Responding to Unilateral Challenges to International Institutions

Abstracts

How do international institutions respond to unilateral challenges by its member states, such as non-compliance, blocking of reforms, renegotiation requests, or withdrawal? This paper argues that this response depends on a trade-off between the risks of not accommodating the challenge, which could disrupt cooperation gains, and the risks of accommodating, which may embolden future challengers. International institutions aim to minimize costs, accommodating challenges when cooperation losses are high, and resisting when the risk of contagion is significant. When both risks are large, they face an “accommodation dilemma” and politically charged negotiations with the challenging country. We evaluate this framework with a comparative case study of fourteen referendum-endorsed challenges to international institutions, analyzing cases that varied in cooperation gains at risk and contagion risks. The analysis shows that across a range of different issues and institutions, the framework helps us better understand why member states respond differently to such challenges and why some challenges are resolved easily whereas others become conflictual. By developing a widely applicable theoretical framework and a versatile coding scheme, the paper contributes to a better understanding of how international institutions respond to contestation and the populist and nationalist backlash against global governance.

¿De qué manera responden las instituciones internacionales ante los desafios unilaterales de los Estados miembros, tales como el incumplimiento, el bloqueo de reformas, las solicitudes de renegociación o la retirada? Este artículo argumenta que la respuesta a esta pregunta depende de un compromiso entre los riesgos de no acomodarse ante un desafío, lo que podría interrumpir los logros de la cooperación, y los riesgos de acomodarse, lo que podría envalentonar a futuros rivales. Las instituciones internacionales tienen como objetivo minimizar los costes, acomodando los desafíos cuando las pérdidas en materia de cooperación son altas, y resistiendo cuando el riesgo de contagio es significativo. Cuando ambos riesgos son grandes, las instituciones internacionales se enfrentan a un «dilema de acomodación» y a negociaciones políticamente tensas con el país desafiante. Evaluamos este marco a través de un estudio de caso comparativo de catorce desafíos a los que se enfrentaron instituciones internacionales, los cuales fueron respaldados por referéndum, y analizamos casos que variaron con respecto a los logros de cooperación en riesgo y a los riesgos de contagio. El análisis demuestra que, dentro de una amplia gama de temas e instituciones diferentes, el marco nos ayuda a comprender mejor por qué los Estados miembros responden de manera diferente a tales desafíos, y por qué algunos desafíos se resuelven fácilmente mientras que otros se vuelven conflictivos. El artículo contribuye, mediante el desarrollo de un marco ampliamente aplicable y de un esquema de codificación versátil, a una mejor comprensión de cómo las instituciones internacionales responden a la impugnación y a la reacción populista y nacionalista contra la gobernanza global.

Comment les institutions internationales répondent-elles aux défis unilatéraux d’États membres, qu'il s'agisse de non-respect, de blocage de réformes, de demandes de renégociation ou de retrait unilatéral ? Cet article affirme que la réponse est fonction d'un compromis entre les risques du fait de ne pas céder devant un tel acte, c'est-à-dire nuire aux avantages de la coopération, et les risques du fait de céder, c'est-à-dire encourager d'autres États à faire de même. Les institutions internationales visent à réduire les coûts, accéder aux demandes quand les pertes pour la coopération seraient élevées et résister quand le risque de propagation est important. Quand ces deux risques sont conséquents, elles se trouvent confrontées à un « dilemme de réponse » et à des négociations lourdes de sens sur le plan politique avec le pays contestataire. Nous évaluons ce cadre à l'aide d'une étude de cas comparative de quatorze contestations des institutions internationales s'appuyant sur un référendum, en analysant les cas différents en termes de dangers pour les avantages coopératifs et de risques de propagation. L'analyse montre que sur un éventail de problématiques et d'institutions différentes, le cadre nous permet de mieux comprendre les différentes réponses des États membres à de tels actes, et pourquoi certaines contestations se résolvent facilement tandis que d'autres engendrent une situation conflictuelle. En développant un cadre à l'application large et un schéma de codage polyvalent, l'article contribue à l'amélioration de notre compréhension de la réponse des institutions internationales à la contestation ainsi que le retour de bâton populiste et nationaliste pour la gouvernance mondiale.

Introduction

International institutions regularly face challenges by individual member states, who contest their international legal commitments. States fail to comply with commonly decided rules (Simmons 2010; Börzel 2022), they resist reform (Hanrieder 2015; Eilstrup-Sangiovanni and Verdier 2024), they try to renegotiate membership terms (Lipscy 2017; Haftel and Thompson 2018), and sometimes even withdraw from international institutions (Helfer 2005; von Borzyskowski and Vabulas 2019, 2025; Cooley and Nexon 2020).

International institutions and their member states respond to such unilateral challenges in very different ways (Debre and Dijkstra 2021; Hirschmann 2021; Heinkelmann-Wild and Jankauskas 2022). Sometimes, they compromise and accommodate the challenging state’s demands. For example, when pressured to renegotiate the North American Free Trade Agreement (NAFTA) by US president Donald Trump, Canada and Mexico agreed to several changes that mostly favored the United States at their expense (Flores-Macías and Sánchez-Talanquer 2019). At other times, the response is to do nothing and to largely continue with business as usual. For example, after withdrawing from the International Whaling Commission (IWC) in 2019, Japan was allowed to maintain close relations with the organization in terms of funding and scientific research, and received observer status in the IWC (Von Borzyskowski and Vabulas 2025). But sometimes, international institutions respond much harsher and more forcefully to unilateral challenges. After Venezuela’s increasingly authoritarian Maduro government withdrew from the Organization of American States (OAS) in 2019, the OAS and several of its member states made sure that the country would face diplomatic isolation and lose its ability to participate in regional diplomatic forums. And the European Union (EU) has reacted harshly to challenges from countries as diverse as Austria, Greece, Hungary, Poland, Switzerland, and the United Kingdom.

This paper proposes a framework for understanding how and why responses to unilateral challenges to international institutions vary. It argues that the other states of a challenged international organization (IO) or international treaty face a trade-off between the consequences of not accommodating and the consequences of accommodating the challenge. Not accommodating the challenging state can be very disruptive, thus potentially putting considerable cooperation gains at risk. Accommodation, on the other hand, carries the risk that it may encourage similar challenges in the future when the challenging country’s experience turns out to be an attractive example to follow.

Faced with the question of whether or not to accommodate the challenger, international institutions choose the response that minimizes expected costs. The choice between a more or less accommodative strategy is straightforward when one factor dominates—challengers will be accommodated when cooperation gains at risk are much higher than contagion risks and vice versa. However, when both responses are costly, international institutions face an “accommodation dilemma” (Walter 2020, 2021a; Jurado, Léon, and Walter 2022). These situations tend to result in politically charged negotiations with the challenging country and attempts by the institution to convince the challenger to back down.

To illustrate the merits of this argument, this paper conducts a comparative case study of fourteen successful unilateral challenges to international institutions that were endorsed by a referendum vote to reject closer or continued cooperation with the respective institution. Referendums are a useful context to examine our framework, because they allow us to identify a set of comparable cases that identify a clear moment in which a challenge to an international institution became public and visible, forcing it to respond. At the same time, these referendums occurred in different countries, covered different issues—from debt repayment and migration policy to IO membership—and targeted different international institutions. Not surprisingly, these referendum-endorsed challenges vary considerably, both with regard to the cooperation gains at risk and the contagion potential, as well as the extent to which each challenge was accommodated or not. They thus provide a good context for testing our theoretical argument about the role of the accommodation dilemma in shaping responses to unilateral challenges to international institutions. The analysis shows that as expected, responses varied considerably across this diverse set of referendum-endorsed challenges in ways that are systematically related to the relative size of cooperation gains at risk and contagion risks.

Our paper makes several contributions. First, it contributes to the emerging literature on how international institutions respond to contestation and the backlash against globalization more generally. Most of this research focuses either on one or a few individual IOs (Lipscy 2017; Hirschmann 2021; Laffan and Telle 2023; Dijkstra et al. 2025) or on varied responses to the same challenger (Dijkstra et al. 2022; Heinkelmann-Wild and Jankauskas 2022). In contrast, building on earlier work that has explored the accommodation dilemma on the individual level (Walter 2020, 2021a; Jurado, Léon, and Walter 2022), this paper develops a theoretical argument about the role of the accommodation dilemma on the international level and provides both a detailed conceptualization and a versatile coding scheme that allow for a comparative analysis. It thus paves the way for a more comparative approach to studying IO's responses to unilateral challenges.

Second, our analysis underscores the difficulties of sustaining international cooperation in an environment where nationalist challenges to international institutions are growing. By exploring the conditions under which strong responses are possible and likely, we contribute to the growing literature on the life cycle of IOs (Gray 2024) and the large literature about the contestation of IOs and global governance more generally (Morse and Keohane 2014; Zürn 2018; Sommerer et al. 2022; Tallberg, Walter, and Zürn 2024; Von Borzyskowski and Vabulas 2025). Our analysis shows that the dynamics observed around high-profile cases such as Brexit are not unusual, but are likely to generalize more broadly.

Finally, our empirical analysis of responses to referendum-endorsed challenges speaks to the vast work on the populist and nationalist backlash against global governance (Copelovitch and Pevehouse 2019; De Vries, Hobolt, and Walter 2021; Dellmuth et al. 2022; Hutter, Grande, and Kriesi 2016; Trubowitz and Burgoon 2023). Popular demands to slow down, change, or even reverse institutionalized international cooperation have not just been at the center of several referendum campaigns (Schneider and Weitsman 1996; Hobolt 2009; Mendez, Mendez, and Triga 2014; Qvortrup 2016; Rose 2019), but confront IOs more generally. Voter-endorsed challenges do not just confront the EU, where Euroskepticism has become widespread (Hobolt and de Vries 2016; Hodson and Maher 2018; Rooduijn and Kessel 2019), but occur across the world. By electing parties and politicians who openly question, defy, and challenge international institutions—from Donald Trump in the United States to Rodrigo Duterte in the Philippines—voters increasingly exert pressure on international institutions. This paper contributes to our understanding of how these institutions respond to such pressure.

Responding to Challenges to International Institutions

Member states of an international institution sometimes decide to unilaterally challenge that institution, for example, by openly breaching the institution’s rules, by unilaterally blocking reform, by seeking to renegotiate membership terms, or by withdrawing from the institution altogether. The reasons behind such challenges are diverse and range from dissatisfaction with member states’ position in the institution (McArthur and Werker 2016; Stephen and Zürn 2019), criticism of substantive policies (Lipscy 2017) and an IO’s authority (Zürn 2018), a perceived decline in IO legitimacy more generally (Sommerer et al. 2022), to pressure from domestic voters and elites (De Vries, Hobolt, and Walter 2021; Trubowitz and Burgoon 2023).

Unilateral challenges therefore typically have consequences not just for the challenged institution, but also for its member states (Oye 1986). They put the size and distribution of cooperation gains in question and also risk incentivizing other states to follow the challenging states’ example. The extent and nature of these consequences hinge not just on the challenger’s actions, however, but also on the response of the challenged institution and its other member states, which play an important role in shaping the outcome and impact of the challenge.¹ While a single country can choose to confront an international institution, the institution’s reaction strongly influences what happens next. When confronted with a challenge, the institution and its other member states must decide on how to respond. This response strongly affects the direction and the consequences of the resulting process, including whether the challenging country continues or drops its confrontation, how the challenge affects the institution in the end, and whether the challenging country is ultimately better or worse off as a result.

Response options can be conceptualized along a continuum from accommodation to non-accommodation.²  Accommodation involves meeting the demands of the challenging state with minimal or no penalty, allowing it to continue to benefit from the international institution while addressing its main concerns. For instance, an IO bureaucracy and/or the other member states may opt to overlook non-compliance with certain rules, to make concessions without expecting much in return, or to minimize the potential losses associated with non-cooperation. This approach minimizes the costs for the challenging state, effectively rewarding its behavior. In contrast, non-accommodation involves refusing to meet the challenging state’s demands and imposing consequences, such as withholding cooperation benefits, refusing concessions, or applying punitive measures like sanctions or fines. This approach aims to make non-cooperation costly, signaling that the privileges of membership in the institution are conditional on adherence to its collective rules.

Weighing the Consequences: Cooperation Gains and Contagion Risks

Both of these response strategies can have considerable consequences for the challenged institution and its member states. Therefore, they need to carefully evaluate and weigh the consequences of different strategies when deciding how to respond. Two sets of consequences stand out: the short- to medium-term effects of the response on cooperation gains and the likelihood that the challenge will reverberate beyond the current challenge, either as an encouragement or as a deterrent. The likelihood with which each of these consequences is likely to materialize and their likely impact depends both on the characteristics of the specific situation at hand and on the institution’s response.

Loss of cooperation gains. The first type of consequence relates to the (opportunity) costs of not entering new cooperative arrangements or of reducing or fully dismantling existing cooperative relations. International cooperation exists because all parties benefit, even if the benefits are not always equally shared (Keohane 1984; Abbott and Snidal 1998; Gruber 2000). Countries typically challenge international institutions because they believe that this challenge will leave them better off than under the status quo. While such challenges can sometimes result in reforms that are beneficial for the institution, this means that such challenges also often leave the other member states worse off, especially when the challenge seeks to redistribute cooperation gains in the challenging country’s favor. If one member tries to change the status quo in its favor and this attempt is accommodated, the other members typically lose some of these advantages (Daßler, Heinkelmann-Wild, and Kruck 2024).

At the same time, non-accommodation can also lead to significant costs when the institution’s harsh response stifles the development of the institution, leads to a reduction of existing cooperation levels, or even risks the death of the challenged institution (Debre and Dijkstra 2021; Heinkelmann-Wild and Jankauskas 2022). Risks to cooperation gains can arise as opportunity costs, for example, when failure to ratify an agreement makes it impossible for the other member states to realize cooperation gains foreseen under that agreement. Costs can also come in the form of actual costs, especially when the challenge puts existing cooperation gains at risk. These costs can be relatively minor, such as small fines for non-compliance, but they can also become substantial, especially if the challenging state does not back down or compromise quickly. For example, when the challenging state uses this opportunity to deregulate, its businesses may enjoy a competitive advantage over their competitors in the other member states. If trade barriers are reintroduced, exporters will be hurt, and international supply chains will be disrupted. Other forms of cooperation and policy coordination—from police cooperation to environmental protection—between the other countries and the challenging country can also suffer. Whether opportunity or actual costs, adjusting to the challenge will also often increase transaction costs as economic agents adjust to the new environment.³ Because IOs often benefit from economies of scale, costs can increase disproportionately when member states, especially large and powerful ones, withdraw or are excluded from the IO (Daßler et al. 2024; von Borzyskowski and Vabulas 2024). A serious breakdown of existing cooperation can be particularly costly, as all involved parties lose out on the benefits of cooperation that they had so far mutually enjoyed.

The extent to which a challenge puts cooperation gains at risk depends both on the characteristics of the specific situation at hand and on the institution’s response. A non-accommodation strategy has the potential to put most cooperation gains at risk. This risk is particularly high when the cooperative gains in question are large, broad-ranging, and when the potential worst-case losses accrue not on a one-off basis but rather constitute long-term or repeated losses. Sizeable and broad cooperation gains are likely to affect member states in a much more profound way than smaller and narrower cooperation gains, which can be offset more easily. As a result, when debating whether to take a firm response and not to accommodate the challenge, the institution and its member states will evaluate how many resources they stand to lose from a non-accommodating response, how big the potential impact is on individual member states, how powerful or important the challenging state is, and how many issue areas are affected by the challenge. Moreover, cooperation gains that do not just accrue in a one-off manner but instead have long-term consequences increase the level of cooperation gains at risk. In such contexts, not accommodating a referendum-endorsed challenge will be very costly for everyone involved.

Political contagion risk. A second type of consequence is possible political contagion effects. They refer to the risk that other member states may be encouraged to launch similar challenges against the international institution in the future. Contagion risks emerge because observing another country’s challenge unfold allows other governments, political elites, and voters insights into what the economic, social, and political consequences would be if they launched a similar challenge. Such expectations play an important role in whether actors in other countries endorse or oppose similar challenges: Those who expect an accommodating response are much more likely to support a challenge than those who expect a non-accommodating response (Tingley and Walter 2011; Walter et al. 2018). This can generate encouragement effects and can evolve into a serious problem for international institutions when challenges proliferate. Such negative dynamics create risks and ultimate weakening of the institution (Downs, Rocke, and Barsoom 1996; Hobolt 2016). Negative contagion dynamics have occurred in a variety of settings, from the export finance regime (Bunte, Gertz, and Zeitz 2022), over the EU’s Stability and Growth Pact (Baerg and Hallerberg 2016), to the nuclear non-proliferation regime (Eilstrup-Sangiovanni 2023). Likewise, the exit of a leading member state significantly increases the chances that other exits will occur in the following year (von Borzyskowski and Vabulas 2019).⁴

Political contagion risks are likely to be larger in contexts in which an accommodative solution makes the challenging state clearly better off than the status quo, where the potential for politicization is high in other states, and where the situation easily travels to other contexts and/or countries. When challenging an institution clearly improves the situation for the challengers, actors in other states learn that it may be both feasible and desirable to engage in a similar challenge (Bamert, Gilardi, and Wasserfallen 2015; Adler-Nissen, Galpin, and Rosamond 2017; De Vries 2018; Walter 2021b). Such encouragements effects are likely to particularly strong when a weak state manages to secure significant concessions, as this signals to most remaining states that it will be easy to get similar concessions, resulting in a strong encouragement effect. In contrast, a failed challenge is likely to make voters and political elites abroad more pessimistic about the viability of such challenges, deterring similar demands abroad (Hobolt et al. 2021; Walter 2021b; Malet 2022; Martini and Walter 2023; Hunter 2024; Malet and Walter 2024). Such deterrence effects will be strongest when a powerful state fails to gain concessions. Contagion dynamics are also more likely in contexts where the potential for politicization is high, especially when influential political actors in the other member states are likely to use the case as an example to push for similar challenges in the other member states. Finally, contagion effects are stronger the less unique the grievances are that the challenge addresses and the easier it is to apply a similar narrative to other contexts and/or countries. In contrast, political contagion risks are less likely to emerge when the case at hand is very unique and does not travel easily.

The Decision Calculus of How to Respond to Unilateral Challenges

Because both ideal-typical response strategies—accommodation and non-accommodation—can have considerable adverse consequences for the challenged institution and its member states, they need to carefully evaluate each option when deciding how to respond to unilateral challenges.

Non-accommodation has the advantage that it reduces contagion risks, but the disadvantage that it raises the risks that cooperation gains may be lost (Keohane 1984). This strategy tends to result in an unattractive outcome for the challenger country. This not only increases the likelihood that it will retract its challenge—the best outcome for the challenged institution—but is also likely to lower contagion risks because it makes the outcome less attractive for the challenging state. Non-accommodation thus demonstrates to cooperation-skeptic voters and political entrepreneurs in other states that such challenges do not necessarily produce desirable outcomes. However, non-accommodation also comes with a significant downside: It can be very disruptive and thus potentially very costly in terms of cooperation gains.

In contrast, an accommodative strategy can preserve many cooperation gains. By accommodating the challenge, both the challenger and the institutions’ other member states can avoid major disruptions. But accommodation also comes with two downsides for the other member states: First, because the distribution of cooperation gains will often change in favor of the challenging country, it tends to leave them worse off than under the status quo. Second, it increases the likelihood that the episode will become an attractive precedent that is likely to encourage support for similar challenges in other member states. Accommodation thus can produce considerable political contagion risks.

The institution’s response to a unilateral challenge is crucial, as it heavily influences the ultimate consequences associated with the challenge. When choosing how to respond, international institutions and their member states must therefore weigh the risks of political contagion against the potential loss of cooperation benefits (Keohane 1984; Walter 2021a; Jurado, Léon, and Walter 2022).⁵ We argue that the relative size of the consequences influences whether the institution will respond in a more or less accommodating fashion.⁶  Table 1 summarizes this argument.

Table 1.

Predicted responses to unilateral cooperation challenges

		Cooperation gains at risk
		Low	High
Contagion risks	High	Non-accommodation	Accommodation dilemma
	Low	Passive/neutral response	Accommodation

In situations in which the potential loss of cooperation gains clearly dominates contagion risks (lower right-hand corner), accommodation is the most likely response. In contrast, when contagion risks dominate the loss of cooperation gains (upper left-hand corner), there are strong incentives to pursue a non-accommodation strategy.⁷ High contagion risks in any other member state thus increase support for a hawkish negotiation strategy, and this effect will be particularly pronounced when the challenge against the institution is highly politicized. At the same time, such pressure will make it easier for the institution and the other member states to coordinate and to develop an efficient response (Hix 2018; Drezner 2000; Laffan and Telle 2023). When neither strategy is likely to carry grave consequences (lower left-hand corner), the choice is much less consequential. In these instances, we expect the institution to react rather passively, neither actively accommodating the challenging state nor expending many resources to stand firm.

The choice is much more difficult when serious concerns about contagion exist, but at the same time, significant cooperation gains are at risk (upper right-hand corner). In these situations, the challenged institution and its other member states face an accommodation dilemma (Walter 2020, 2021a; Jurado, Léon, and Walter 2022). This dilemma is based on the trade-off between minimizing cooperation losses and minimizing contagion risks and presents a real challenge for policymakers (Oye 1986). In situations characterized by an accommodation dilemma, all responses come with considerable downsides: Whereas accommodation minimizes the disruption, it makes the challenging country better off by allowing it to enjoy the benefits of cooperation without sharing the costs. This not only leads to a redistribution of cooperation gains but is also likely to spark political contagion, which may threaten the long-term stability of the international institution. This speaks in favor of non-accommodation. However, this is a strategy that—unless the challenging state aborts its cooperation challenge—creates high costs not just for the challenging state but also for the remaining member states. This creates incentives to pursue a non-accommodation strategy with the goal of persuading the challenging country to withdraw its cooperation challenge and thus reducing contagion risks, but also some incentives to compromise if this strategy fails. In these instances, we can expect institutions to offer to maintain cooperative relations in some areas where the potential loss of cooperation benefits would be particularly painful, but to overall maintain an uncompromising stance so as not to encourage further challenges by other states that may ultimately weaken the institution in the long run. In any case, given the high stakes, these therefore tend to lead to politically charged negotiations with the challenging state.

A Comparative Case Study of Referendum-Endorsed Challenges: Research Design

To empirically examine the usefulness of this framework, we conduct a comparative case study of responses to fourteen referendums that successfully endorsed a challenge to an international treaty or organization. This research design is well-suited for testing our framework because it allows us to carefully assess the issues at stake in each referendum and the targeted institution’s response for a similar type of challenge—a public, high-stakes challenge endorsed by a referendum—in a variety of different contexts.

Although such referendum-endorsed challenges do not happen frequently, they constitute a particularly consequential form of the popular backlash against international institutions (De Vries, Hobolt, and Walter 2021). They also constitute a useful context to study our argument because referendums about international cooperation have become more frequent over the years (Hobolt 2009; Mendez, Mendez, and Triga 2014; Mendez and Germann 2016). Although voters have expressed support for international institutions in about two-thirds of these referendums, about one-third of them have directly challenged international institutions (De Vries, Hobolt, and Walter 2021). This has been the case both for referendum outcomes that veto the creation or deepening of such institutions and those that question existing institutions. These referendum-endorsed challenges compel the government to openly contest the targeted international institution (sometimes against its own preferences) and thus generate high-profile challenges, which become salient and visible at a clearly defined moment and which the targeted institution cannot ignore.⁸

As a result, referendum-endorsed challenges constitute a set of comparable cases in which we can observe variation in the extent to which each challenge was accommodated or not. The responses of international institutions to such challenges vary widely (Schimmelfennig 2019). Sometimes, successful votes are accommodated and thus result in a change of the international arrangement along the lines demanded in the referendum. Sometimes they result in a significant rupture between the challenging state and the international institution, and sometimes they are retracted when the challenging state aborts its plans to renegotiate, does not comply with, or withdraws from the institution. Moreover, referendum-based challenges vary both with regard to the issues and the institutions they target, allowing us to increase the external validity of our analysis. Finally, referendums increase the visibility, politicization, and legitimacy of challenges directed against international institutions and therefore have a particularly high potential for political contagion. Referendum-endorsed challenges hence provide a good context for testing our theoretical argument about the role of the accommodation dilemma in shaping responses to unilateral challenges to international institutions.

Finally, the research design and insights generated by this setup are likely to travel to other high-profile, voter-endorsed challenges, including those from campaign promises rather than referendums (e.g., Trump’s pledges on NAFTA or the Paris Agreement). Our study thus provides insights into the dynamics associated with the broader populist backlash against international institutions globally (Copelovitch and Pevehouse 2019).

Case Selection

Referendums on issues related to international institutions are not rare. Between 1972 and 2019, eighty-one such referendums were held on issues related to accession to, reform of, compliance with, and exit from various international institutions (Center for Direct Democracy Aarau 2019). Many of these referendums have been held on issues related to European integration, but referendums on international institutions have also been held in countries such as Costa Rica, Taiwan, or Georgia and on institutions ranging from the UN and NATO to various trade agreements.

In about a third of all referendums related to international institutions (twenty-five referendums), voters decided against international cooperation (De Vries, Hobolt, and Walter 2021). Because of feasibility concerns, we selected fourteen of these referendum-endorsed challenges for our analysis. These referendums share a common feature—each one challenges an international institution—but differ in the dependent variable, the response of the respective international institution. Because referendums aimed at renegotiating the terms of or even withdrawing from existing international institutions are the most consequential, we include all of these referendums in our analysis. In addition, we include five cases in which a proposal to deepen cooperation was rejected at the polls and three cases in which a proposal to join an international institution was voted down in a referendum.⁹

The referendum-endorsed challenges included in our study cover a large variety of different issues, most notably membership in an IO or agreement (Greenland 1982, Switzerland 1986, Switzerland 1992, Norway 1992, UK 2016), a deepening of existing cooperation (Denmark 1992, Ireland 2001, France 2005, Ireland 2008), debt repayment and austerity (Iceland 2010 and 2011, Greece 2015), foreign policy cooperation (Netherlands 2016), and migration (Switzerland 2015). Table 2 provides an overview of all fourteen referendums covered by our analysis, including some short information on the topic and the context under which they took place.

Table 2.

Fourteen referendum-endorsed challenges to international institutions

	Targeted institution	Referendum	Outcome
1982 Greenland	European Community	Leaving the European CommunitiesGreenland, initially incorporated into the EC as a Danish territory in 1972 against its will, gained self-rule in 1979 and voted in 1982 to withdraw.	EC Withdrawal in 1985.The “Greenland Treaty” ensured Greenland’s market access and aid, while the EC retained fishing rights.
1986 Switzerland	United Nations	United Nations membershipNeutral Switzerland was not a UN member, despite being host to many UN institutions and participating in many initiatives. In 1986, 75.7 percent of Swiss voters rejected a proposal for full UN membership.	Status quo maintainedSwitzerland continued to contribute to UN agencies and engage in international diplomacy without full membership.
1992 Denmark	European Community (Maastricht Treaty)	The Maastricht TreatyThe Maastricht Treaty sought to establish the European Union, but 50.7 percent of the Danish voters rejected it, citing concerns over sovereignty, particularly regarding currency and defense.	Negotiation of opt-outsEU leaders negotiated opt-outs for Denmark’s key concerns, leading to voter approval in a second referendum on May 18, 1993.
1992 Switzerland	European Economic Area	European Economic Area AgreementThe EEA agreement would have allowed Switzerland to participate fully in the EC’s internal market, but was narrowly rejected by 50.3 percent of Swiss voters due to concerns over sovereignty, unrestricted freedom of movement, and the EC’s monetary crisis.	Bilateral agreementsThe EEA entered into force in 1994 without Switzerland. Switzerland negotiated a broad set of bilateral, sector-specific agreements with the EU that allowed for close cooperation.
1994 Norway	European Union	European Union MembershipDespite already close ties through the EEA, 52.2 percent of Norwegians voted against EU membership in November 1994, citing concerns over sovereignty, immigration, and cultural values.	Status quo maintainedNorway remained in the EEA. Over time, it deepened its ties with the EU through new agreements, such as the Schengen Agreement.
2001 Ireland	European Union (Nice Treaty)	Treaty of NiceThe Treaty of Nice reformed EU institutions to be able to cope with enlargement and become more efficient, but 53.9 percent of Irish voters rejected it over sovereignty and neutrality concerns.	Opt-outs and guaranteesAfter securing the Seville Declaration protecting Irish neutrality, 60 percent of voters accepted the treaty in a second referendum.
2005 France	European Union (EU Constitution)	European Union ConstitutionThe EU Constitution aimed to consolidate EU treaties and enhance integration. Ratification was derailed when 55% of French voters (and days later also Dutch voters) rejected it in a referendum, citing concerns over jobs, public services, and Turkey’s potential EU accession.	Treaty abandonedThe Constitution was abandoned and replaced with the Lisbon Treaty, which retained much content from the Constitution but dropped aspects suggesting a shift towards a federal Europe.
2008 Ireland	European Union (Lisbon Treaty)	Lisbon Treaty53.4 percent of Irish voters rejected the Lisbon Treaty, meant to streamline and enhance EU decision-making, citing concerns over neutrality and less control over tax and social laws.	Opt-outs and guaranteesAfter securing binding guarantees from the EU, 67.1% of Irish voters approved the treaty in a second referendum.
2010 Iceland	IC-UK-NL Loan Agreement I	First loan guarantees referendum (Icesave I)The collapse of Iceland’s Landsbanki left foreign depositors with major losses. The United Kingdom and Netherlands initially compensated their citizens and then sought repayment from Iceland. However, 93 percent of Icelandic voters rejected the negotiated repayment agreement.	Renegotiated agreementThe Icelandic government and the United Kingdom and NL renegotiated the repayment agreement with better terms for Iceland
2011 Iceland	IC-UK-NL Loan Agreement II	Second loan guarantees referendum (Icesave II)Although the renegotiated loan repayment agreement offered better terms, 59.8 percent of voters rejected the new agreement again in a second referendum.	Legal resolutionThe United Kingdom and the Netherlands brought the case before the EFTA court, which ruled in Iceland’s favor in 2013.
2014 Switzerland	European Union (Bilateral Treaty)	Initiative “Against Mass Immigration”The popular initiative aimed to limit immigration by reintroducing quotas for foreign workers and hiring priority for Swiss citizens was narrowly accepted with 50.2 percent of the vote. It conflicted with the Swiss-EU “Free Movement of Persons” Treaty, putting Switzerland’s custom-made relations with the EU at risk.	Non-implementationSwitzerland introduced quotas and Swiss worker priority only for non-EU nationals, preserving the status quo for EU/EFTA nationals, in line with the CH-EU Treaty.
2015 Greece	European Monetary Union	Greek Bailout referendumAfter years of economic crisis, Greek voters rejected a renewed bailout agreement from the “troika” (EC, ECB, IMF), primarily opposing the austerity measures tied to financial aid. The vote put Greece’s Eurozone membership and the Eurozone’s crisis management at risk.	Non-implementationFacing a choice between the bailout package or leaving the euro, Greece accepted the bailout package.
2016 Netherlands	European Union (Association Agreement)	Ukraine–EU Association AgreementThe Ukraine–EU Association Agreement established political and economic ties between Ukraine, the EU and Euratom. 64 percent of Dutch voters rejected the agreement, citing concerns over Ukraine’s readiness and deeper EU integration more generally.	Clarifications and guaranteesThe EU provided political guarantees, clarifications, and a reaffirmation of key principles, allowing the Dutch government to ratify the agreement.
2016 UK	European Union	Brexit - Leaving the EUAfter failed attempts to renegotiate the UK’s EU membership terms, 51.9 percent of UK voters opted to leave the European Union. Key issues included sovereignty, immigration control, economic impact, and EU regulations.	EU withdrawalThe negotiated Brexit deal significantly reduced cooperation between the EU and the United Kingdom.

Operationalization

We developed detailed coding schemes,¹⁰ which specify clear indicators to operationalize both our dependent and independent variables. We use documents, media coverage, and secondary literature to assess each case and provide a detailed analysis of each of the fourteen referendum-endorsed challenges, including broader details about each case, references to relevant sources, and justification for each coding decision in the online appendix.

Dependent Variable: Responses to Referendum-Endorsed Challenges

To measure the dependent variable—international institutions’ responses to referendum-endorsed challenges—we classified responses into five categories: weak and strong non-accommodation, a neutral/passive response, and weak and strong accommodation. Responses were coded as strong non-accommodation when the institution and/or its member states strictly refused to alter the agreement or membership terms, with no or tightly limited concessions or flexibility. In contrast, weak non-accommodation means that although most demands are still rejected, minor and symbolic concessions are possible. Institutions can also opt to not respond and to accept the referendum results without taking any specific action or offering concessions. We classify such responses as passive or neutral responses. Finally, institutions can also accommodate challenges. We classify such responses as weak accommodation when the challenging state is granted small deviations from certain rules, timelines, or obligations, although some key demands remain unmet and the challenging state is required to make some concessions as well. Finally, when most or all of the challenging state’s demands are met without any major concessions required in return, we classify the response as strong accommodation.

Using this coding scheme, we classify five referendums as strong accommodation (Greenland 1982, Switzerland 1992, Denmark 1992, Ireland 2001, and France 2005) and two as weak accommodation (Ireland 2008 and Iceland 2010). Two referendums (Switzerland 1986 and Norway 1994) were met with a neutral/passive response, one with a weak non-accommodation response (Netherlands 2016), and in four cases (Iceland 2011, Switzerland 2014, Greece 2015, and UK 2016), the targeted institution and/or the other member states reacted with strong non-accommodation.

Independent Variables: Cooperation Gains at Risk and Contagion Risks

To operationalize the level of cooperation gains and contagion risk posed by each referendum-endorsed challenge, we developed a detailed but general coding scheme that can be used to assess the potential consequences of referendum-endorsed challenges. In line with our conceptualization of cooperation gains at risk and political contagion risks as multidimensional concepts, we develop indicators that capture each concept’s different dimensions. By carefully operationalizing each dimension of these concepts, the coding scheme captures variation in potential consequences across different cases and contexts. Table 3 presents a summary of the coding scheme (the full scheme is available in the online appendix). After applying these indicators to assess the issues at stake in each referendum and classifying each of them on a 1 (low) to 3 (high) scale, we aggregate the scores for all dimensions using unweighted averages. This process yields a comprehensive and comparable overall assessment of the potential cooperation gains at risk and the contagion risks associated with each referendum. Table 4 presents these scores; more detailed information for each of the fourteen cases is available in the online appendix.

Table 3.

Coding scheme summary: dimensions and indicators

Cooperation gains at risk

Size Size of the cooperation gains at risk for the institution and other relevant states if the referendum-endorsed request is not accommodated. a) Depth of cooperation gains at risk: How much money or other benefits does the institution and other relevant states stand to lose or gain if cooperation collapses completely? b) Impact on member states: How big is the impact on other state’s domestic economies? c) Challenger importance: How powerful or important is the challenging state in the specific area or for the specific institution? Breadth The extent or range of issues, topics, or areas covered by the cooperative relationship. How many issue areas are put at risk through non-accommodation? Time profile The temporal characteristics of potential losses incurred by the remaining member-states due to non-accommodation. Do the potential losses for the remaining member-states accrue on a one-off basis or are they long-term or repeated losses?

Political contagion risk

Attractiveness of the accommodative outcome The extent to which the challenging state is better off after an accommodated challenge compared with the status quo. How does the challenging state’s utility compare with its status quo utility if its challenge is accommodated? Potential for politicization Presence of political actors who are likely to use the case as an example to push for similar challenges in the other member states. Are there political actors among the other member states who would desire the same accommodative outcome and are likely to seize this response as an opportunity? Challenging state bargaining power The challenging states bargaining power compared to other states. How does the challenging state’s bargaining power compare to the bargaining power of the other member states? Uniqueness Ease with which the accommodative response can be applied to other contexts and/or countries How many other member states share similar grievances?

Table 4.

Coding overview—cooperation gains at risk

	Cooperation gains at risk						Contagion risks
Case/indicator	Size			Breadth	Time profile	Overall score	Attractiveness	Politicization potential	Bargaining power	Uniqueness	Overall score
	Depth	Impact	Challenger importance				relative to status quo		relative to other relevant states	applicable to…
Scoring:	1.0: Very few 1.5: Few 2.0: Some 2.5: Substantial 3.0: Critical	1.0: None 1.5: Minimal 2.0: Some 2.5: Large 3.0: Very large	1.0: Not important in any area 1.5: Not important, single area 2.0: Important in few areas 2.5: Important in several areas 3.0: Important in most areas	1.0: Single 1.5: Few 2.0: Several 2.5: Many 3.0: All	1.0: Isolated 1.5: Isolated, slight impact over time 2.5: Recurring, period vaguely defined 3.0: Recurring indefinitely	Unweighted average	1.0: Utility lower 1.5: Utility same 2.0: Utility somewhat higher 2.5: Utility higher 3.0: Utility much higher	1.0: No actors 1.5: Few actors, politically weak 2.0: Some pol. weak actors 2.5: Some pol. powerful actors 3.0: Many pol. powerful actors	1.0: More powerful than all others 1.5: More powerful than some 2.0: As powerful 2.5: Somewhat less powerful 3.0: Less powerful	1.0: No other state 1.5: Few other states 2.0: Some other states 2.5: Many other states 3.0: Most other states	Unweighted average
Greenland 1982: EC Membership	1.5	1.5	1.5	2.5	3.0	2.0	2.0	1.5	3.0	1.0	1.9
Switzerland 1986: UN membership	1.0	1.5	1.0	1.0	1.0	1.1	2.0	1.0	2.0	1.0	1.5
Denmark 1992: Maastricht Treaty	3.0	3.0	3.0	3.0	3.0	3.0	2.5	2.0	2.0	2.0	2.1
Switzerland 1992: EEA membership	2.5	2.0	2.0	2.5	2.0	2.2	2.5	2.0	2.0	1.0	1.9
Norway 1994: EU membership	1.5	1.5	1.5	2.0	3.0	1.9	2.0	2.0	2.0	1.5	1.9
Ireland 2001: Treaty of Nice	2.0	2.0	3.0	2.5	2.5	2.4	2.0	2.0	2.5	1.5	2.0
France 2005: EU constitution	3.0	3.0	3.0	3.0	3.0	3.0	2.0	2.0	1.5	2.0	1.9
Ireland 2008:Treaty of Lisbon	2.5	3.0	3.0	3.0	3.0	2.9	1.5	2.5	2.5	2.0	2.1
Iceland 2010: Icesave I	1.0	1.0	1.0	1.0	1.0	1.0	2.5	2.0	3.0	2.5	2.5
Iceland 2011: Icesave II	1.0	1.0	1.0	1.0	1.0	1.0	2.5	2.0	3.0	2.5	2.5
Switzerland 2014: Free Movement of People	2.5	2.5	2.0	2.5	3.0	2.5	3.0	3.0	2.5	3.0	2.9
Greece 2015: Bailout Referendum	3.0	3.0	2.5	2.5	3.0	2.8	3.0	3.0	2.5	2.5	2.8
Netherlands 2016: Ukraine–EU Association Agreement	1.5	1.5	1.5	1.5	3.0	1.8	2.0	2.5	2.0	2.5	2.3
UK 2016: Brexit Referendum	3.0	3.0	3.0	3.0	3.0	3.0	3.0	3.0	1.5	3.0	2.6

Cooperation gains at risk. The impact of a referendum-endorsed challenge on cooperation gains for other member states depends on three broad factors: the size, breadth, and duration of the potential losses. We operationalize size by evaluating the depth of cooperation gains at risk, the impact on other states’ economies or sectors, and the importance of the challenging state in a specific policy area or IO. The breadth refers to whether the challenge affects one issue or multiple areas of cooperation. Finally, long-term or repeated disruptions are considered more significant than one-off events in determining the overall risk to cooperation.

There is considerable variation in the extent to which the referendums we study put cooperation gains at risk (see Table 4). In some cases, they were minimal, such as in the two Icesave referendums in Iceland, which placed a one-off payment of €4.5 billion to the United Kingdom and the Netherlands at risk, or when the Swiss rejected UN membership, which did not affect Switzerland’s membership in many specialized UN agencies or its UN contributions.

In other cases, referendums put many more cooperation gains at risk. The stakes were highest in six referendums that raised fundamental questions about key European institutions. The Danish 1992 rejection of the Maastricht Treaty put the establishment of the EU at stake because the Danish veto put significant potential cooperation gains regarding economic and monetary union as well as political integration at risk. Likewise, the 2005 French and 2008 Irish referendums prevented closer cooperation and reforms, raising doubts about the EU’s ability to adapt its structures and institutions to accommodate new members and deepen integration. Ireland’s rejection of the Treaty of Lisbon threatened essential reforms needed to streamline EU decision-making and ensure effective functioning as a twenty-seven-member union. French voters’ rejection of the EU Constitution in 2005 put the EU’s entire constitutional reform process in question. The Swiss 2014 anti-immigration referendum put high cooperation gains at risk because it jeopardized the six bilateral treaties that form the foundation underpinning Switzerland’s close relations with the EU. Greek voters’ 2015 refusal to accept a new bailout package with harsh conditionality put Greece’s eurozone membership at risk, an outcome that European policymakers had long sought to avert. A “Grexit” was expected to be hugely disruptive not just for Greece but also for the Eurozone as a whole because it would demonstrate that euro membership was not irreversible, provide a huge blow to the common currency project, and likely cause huge risk premia to emerge on international financial markets for other European debtor countries. Finally, a non-accommodated “no-deal” Brexit outcome posed long-term risks that threatened to disrupt deep economic, political, and security relations, a very costly outcome not just for the United Kingdom but also for the remaining EU-27 member states.

The remaining referendums fall in between these two extremes. The 2016 Dutch rejection of the Ukraine–EU Association Agreement jeopardized the portion of the agreement concerning deep and comprehensive free trade between the EU and Ukraine at risk, but its overall impact was limited because most of the content of the agreement was not subject to national ratification and had already been provisionally implemented. Ireland’s 2001 rejection of the Treaty of Nice delayed institutional reforms needed for Eastern enlargement and posed slightly less critical challenges than the rejection of the Lisbon Treaty, as enlargement could still proceed under existing frameworks. Norway’s EU membership meant that the political, strategic, and resource-based benefits of a full Norwegian EU membership would not be realized, but the potential fallout was limited by the fact that the country remained closely integrated with the EU through the EEA and the Western European Union. The impact of Greenland’s departure from the EC was limited by its small market size, remote location, and limited integration with EC countries beyond Denmark (with the exception of the fishing industry). Switzerland’s 1992 rejection of EEA membership jeopardized close cooperation between the EC and Switzerland and disrupted plans for a unified EEA, and thus put somewhat higher cooperation gains at risk.

Contagion risks. Political contagion risks are higher when an accommodative solution is attractive to other countries and can easily spread to other cases. These risks are also assessed across four dimensions: the attractiveness of accommodation, the potential for politicization, the challenger state’s bargaining power, and the uniqueness of the case. The attractiveness is evaluated by determining if accommodation benefits the challenging state compared to the status quo. The potential for politicization is higher when more political actors, especially powerful ones, are likely to use the case to push similar challenges. A high bargaining power means weaker states are less likely to be able to secure similar accommodations. The uniqueness of the case is assessed by considering how many other states share a similar grievance, with more unique cases posing lower contagion risks, and by considering the challenging state’s bargaining power relative to the other member states.

Table 4 shows that there is also considerable variation in the level of contagion risks associated with each cooperation challenge. Reflecting the politicized nature of many referendums, our analysis shows that there is only one case in which contagion risks were rather low: Switzerland’s 1986 decision to not join the UN did not have much potential to reverberate abroad, not just because the overwhelming majority of countries were already UN members, making Switzerland’s non-membership in the UN a significant anomaly, but also because Switzerland did not make any demands for a special status with regards to the UN.

Contagion risks can be classified as moderate in seven cases. Greenland’s vote to withdraw from the EC, though politically significant, was unique: As a former Danish colony, the territory had joined against its will and held the referendum right after it gained the right to home rule. This created a specific situation unlikely to apply to other member states, limiting the potential for contagion. Swiss (1992), Danish (1992), and Norwegian (1994) voters all rejected closer integration with the EU in the early 1990s (Hobolt 2009). Renegotiating treaty and membership terms to let countries selectively adopt beneficial aspects while opting out of costly ones would likely appeal to Euroskeptic political actors in other countries. However, the EU’s popularity and limited influence of Euroskepticism in the early 1990s curbed contagion risks in these cases, as did some unique characteristics: Switzerland had already submitted a membership request to the EU, so that EEA rejection was viewed as a temporary disruption, and no other EEA country exhibited similar levels of self-sufficiency as Norway. Accommodating Irish (2001, 2008) and French (2005) concerns also had a potential for political contagion because renegotiating specific exceptions, opt-outs, and treaty changes could be seen as an attractive outcome for political entrepreneurs from other countries as well. However, Ireland’s neutrality and referendum requirements reduced contagion risks, as did France’s bargaining power, which made similar accommodations less viable for others.

Finally, contagion risks were high in six referendum-based challenges. Because the opponents to the EU–Ukraine Association Agreement rejected by the 2016 Dutch referendum had focused less on the specific issue at hand, but rather on broad concerns related to the EU, the referendum galvanized Euroskeptics across Europe and thus had a considerable potential for contagion. The Icesave referendums had the potential to turn into an attractive precedent for other countries saddled with external debt, especially at a time when taxpayer outrage about having to repay private debts ran high. Greece’s 2015 bailout rejection threatened to inspire similar actions by other crisis-stricken eurozone economies intent on receiving a bailout without harsh conditionality, giving it a very high contagion potential. As influential political actors in other EMU countries such as Italy, France, and Slovakia watched the EU’s response closely and threatened to hold their own referendums, the matter became heavily politicized in both debtor and creditor countries and threatened to undermine Eurozone crisis management. Finally, contagion risks were also high in the British (2016) and especially the Swiss (2014) referendums because both sought to limit obligations arising from membership in the targeted institution (such as commitments to free movement or budget contributions), while at the same time continuing to enjoy far-reaching benefits, particularly broad access to the EU’s Single Market and to EU programs. By shifting the costs and benefits so clearly in the challenging state’s favor, this would have created a very attractive outcome for both countries relative to the status quo. This was especially true as demands for limitations on immigration, contentious in many EU states, were central to both the Swiss and Brexit referenda. An accommodating outcome would thus have provided an appealing precedent for Euroskeptic leaders in other EU countries. Political actors across Europe, from France’s National Front to Norway’s Progress Party, closely followed these developments. Despite these similarities, Switzerland’s limited bargaining power amplified politicization potential, while the UK’s status as one of the most powerful countries in Europe made it distinct, thus reducing contagion risks to some degree.

Analysis: Why Do Responses of International Institutions to Unilateral Challenges Vary?

Our theoretical framework suggests that the extent to which international institutions and their member states face an accommodation dilemma influences how they respond to unilateral challenges. The response is expected to be particularly hawkish in cases in which the net costs of non-accommodation are small, and more dovish when non-accommodation is very costly for its other member states’ domestic economies and societies. When the expected consequences are minor, the response is likely to be muted and passive. But when institutions face a pronounced accommodation dilemma, the whole process is likely to be difficult and protracted. The institution’s willingness to accommodate will be very limited, especially initially and when there is a good chance that the challenger will back down.

Figure 1 provides an overview of the risk profiles associated with each of the fourteen referendums. It places the referendums in a two-dimensional grid that shows both the extent of cooperation gains at risk (x-axis) and the contagion risks involved in each referendum (y-axis). The figure suggests that in three cases (Iceland I and II and the 2016 Dutch referendum), contagion risks dominate the cooperation gains at risk. For these cases, our framework predicts a non-accommodating response. Two cases (Switzerland 1992 and Norway 1994) are situated in the lower-left quadrant, suggesting a neutral or passive response. A third set of cases clusters in the mid- to lower-right of the grid, suggesting that these referendums (Greenland 1982, Switzerland and Denmark 1992, Ireland 2001 and 2008, and France 2005) put more cooperation gains at risk than they carried contagion risks. For these cases, our framework suggests a more accommodating response. Finally, three referendums (Switzerland 2014, Greece 2015, and UK 2016) are clustered in the upper right-hand corner, suggesting a strong accommodation dilemma. Here we should see politically difficult processes and significant efforts aimed at encouraging the referendum countries to retract their cooperation challenge, typically by taking a decidedly non-accommodative stance. If these efforts are not successful, we expect mostly an uncompromising response designed to deter further challenges by other states, while cooperative relations are selectively maintained in a few areas where potential costs are particularly large.

Figure 1.

Risk profiles and responses to unilateral referendum-endorsed challenges.

Examining the extent to which these predictions square with actual outcomes, Figure 1 suggests that many of these expectations are borne out, even though some interesting anomalies exist. Accommodation occurred far more frequently when cooperation gains clearly outweighed contagion risks, whereas cases with few consequences were met with passivity. The politics following upon referendums that confronted the challenged institutions with a pronounced accommodation dilemma were most contentious, and ultimately resulted in strong non-accommodation. A more surprising pattern emerges for challenges with low cooperation gains at risk but a high potential for political contagion. Contrary to expectations, one of these challenges was weakly accommodated, while the remaining two exhibit varying levels of non-accommodation. To probe our argument, we briefly discuss responses in each of these categories.¹¹

Strong accommodation. In all referendum-endorsed challenges that were strongly accommodated by the targeted institutions, cooperation gains at risk outweighed contagion risks. For example, Switzerland was allowed to pursue closer ties with the EU (and by implication, EEA) after it had rejected EEA membership in 1992 in a more tailor-made manner. Because Switzerland had deposited an EU accession request in Brussels in 1992, the EU was relatively unconcerned about contagion risks and saw these alternatives as a bridge to Swiss EU membership. Additionally recognizing Switzerland’s economic importance, the EU accommodated Switzerland’s preference for a sectoral, bilateral approach. The series of sector-specific agreements, collectively known as the “Bilateral Path,” that were subsequently negotiated allowed Switzerland to access the EU’s internal market in key areas like trade, transport, and research without full EEA or EU membership.¹²

Faced with votes that threatened to derail key institutional reforms and major integration steps, EU member states also showed themselves flexible to accommodate challenges raised by referendum votes in Denmark 1992, Ireland 2008, and France 2005. Following Denmark’s 1992 rejection of the Maastricht Treaty, the other EC member states granted Denmark specific opt-outs and derogations on its main four demands regarding the single currency, defense policy, justice and home affairs, and EU citizenship. These concessions were compatible with the treaty but applied only to Denmark, allowing the EC neither to renegotiate nor modify the treaty itself, and thus to pave the way for the establishment of the EU. This accommodative stance is consistent with the fact that cooperation gains at risk clearly outweighed contagion risks. That said, in light of the fact that contagion risks were not trivial, our framework would have predicted a more controversial process as ultimately played out. The EU member states also accommodated France’s (and the Netherland’s) 2005 referendum vote against the European Constitution. The vote compelled European leaders to abandon the treaty and to drop some of the most controversial aspects of the EU Constitution altogether. They then implemented some of the core institutional reforms in a much more limited form through the Lisbon Treaty. This treaty, however vital for implementing key reforms to modernize the EU’s decision-making processes, was rejected by a further referendum, the 2008 Irish vote, which threatened to leave the Union with outdated and inefficient decision-making mechanisms. In response, the EU accommodated Irish concerns by providing legally binding guarantees, annexed to the treaty, addressing neutrality, taxation, and social policies. Crucially, the treaty was amended to ensure that every member state would retain a permanent Commissioner, reflecting a strong accommodative approach as institutional structures were adjusted to meet Ireland’s demands. These concessions enabled Ireland to hold a second referendum, securing the treaty’s ratification while addressing its specific concerns.

Finally, our analysis also shows that international institutions can show themselves willing to accommodate even major challenges, such as a unilateral withdrawal. After Greenland had voted to end its EC membership, the EC granted Greenland privileged “Overseas Country and Territories (OCT)” status by concluding the Greenland Treaty. The EC countries negotiated retained fishing access to Greenland’s waters, while giving Greenland continued access to the EEC market for fishery products and continued development aid declaring Greenland a “special case” in the treaty, the EC managed to limit the loss of cooperation gains without risking major contagion abroad.

Weak accommodation. We also observe several cases of weak accommodation in our sample, in which the challenging state was granted small exceptions but had to make some concessions as well. Ireland’s 2001 rejection of the Treaty of Nice, designed to prepare the EU for enlargement, prompted the EU to issue the Seville Declaration, an official document providing formal assurances to address Irish concerns about neutrality. While not legally binding, these guarantees went beyond a mere clarification of the treaty, emphasizing that Ireland would not be required to join military alliances or participate in EU defense commitments. With cooperation gains at risk only slightly outweighing the risk of contagion, the EU opted for a measured response, providing minimal but formal accommodation to satisfy Irish voters without establishing a strong precedent for other member states. This allowed Ireland to address domestic opposition and hold a second referendum, securing ratification without altering the treaty itself. After Iceland’s voters had rejected the agreement to repay debts incurred by the country’s collapsed Landsbanki, the United Kingdom and the Netherlands made some concessions regarding interest rates and repayment period. Given the high potential for contagion, and the low cooperation gains at stake, this is at odds with our argument, but may be linked to the fact that British and Dutch claims against the Icelandic government were legally not rock solid, as the subsequent EFTA Court ruling in favor of Iceland revealed.

Passive/neutral response. Faced with rejections of UN (Switzerland) and EU (Norway) membership, the targeted institutions reacted passively and did not change their relations with both countries after the vote. For example, Switzerland continued as an observer state at the UN that closely worked with UN specialized agencies and made substantial financial contributions. Likewise, Norway maintained close relations with the EU as an EEA member. As both cooperation gains at risk and political contagion risks were small in these cases, this response is in line with our framework’s expectation.

Weak non-accommodation. In 2016, Dutch Euroskeptics had launched a referendum against the EU–Ukraine Association Agreement. Because the campaign was focused less on the agreement itself and more on criticism of the EU as such, contagion risks were considerable. As a result, the other EU member states were careful not to accommodate the challenge too much. Rather, they decided that the European Council would provide interpretative clarifications of the agreement. Unlike the legally binding guarantees given to other countries (such as Ireland), these assurances were political clarifications that did not alter the agreement’s substance. Because this approach meant that demands to change the agreement were rejected, while making symbolic concessions, we classify this response as weak non-accommodation, which is in line with expectations for cases in which contagion risks surpass cooperation gains at risk.

Strong non-accommodation. In four instances, the challenged institutions refused to make any major concessions or to alter the agreement or membership terms in question. After Icelandic voters rejected the renegotiated repayment agreement in the Icesave II referendum and, as expected for a case where contagion risks outweigh cooperation gains, the United Kingdom and the Netherlands showed themselves unwilling to compromise further. Instead of accommodating Iceland, they sought legal dispute resolution and brought the issue before the EFTA Court, which, however, eventually freed Iceland from the obligation to repay the deposit guarantees.

The other three cases of strong non-accommodation all occurred in cases that confronted the challenged institution with a strong accommodation dilemma: Switzerland 2014, Greece 2015, and UK 2016. Given the high cooperation gains at stake and the very high potential for political contagion, the best outcome in all three cases would have been for the challenging state to withdraw its challenge. While the EU’s and EMU’s tough, non-accommodating stance succeeded in securing this goal with respect to the Swiss and Greek challenge, however, it was less successful in the Brexit case.

Following Switzerland’s 2014 anti-immigration referendum, which conflicted with the principle of free movement of people, the EU refused to negotiate a compromise and instead presented Switzerland with a stark choice: either adhere to existing agreements or face significantly reduced cooperation. This would have been possible because the so-called guillotine clause gave the EU the power to respond to a violation of the free movement of people treaty with a termination of all bilateral agreements. Switzerland initially resisted, even violating its treaty commitments by declining to extend freedom of movement to Croatian nationals (a new EU member). But the EU’s non-accommodating response, during which it even barred Switzerland’s access to the new Horizon 2020 research program and the ERASMUS program, ultimately convinced Swiss policymakers to not implement the referendum vote with respect to EU citizens and to maintain the bilateral treaties instead. Similarly, the EMU member states took a hard line in response to Greece’s 2015 referendum, with which the country had sought to achieve less painful conditions for a new bailout. Rather than engage in serious negotiations, EMU member states threatened that Greece would have to leave the Eurozone if it did not back down. Confronted with huge economic and political risks associated with a “Grexit,” the Greek government quickly accepted the slightly modified bailout agreement, effectively disregarding the referendum vote.

The Brexit referendum in 2016 posed an even more complex challenge to the EU, which generated an extended negotiation process full of political drama. Throughout the Brexit negotiations, the EU made it clear that it was not going to accommodate the British Have your cake and eat it approach that Brexiteers had championed. The negotiations were protracted and much more difficult than observers had originally expected. One major concern, mentioned repeatedly by the EU-27 side, was possible contagion risks that would put the workings of the Single Market at risk in the long run. As a result, both the talks about the terms of withdrawal and about the future relationship were close to failure several times. But even in these moments, when the risks of a “No Deal Brexit” and the associated losses in cooperation gains became substantial, the EU-27 did not make major concessions to the United Kingdom. In line with the accommodation dilemma argument, the EU did show some flexibility regarding Northern Ireland, where the potential loss of cooperation posed the most serious risks, including renewed civil conflict. However, in other areas, the EU prioritized preserving its institutional integrity and deterring contagion risks over maintaining close ties with the United Kingdom.

Conclusion

This paper has explored how international institutions respond when individual member states challenge the institution by blocking reforms, not complying with their legal commitments, seeking to change membership terms in their own favor, or withdrawing from the institution. It has proposed a framework for understanding these responses, suggesting that institutions face a trade-off between not accommodating the challenging state, which risks disrupting cooperation gains, and accommodating the challenge, which could set a dangerous precedent for future challenges. A comparative case study of fourteen rather different referendum-endorsed challenges to international institutions showed that the relative size of potential loss of existing cooperation gains and contagion risks influenced responses to these challenges. This illustrates that the accommodation dilemma framework provides a parsimonious but powerful framework to understand the responses to such challenges and the negotiation dynamics surrounding these processes.

While this paper has focused on challenges that have been endorsed by voters, the argument and our empirical approach can be applied to a much wider set of instances in which countries unilaterally challenge international institutions. From notifications to withdraw from the International Criminal Court, over the difficulties of extending the new START arms control treaty beyond its expiry date, to renegotiations about IMF quotas in favor of rising powers such as India and China, international institutions have faced challenges in which countries unilaterally seek to change the costs and benefits of international cooperation in their favor. Whether these challenges are directly endorsed by voters or not, they confront the affected international institutions and their member states with the question of how to respond. To the extent that accommodation creates contagion risks—either by encouraging other member states to pursue a similar path or by encouraging the challenging state to repeatedly seek renegotiations in its own favor, the other member states may also confront accommodation dilemmas when weighing the benefits of accommodation against these potential contagion effects. At the same time, we show that countries and international institutions will be more willing to accommodate challenges and defections from jointly agreed rules when resistance is likely to put large cooperation gains at risk. This framework will therefore also be useful in studying the response of the world community under the more assertive Trump 2.0 presidency, in which the world’s hegemon non-cooperative approach puts major cooperation gains, from NATO’s security guarantees to the rules-based multilateral trade order, at risk. The accommodation dilemma framework suggests that states are ultimately going to be willing to accommodate many of these challenges.

In sum, the paper underscores the dilemmas confronted by countries faced with unilateral cooperation challenges and the difficulties associated with maintaining good, cooperative relations in a context where nationalist challenges to international institutions proliferate.

Author Biography

Stefanie Walter is a Full Professor of International Relations and Political Economy at the Department of Political Science at the University of Zurich. Her research examines the backlash against globalization and challenges to international cooperation.

Nicole Plotke-Scherly is a PhD student in International Relations at the Department of Political Science, University of Zurich. Her research focuses on the challenges to international organizations and the rise of the Global South.

Notes

Authors’ note: We would like to thank Tanja Börzel, Loriana Crasnic, Fabrizio Gilardi, Gorana Grgic, Belén Gonzalez, Sara Hobolt, Tuuli-Anna Huikuri, Tom Hunter, Erik Jones, Tobias Lenz, Giorgio Malet, Marco Martini, Sean Müller, Sujeong Shim, Jon Slapin, Michael Strebel, Thomas Sommerer, Jonas Tallberg, Dustin Tingley, and Denise Traber, as well as participants in workshops at the LAGAPE seminar, the UCL European Politics series, the Europe@LSE seminar, the IR colloquium at the University of Wisconsin-Madison, the IPZ publication seminar, and APSA and DVPW 2021. This project has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme grant agreement no. 817582 (ERC Consolidator Grant DISINTEGRATION).

Footnotes

1

To improve readability, we will refer to the "institution’s response" in the discussion below, referring to the coordinated response between the institutions’ bureaucracy and its member states.

2

Others distinguish between conciliatory and adversarial responses (Heinkelmann-Wild and Jankauskas 2022).

3

The extent of these costs will often vary across member states; reduced cooperation can sometimes also create upsides for some member states.

4

To the extent that contagion risks can significantly weaken existing institutions, they can thus also be interpreted as a long-term threat to cooperation gains.

5

The politics of economic sanctions follows a similar rationale; see, e.g., Hufbauer (1990).

6

In addition, traditional determinants of negotiation strategies, such as the distribution of bargaining power, will also influence how good a deal a withdrawing country can expect from the remaining member states (Bailer 2010; Lipscy 2017).

7

The extent to which governments are concerned about potential encouragement effects also depends on their own ideological leanings. More cooperation-skeptic governments tend to be less concerned about contagion risks and therefore more likely to accommodate, not least because they seek to set a precedent for their own potential future challenges (Jurado, Léon, and Walter 2022).

8

Of course, such referendums are often preceded by more low-profile challenges and high levels of contestation within the referendum country. We begin the analysis at a point when the challenge becomes particularly consequential for the targeted institution.

9

In selecting cases, we prioritized cases that maximized variation in terms of countries and issues and referendums whose legitimacy was not contested (as in Brazil 2000, Taiwan 2008, and Hungary 2016). We also excluded cases that did not provide enough material to study (for example, the referendums in Antigua & Barbuda and Grenada that rejected the Caribbean Court of Justice as the final court of appeal). Note that this case selection has the drawback that all analyzed cases are from Europe, although referendums also occur in other parts of the world.

10

See online appendix, Tables A3 and A4.

11

Details on and a bibliography for all cases can be found in the online appendix.

12

In contrast to the EU’s expectations, however, Switzerland did not ultimately become an EU member. Rather, Swiss Euroskeptics felt vindicated by the EU’s response and have since been strengthened. Because Switzerland greatly benefited from the EU’s accommodative approach (despite unexpected problems, especially with regard to a huge influx of EU migrants), the Swiss model has also become a poster child for Euroskeptics across Europe. This development suggests that the EU may have underestimated the risks associated with accommodating Switzerland at the time.