikingly, neither the Secretary-General nor the like-minded two remaining permanent Council members, France and the United Kingdom, did much to defend the interests of the League.France and the United Kingdom, for instance, dealt with Italy bilaterally during the Ethiopian crisis, preferring to maintain good relations in the context of the threat from Nazi Germany ( Henig 2010 , 163-66).Except for the expulsion of the Soviet Union, "[k]ey events were kept away from Geneva […] which was steering, or rather being steered by leading members , clear of major international problems" ( Beck 1995 , 175).Avenol was described as an "unofficial agent in the League" of the great powers (James Barros in Beck 1995 , 178) rather than an autonomous operator with agency.On the day of the Anschluss in 1938, he was planting trees on the lawn of the Palais des Nations in Geneva ( Beck 1995 , 185).
The League continued operating during WWII, but was severely affected by it.Its host Switzerland had opted for neutrality and could therefore not be seen as supporting the League with its bias toward France and the United Kingdom ( Walters 1952 , 801-802).Secretariat staff had already been significantly reduced when by mid-1940, Geneva became encircled and isolated.Avenol resigned and was replaced by his deputy Sean Lester who remained in Geneva ( Fosse and Fox 2016 , 171-88).Many other League officials fled Geneva and set up shop elsewhere including Princeton, Montreal, London, and Washington ( Walters 1952 , 809;Clavin 2013 , 258-66;Fosse and Fox 2016 , 191-93).The departure from Geneva and the "globalization" of the League meant that many officials got involved in the new postwar institutions ( Clavin 2013 , 267-340;Kott 2014 ).At the same time, cut off from the world, with very little resources, and no Swiss support, the League in Geneva had little agency to fight for survival.Officials still tried with a range of resistance strategies.Lester, for instance, considered it "essential that some element of the League remained alive in Geneva […] to keep the ideals of international cooperation and organization alive for the post-war era" ( Clavin 2013 , 260).League officials also drafted the 1944 London Report "to counteract the declining image of the League of Nations" ( Auberer 2016 , 393).We can therefore identify the presence of the IO-centric institutional logic, yet the League's agency and ability to pursue it were heavily constrained.
Continued wartime activity raises the question whether the League died or was merely replaced by the UN.During its last assembly in Geneva, British statesman Robert Cecil paraphrased an old French saying by proclaiming that "The League is dead, long live the United Nations."The UN also took over the assets (office buildings, furniture, books, archives, and some capital) of the League in Geneva ( UN General Assembly 1946 , articles 1-2; Myers 1948 ), which would become the second headquarters location of the UN, and 200 League officials went on to work for the UN ( Gram-Skjoldager and Ikonomou 2019 , 421).Clavin (2013 , 267-340) furthermore shows how the 1939 Bruce Report of the League provided the blueprint of the UN Economic and Social Council.On the other hand, the UN was a different IO with the United States and Soviet Union as members, a new UN Charter, and the UN headquarters eventually built in New York.The UN was therefore a qualitatively different IO.
From the perspective of institutional stickiness, the question remains why the League was not revived under its own name and ultimately dissolved in 1946 or, in other words, why the member states went to great lengths to negotiate a new IO.The answer is ultimately two-fold.First, states wanted to write off the doomed League and give the UN a fresh start ( Goodrich 1947 , 3-4).During the negotiations about the UN and its Security Council, which took place largely during the war, there was thus little talk about continuity from the League.While League officials attended the San Francisco UN conference negotiated in the Opera House in 1945, they initially found themselves without assigned seats, thereby being spectators at their own funeral ( Clavin 2013 , 341).The absence of the League was, second, critical from a political perspective.Being expelled from the League in 1939, the Soviet Union was openly hostile ( Goodrich 1947 ).Furthermore, the US Senate had failed to ratify the Treaty of Versailles resulting in the absence of the United States from the League.The UN Charter therefore had to be something new and member states were willing to accept the replacement costs.

International Refugee Organization
The IRO was set up as a temporary IO shortly after WWII in 1946 to deal with 1.5 million refugees and displaced persons in Europe.It was closed in 1952 ( Holborn 1956 , 559-61).We use this case as an instance of lack of demand : the IRO performed reasonably well, but the problem of refugees in (Western) Europe simply decreased as time passed by.Still, from the institutional perspective, the case of the IRO remains a puzzle.While other major IOs-such as the OEEC, NATO, or World Bank-also have time-limited mandates or are tasked to address a specific problem, it rarely happens that the member states establish a major IO with high levels of institutionalization and substantial bureaucratic resources and subsequently close it down after a couple of years.This thus raises the questions why the member states did not value the institutional assets of the IRO and why IRO officials and like-minded actors did not adopt a mandate adaptation strategy to deal with refugee problems elsewhere beyond Western Europe.What this case illustration makes clear is that the assets (refugee camps in Western Europe) were rather specific and became sunk costs.Moreover, the membership constellation became more complicated as the Cold War took off and the United States ended up paying much of the costs of the IRO.The United States then focused on reducing and dissolving this American-led IO.
The problem of refugees became increasingly pronounced already during the interbellum and resulted in various forms of international cooperation including under the League (see Orchard 2014 , chapter 5).In addition, the Intergovernmental Committee on Refugees (IGCR) and UN-RRA were established, respectively, in London in 1938 andWashington in 1943, to deal with refugees and displaced persons during and particularly after WWII.The scale of the problem was immense with 6.6m displaced persons outside their countries, 33m displaced persons inside their countries, and more than 750,000 refugees from German and Soviet territories ( Orchard 2014 , table 6.1).While many displaced persons were soon repatriated, once WWII had ended, there were difficulties with refugees, displaced persons, and prisoners of war from Central and Eastern Europe and the Soviet Union ( Ristelhueber 1951 , 177).Particularly the United States objected to forced return on the basis of humanitarian principles ( Orchard 2014 , 146-52), whereas the Soviet Union expected all its citizens to come home.
Conflict over the course of the UNRRA, the end of the League's High Commissioner for Refugees, British unwillingness to sustain the IGCR, and the new problem of how to resettle refugees and displaced persons (rather than return them) resulted in the creation of the IRO as a temporary UN agency in February 1946.Yet, during the negotiations of the IRO states split along East-West lines on the definition and level of support to refugees and displaced persons ( Ristelhueber 1951 , 178-80).When finally established, the IRO therefore had only member states from the Western hemisphere.The IRO nevertheless provided refuge through camps and resettlement support to 1.5 million displaced persons until the end of its mandate (see Rucker 1949 ;Ristelhueber 1951 ;Holborn 1956 for a discussion of the activities).It employed nearly 2,900 international officials in 1949 ( Holborn 1956 , 99) and had a total budget of around $400m during its less than five years of existence ( Holborn 1956 , 122).
More than half of the budget was paid for by the United States ( Holborn 1956 , 122), also because the IRO membership remained smaller than anticipated, and the United States soon started behaving like a dissatisfied customer ( Orchard 2014 , 168-70).At the same time, with the emergence of the Cold War, the United States wanted to avoid handing over more responsibilities to the UN where the Soviet Union had a strong position.This tension became critical once negotiations over the dissolution of the IRO started.While the United States could not prevent the establishment of the UN High Commissioner for Refugees in 1949, highlighting the continued need for refugee cooperation worldwide, the role of the High Commissioner was more restricted and the office did not receive any operational funds from the UN.9When the IRO disbanded in 1951-1952, it handed over responsibilities mainly to host countries, such as France, Germany, Austria, and Italy ( Ristelhueber 1951 , 221-22).The staff had already been reduced dramatically to 1,684 officials at the end of 1950 and 677 officials at the end of 1951 ( Holborn 1956 , 99).IRO was therefore time-limited and its function in Europe had largely been fulfilled leaving assets (such as temporary refugee) camps without much value.The organization was American sponsored and there was no drive by its staff for survival.While the UN High Commissioner for Refugees was a much more restricted office at creation, it soon however expanded its mandate and activities as the worldwide refugee problem continued to demand international cooperation.

Revisiting Institutional Stickiness
This article started off noting that IOs, like other institutions, often stick around even if demand for IOs runs out or IOs are perceived to insufficiently supply the functions for which they were created.Member states may not want to incur replacement costs to set up other (more efficient) IOs or they may value the institutional assets of IOs.At the same time, the IOs themselves have also some agency to adapt to changing environments or to resist challenges to their organization.These logics are relevant for major IOs, as their replacement costs, assets, and agency are exponentially higher than those of minor IOs.As such, major IOs are likely particularly sticky and this article has empirically indeed shown that IOs with a large membership, or high institutionalization, or large resources are dissolved considerably less than their smaller cousins ( table 2 ).The examples of longevity and stickiness are plentiful, from NATO to the World Bank, which have adapted to changing demands to the many "zombie" IOs ( Gray 2018 ) that live on in spite of little real-world impact.Nevertheless, the article has also identified some twenty-one major IOs that have died and discussed two case illustrations.This brings us back to the question what it takes for major IOs to get dissolved.
While the causes of death of major IOs are varied, they are well captured in terms of the demand for and/or supply of cooperation.The challenge of changing demands for cooperation among states is part of most IR theories: from realists pointing at power transitions and changing alliance patterns, liberals at changing societal problems, and constructivists more generally at transformative potential.Regardless of the origins of these changing demands for cooperationbe it that the member states are no longer friends or yesterday's problems have been solved-this puts pressure on IOs.This article has evaluated two institutional logics that help to explain the survival of IOs even when demands for cooperation change: IOs may possess valuable assets and IOs may have agency for adaptation .This article has used the case illustration of the IRO, a major IO whose mandate ran out and that was dissolved.
This IRO case highlights the need for a better consideration of the value of institutional assets and when they are, in fact, sunk costs.While many of the examples of surviving IOs (e.g., NATO, World Bank) have assets that include centralized headquarters and relatively generic (economic/security) expert bureaucracies, the assets of the IRO (decentralized refugee camps across Europe) proved less valuable to a continued global refugee problem.The fact that IOs have assets, such as buildings or institutional structures, is not enough; the question is how well assets fit to the continued task at hand.The other question is why the IRO-as a formal UN organization-did not develop into a more global refugee IO.There is little evidence, in this respect, of any such attempts on the side of IRO staff.The United States as the main sponsor wanted this IO closed.Since there was no possibility for further prolongment, IRO staff focused on wrapping up responsibilities in the best way possible ( Ristelhueber 1951 , 221-22).
Supply problems present a second challenge to the survival of IOs.As Klabbers (2009, 298) notes, "the rationale behind an organization will usually continue to exist; it is merely the institutional arrangements which are deemed unsuitable."Yet, in spite of many IOs that do not optimally perform ( Gutner and Thompson 2010 ;Lall 2017 ;Sommerer et al. 2021 ), there are two institutional logics that help to explain persistence: member states often find it too cumbersome to replace major IOs and IO staff and other like-minded invested parties have incentives to resist and to fight for survival.
The case illustration of the League shows the limits to the institutional logics of replacement costs and IO agency to pursue resistance strategies.The founding member states of the UN did not seem to consider the costs of negotiating a new institution and dissolving the League as a factor.Important was not just the overall perception of the League as an institutional failure, but also the varying membership and the new great power composition after WWII.While there were clearly attempts by League officials to promote the continuity of their institution, they were limited in their ability to do so when Switzerland became encircled.Even though secretariat officials can be powerful advocates for their organizations, simply having staff is not enough.For major IOs to pursue resistance strategies, they also require leadership, organizational direction, and simply to be present.
By critically studying outlier cases of major IO death, we can thus further refine institutional theory.Outlier cases provide theory with a tough time and allow us to better understand the underlying mechanisms.This article shows that we should consider the actual negotiation and replacement costs, the value of institutional assets, and the ability of secretariats to respond strategically when analyzing the survival and death of IOs.In general, institutional theory remains convincing when explaining the longevity of major IOs.Even in the cases empirically discussed, we see some institutional stickiness.League officials continued their careers in all sorts of newly established UN institutions, thereby shaping the postwar environment.Furthermore, there was some continuity from the IRO to the UNHCR against the wishes of the United States and the UNHCR soon became the focal institution for global refugee problems.Nevertheless, stickiness was hardly absolute and should be considered the cumulative effect of several institutional variables.This in turn also allows for an institutional theory on the survival and death of IOs.

Conclusion
In the context of the current crisis of liberal international order, several scholars have empirically started studying the death of IOs and they show that IOs are not immortal.The trouble with these large-N studies is that they hardly distinguish between major and minor IOs.Yet, there are good reasons to assume, based on institutional theory, that major IOs-those with many member states, or high levels of institutionalization, or substantial administrative resourcesare a class of their own.The four identified logics of stickiness are substantially more relevant to major than minor IOs, which may be easier to replace, have only very limited assets, and typically do not have secretariat agency.To complement the recent large-N empirical studies on the death of IOs, this article therefore wondered what it takes for major IOs to die.
The article has identified some twenty-one major IOs that have died since 1815.The percentage of major IOs that have died (10 percent) is, however, considerably smaller than the percentage of minor IOs that have died (33 percent).This confirms the institutionalist logics and implies that these twenty-one major IOs are outlier cases.The death of a major IO, nevertheless, remains an important event in IR, and this article has zoomed in on two instances by providing case illustrations on the basis of secondary literature and official documents.What has become clear is that the principal causes of death vary greatly among these major IOs.Yet, the case illustrations neatly show that there are limits to institutional stickiness.As a final step, the article has revisited institutional theory.A key finding is that institutional theoryand institutional stickiness-works differently across cases.In some instances, states may not worry about possible replacement costs; they may consider assets as sunk costs, or secretariat officials may not have the ability to fight for survival.
What do we make of these findings?First, institutional theory is and remains critical to understand the robustness of IOs.For all the talk of the crisis of liberal international order, major IOs tend to be robust and they will likely stick around.Second, where major IOs do occasionally die, we need to pay much more attention to how their precise causes of death relate to their institutional design.Even within the relatively coherent theory of realism, for instance, some logics may focus on the divergence of the member states, while others privilege hegemonic decline, yet both suggest different causal pathways to death.This is important because, third, the causes of death help to explain how the dissolution process unfolds, including whether major gaps can be left in global governance, whether institutional assets are valuable, and whether secretariat staff plays a relevant role.To conclude, the interaction between institutional theory and causes of death needs to be accounted for and further explored.
These findings qualify discussions on the crisis of liberal international order.IOs, or at least the major IOs that matter significantly to IR, do not die in large numbers.Even if the failure of the League presents an episode that always looms large over IOs, including the UN whenever the Security Council appears again in gridlock following the Russian war in Ukraine (e.g., Patrick 2022 ; Vindman 2022 ), this article highlights that the League was an outlier case subject to fairly specific circumstances.At the same time, the findings also show that some major IOs are at risk in case of changing member states constellations.The change from France and the United Kingdom as key states in the League to the Soviet Union and United States in the UN, and the restricted membership of the IRO in what was supposed to be a universal organization, caused real problems for both IOs.The ongoing power transitions will, in this respect, also likely put pressure on the demand for cooperation between states.Coupled with a perceived underperformance of existing IOs, this may lead to simultaneous supply and demand challenges for some IOs.

Table 1 .
Mechanisms of institutional stickiness and the demand for and supply of IOs

Table 3 .
Major dead IOs Missing data on inclusion variables or not coded as "dead" in