1 Introduction

International organizations promoting the Liberal International Order (LIO)Footnote 1 are increasingly contested by some of their own members (De Vries et al., 2021). In several democracies, liberal democratic institutions have been questioned or undermined by illiberal governments (Haggard & Kaufman, 2021; Przeworski, 2019), and political parties that openly contest democratic principles enjoy considerable electoral support from domestic publics (Greskovits, 2015; Knott, 2018). Autocratization in formerly transitioning and consolidating democratic regimes, many of which joined LIO organizations after the Cold War, is a setback to the idea that membership in these organizations helps consolidate democratic rule. Although international organizations can contribute to consolidation by empowering domestic reform elites, they often lack mechanisms to avert and sanction illiberal conduct (Poast & Urpelainen, 2015). Moreover, and in contrast to non-compliance, illiberal contestation, such as casting dissenting votes, using vetos to block decisions, or forcing lower standards, cannot be sanctioned through institutional safeguarding mechanisms (Panke & Petersohn, 2012). Nevertheless, illiberal contestation has potentially important consequences for the functioning and, ultimately, the legitimacy of liberal international institutions (Dingwerth, 2019).

In this paper, we thus ask: What happens within LIO institutions when members deviate from shared liberal norms? Historically, members showing weak commitment to liberal norms and objectives of LIO institutions were relatively common (Börzel & Zürn, 2021). However, many of these illiberal members opted for a strategy of maintaining substantial decoupling between LIO membership and practical compliance at the domestic level. By contrast, more recently, we have witnessed a qualitative shift towards active contestation among illiberal members, posing a potential threat to the institutions of the LIO and their international policymaking. The politicization of liberal international politics at the domestic level furthermore illustrates that domestic and international illiberal contestation has become increasingly intertwined (for a comprehensive approach to measuring norm robustness, see Simmons and Jo, 2019).

We theorize that while ideological opposition to liberal norms is an important prerequisite for contesting liberal international policies, the regime environment influences the extent to which domestic actors are inclined to show contestation behavior at the international level. Whereas previous scholarship has mostly focused on non-compliant regimes, we turn to examining the international behavior of illiberal political parties while taking their domestic regime environment into account. Specifically, we expect that illiberal parties governing illiberal regimes can benefit on the international stage by refraining from contesting liberal norms and institutions while not risking domestic repercussions (cf. Hafner-Burton et al. 2008). By contrast, illiberal parties governing liberal regimes are likely to face more domestic constraints if they publicly commit to international liberal norms, and thus have incentives to discredit the applicability and validity of such international liberal norms. Further, we expect illiberal government parties to moderate their contestation behavior at the international level compared to their counterparts in opposition. More generally, we argue that the state of domestic liberal democracy and the changing domestic party landscape have repercussions for decision-making at the international parliamentary level, which may pave the way for the gradual internal decay of liberal policymaking within LIO institutions.

To substantiate our theoretical claims about how illiberal trends affect organizations promoting liberal principles, such as institutionalized multilateralism and individual liberties, we analyze a novel dataset of more approximately 500,000 recorded votes in the Parliamentary Assembly of the Council of Europe (PACE). The Council of Europe paradigmatically represents the judicialization of international politics and is among the oldest and the most potent LIO organizations. PACE thus provides an ideal study ground for our analysis in several ways. First, the extent to which parties successfully mobilize voters by attacking liberal democratic ideas and universal human rights differs considerably between PACE’s parties and member countries. Since PACE’s goal is to promote democracy, the rule of law, and civil and minority rights, we obtain an extensive time series of parties frequently voting on liberal-democratic principles, unique in its comparability across time, parties, and countries. With its 47 member states, including various non-EU members such as Azerbaijan, Georgia, and Russia, the Assembly provides us with the largest possible variation of political regimes and party positions. It allows us to disentangle the effects of liberal democracy at the country level versus illiberal positions at the party level and conditioning factors such as political parties’ opposition or government status.

Our findings indicate that contestation has been growing in recent years, but liberal parties continue to dominate the decision-making process. While illiberal parties from illiberal domestic regimes (congruent illiberals) have constantly been a significant source of contestation, the group of illiberal parties operating in liberal political regimes (dissenting illiberals) has grown in recent years and is similarly inclined to contest liberal international policies. Consistent with our theoretical expectations, we find that illiberal government parties moderate their behavior compared to their opposition counterparts, but only when it comes to amendment votes on liberal proposals and only among dissenting illiberal parties. By contrast, we do not find evidence that congruent illiberal parties in government contest liberal international policies less frequently than governing dissenting illiberal parties. Liberal parties do not differ in their voting behavior, irrespective of the political regime in which they operate or whether they are in government, a finding that can be interpreted as an indicator of deeper norm internalization.

Our paper makes several contributions to scholarship on the drivers of contestation to the LIO. First, our study documents the rise of illiberal actors in IOs and provides insights into some of the consequences for international cooperation, which speaks to the broader thematic focus of this special issue on autocratization processes within international organizations. Second, we provide new evidence of the democracy-enhancing effects of multilateralism and, in particular, international parliamentarization (Keohane et al., 2009; Moravcsik, 1995). Third, our paper speaks to earlier contributions arguing that illiberal domestic governments do not necessarily contest liberal norms and universal rights at the international level (Hafner-Burton et al., 2008; Hathaway, 2007) but that the interplay between ideological illiberalism and domestic regime type matters for whether they do. Lastly, our findings carry implications for the ongoing debates among international policymakers on whether LIO organizations should exclude illiberal members to ensure continuity in liberal policymaking.

2 Illiberal political parties and liberal international organizations

Concerns over the legitimate exercise of authority by international organizations are older than the recent illiberal backlash. Criticism, however, originated from a fundamentally liberal concern over the democratic accountability of international decision-making. In response to legitimization pressure, many international organizations have adopted parliamentary branches, resulting in a surge of parliamentarization since the late 1980s (Schimmelfennig et al., 2021). Because they involve elected non-governmental actors, namely, parliamentarians, International Parliamentary Institutions (IPIs) are viewed as only weakly dominated by territorial state interests and characterized by a greater level of assertiveness.Footnote 2 However, to effectively promote liberal norms, international organizations and their parliamentary branches depend on the democratic orientation of their members (Keohane, 2006).

Since the end of the Cold War, membership in the institutions of the liberal international order, including the European human rights regime, has become nearly universal. As we will argue, a new challenge emerges from within LIO organizations as illiberal challenger parties in liberal democracies and trends of democratic backsliding have not only implications for the trajectory of domestic democratic systems but also for the decision-making processes of LIO organizations. Contrasting with previous research, we focus on illiberal political parties—rather than regimes—and the behavioral implications of their illiberal orientation for membership in LIO organizations.

By illiberal parties, we refer to political parties who lack a “public commitment to legal means for gaining power, and rejection of the use of force” (Linz, 1978 p. 29). Such actors commonly refuse to abide by the rules of the democratic game, question political opponents’ legitimacy of participating in the political process, accept or even support violence, and endorse restrictions on civil liberties such as the freedom of the press (Levitsky & Ziblatt, 2018 p. 22). Illiberal parties, including those that have emerged in established democracies in recent years, dislike non-majoritarian democratic institutions, such as an independent judiciary at home or courts at the international level (e.g., Petrov, 2020). They frequently oppose the pooling of authority in multilateral institutions, especially if they are perceived to push for the primacy of universal rights over popular sovereignty (Börzel & Zürn, 2021). As human rights regimes seek to empower individual citizens vis-á-vis the state through the provision of international adjudication with sanctioning power (Moravcsik, 2000), and eliminate policy content that discriminates based on race, ethnicity, gender, or sexuality, illiberal parties criticize them for illegitimately meddling in domestic affairs and therefore undemocratic (Dellmuth & Tallberg, 2021; Koch, 2021). Thus, illiberal parties’ backlash against the liberal international order is often framed as a quest to regain national sovereignty and protect majoritarian decision-making at the domestic level (Koch, 2021; Petrov, 2020).

The concept of illiberal parties is somewhat different from populism (Medzihorsky & Lindberg, 2023). Populist parties are characterized by a “thin” ideology that postulates an antagonism between the people and the elites (Mudde, 2004). Their strategy usually seeks to divide societal groups into ‘good’ and ‘evil’ (Taggart, 2000). Populist parties are thus primarily defined by their polarizing and anti-elite rhetoric and can pursue a variety of ideological preferences. While the majority of populist parties in Western Europe pursue anti-liberal and anti-democratic agendas and can thus be classified as both populist and illiberal, in this paper, we focus on the ideological tension between the illiberalism of parties and liberal international policies.

2.1 A typology of illiberal parties

Because of the ideological tension between their illiberal orientation and the LIO’s liberal character, we would expect illiberal parties to make use of their membership to contest liberal norms and institutions that could interfere with their illiberal agenda. Yet, as long as liberal norms remain a dominant feature that the international order seeks to promote, there are benefits to refraining from open contestation that illiberal parties cannot or will not forego (Debre, 2021; Kelemen, 2020; Nielsen & Simmons, 2015). Since the end of the Cold War, membership in the institutions of the liberal international order, and especially the European human rights regime, has become nearly universal, stretching the regime to its geographic limits. Furthermore, illiberal actors are incentivized to join LIO organizations through material payoffs and substantive legitimizing benefits they receive in return for participation (e.g., Hafner-Burton et al., 2008).

As an example of those incentives, consider the benefits associated with membership in the European Union and NATO. While gaining membership requires the ratification of core human rights treaties, it offers market access, visa liberalizations, and security partnerships in return. Joining human rights regimes is also useful for leaders of illiberal regimes that seek to establish the legitimacy of their international authority, even when membership does not provide any material benefits (Busygina & Kahn, 2020). Additionally, the limited enforcement capacity of international institutions, especially human rights regimes, makes participation in the Liberal International Order (LIO) an economical legitimation strategy (Hafner-Burton et al., 2008; Vreeland, 2019).

Illiberal parties can thus pursue a certain degree of decoupling between their behavior in international fora and domestic conduct, i.e., strategically support liberal policy and contest only a limited scope of selected liberal norms that they perceive as particularly threatening to their illiberal agenda (see also Winzen et al., 2022). We argue that illiberal parties weigh the benefits of participating in, and paying lip service to, LIO decisions against the repercussions for their domestic power status (cf. Debre, 2021; Moravcsik, 2000; Nielsen and Simmons, 2015). The degree to which illiberal parties face domestic political competition, scrutiny from media, civil society, and judicial review may have important implications for their contestation behavior. We advance a power-based logic underlying liberal norm contestation but focus on domestic, rather than international, power competition (cf. Bettiza and Lewis, 2020; Mattes et al. , 2015).

Given the literature’s focus on (non-democratic) governments as actors and the United Nations human rights regime, scholars have usually equated regime type with actors’ illiberal orientation (Hafner-Burton et al., 2008; Hathaway, 2007; Hug, 2016; Kentikelenis & Voeten, 2021; Simmons, 2009; Vreeland, 2019; Voss, 2019). Differentiating between illiberal parties and their regime environment helps us theorize the international behavior of actors that have received little scholarly attention: illiberal parties in liberal democratic regimes and illiberal opposition parties in both liberal and illiberal regimes. Hence, apart from the extent to which a party holds an illiberal orientation, we distinguish whether a party’s domestic political environment is characterized by illiberal or liberal governance and whether it participates in government or not.

Fig. 1
figure 1

A typology of illiberal parties

Full size image

As Fig. 1 illustrates, differentiating parties along the party and domestic regime dimensions leads to four types of political parties. Our argument focuses on the differences in behavior between the two types of illiberal parties, as we are particularly interested in how the domestic regime environment moderates illiberal parties’ behavior. International liberal norms are primarily aligned with the strategic interests of liberal parties, including dissenting liberals such as the Hungarian Socialist Party, the Polish Civic Platform, and the Serbian Democratic Party.

In this paper, we focus on comparing the behavior of dissenting and congruent illiberals (left-hand side of Fig. 1). Dissenting illiberals refer to illiberal parties from liberal domestic regimes, whereby the party’s illiberal orientation conflicts with its regime environment. Examples are the German Alternative for Germany (AfD) and the Greek Coalition of the Radical Left. By contrast, congruent illiberals are illiberal parties that operate in an illiberal regime and are congruent to their environment. United Russia is a case in point for this party type, as well as the Turkish Justice and Development Party (AKP) and smaller opposition parties such as the Serb Democratic Party in Bosnia and Herzegovina. Having established a typology of illiberal parties, the following sections formulate our empirical expectations as to whether different types of illiberal parties show varying levels of contestation behavior.

2.2 International contestation behavior of illiberal parties: Theoretical expectations

We structure our theoretical expectations as follows. First, we develop expectations with respect to illiberal government parties’ contestation behavior depending on their regime environment (H1). Second, we compare illiberal opposition parties to government parties and develop a government moderation hypothesis (H2). We illustrate the comparisons of interest and hypotheses in Fig. 2.

Fig. 2
figure 2

Comparisons and hypotheses

Full size image

2.2.1 Congruent and dissenting illiberal governments

In previous research, regime type has mostly been equated with the government’s orientation towards liberal democracy. As a result, dissenting illiberal governments have so far received scant scholarly attention. Given the literature’s focus on regime type, existing scholarship would categorize dissenting illiberal governments together with liberal governments as liberal-democratic actors. At the same time, there are a number of studies that examine populist backlash against multilateralism (e.g., Koch, 2021). Nevertheless, these studies concentrate on party types, often overlooking the regime environment. Furthermore, as we have already discussed, populism and liberalism represent distinct party dimensions, with the latter more precisely capturing the ideological tension with the Liberal International Order (LIO) from a conceptual perspective.Footnote 3

We expect that the tension between dissenting illiberals’ preference for fewer checks and balances on executive power, on the one hand, and their membership in the community of democratic countries leading the LIO institutions, on the other hand, produces incentives to openly contest liberal norms in international fora for two reasons: to deter liberal opposition and to cater to their illiberal constituency.

Dissenting illiberal governments strategically assess the signaling impact of their international actions, factoring in domestic audiences, including electoral supporters and their elite support base (cf. Mattes et al., 2015). Inking international treaties or supporting liberal policies delineates acceptable behavior, reshaping domestic expectations (Simmons, 2009). The potential reference to rights in a ratified treaty can mobilize domestic groups, potentially leading to litigation for future rights abuses. In a liberal democracy, the opposition is bolstered by free media coverage, resources from civil society groups, and independent courts that may not favor the government. This renders symbolic international commitment a riskier strategy for dissenting illiberal governments compared to congruent illiberal counterparts.

Several examples from compliance literature indicate that governments falling into the category closest to dissenting illiberals—i.e., those facing executive constraints—are generally less inclined to publicly commit to liberal norms. For instance, governments constrained by checks on executive power display a reduced likelihood of signing human rights treaties (Hafner-Burton & Tsutsui, 2007). Notably, fully autocratic regimes, accustomed to routine torture, are not more hesitant than compliant liberal democracies to ratify the UN Convention Against Torture (CAT). However, democratic regimes that engage in torture as a punitive measure are less inclined to ratify the Convention (Hathaway, 2007). Additionally, democratic regimes with recent histories of violent conflict are less likely to join the International Criminal Court (ICC) compared to autocracies, suggesting that fragile democratic regimes are more reluctant to commit to the LIO than their outright autocratic counterparts (Simmons & Danner, 2010). In a similar vein, states possessing robust domestic enforcement mechanisms are also dissuaded from ratifying the treaty prohibiting child labor if they do not comply with its provisions. This is due to the potential risk of facing effective enforcement, resulting in an improvement in children’s rights compared to autocracies that have ratified the treaty (Stein, 2016).

Moreover, rulings from the European Court of Human Rights (ECHR) have been demonstrated to serve as an impetus for the reform of LGBT rights, particularly in nations where national courts can rely on such precedents (Helfer & Voeten, 2014). This phenomenon sheds light on why illiberal entities not only contest international courts and treaties for perceived encroachments on sovereignty but also consistently target domestic courts (Petrov, 2020; Voeten, 2020). Given that dissenting illiberal governments contend with influential domestic actors with both the means and motives to challenge the government’s actions, there exists a potential for a liberal backlash if they are perceived as mimicking a liberal image on the international stage.

As for the second rationale—that is, catering to domestic constituents—, dissenting illiberal governments can also rely on a segment of the electorate that supports or tolerates their illiberal agenda. While dissenting illiberals face constraints in their use of repression tools compared to congruent illiberals, they gain legitimacy through being elected in free and fair elections. Dissenting illiberal governments appeal to a median voter with, on average, pro-sovereignty and often anti-liberal preferences. For our argument to hold, it is not crucial whether voters possess explicit knowledge and preferences regarding politics within a specific LIO institution, although they might. It is sufficient for dissenting illiberals to link their international behavior to significant topics that divide societal groups and mobilize existing discontent. That is, the promotion of liberal core values by international institutions is easily and increasingly integrated into the domestic debate over the loss of national control and sovereignty concerns (De Vries et al., 2021).

Dissenting illiberals may exploit their open opposition to the LIO by employing rhetoric that diminishes the relevance of liberal democratic standards within their country. Labeling their governance as an “alternative form of democracy” and condemning democracy promoters’ involvement as cultural interference enhances domestic narratives of legitimacy (Debre, 2021). The alignment between majoritarian interpretations of democracy and dissenting illiberals’ assertion to represent portions of the electorate contributes to their perceived legitimacy and reinforces their persuasive rhetoric (cf. Guriev and Treisman, 2022). Dissenting illiberal governments can utilize their majoritarian support base as a shield against both international and domestic interventions. Notably, the politicization of liberal norms in other democracies is welcomed by dissenting illiberals, as it fosters (legitimate) skepticism about the universal applicability of norms among domestic audiences (Terman & Byun, 2022). Thus, perhaps ironically, the liberal regime environment positions dissenting illiberals to effectively question and contest liberal norms and the institutions advocating for them.

To summarize, we expect dissenting illiberal governments to show a higher level of contestation in LIO organizations than congruent illiberal governments. Even though liberal LIO organizations’ instruments to sanction illiberal regimes are limited, they have the potential to boost the liberal domestic opposition’s standing, thereby posing a threat to the dissenting illiberal’s agenda. Seeking to mitigate the influence of liberal international regimes on domestic power structures, dissenting illiberals should use contestation to weaken liberal norms at the international level. By contrast, congruent illiberals face more external pressure while having to fear less domestic repercussions, which is why they should have fewer incentives to contest decisions on liberal norms and instruments at the international level.

Hypothesis 1 Dissenting illiberals in government contest international liberal policies more frequently than congruent illiberals in government.

2.2.2 Illiberal government versus opposition parties

The above rationale encounters challenges when applied to illiberal opposition parties. Since we examine a non-governmental assembly, it is also in order to distinguish between government and opposition parties. International Parliamentary Institutions (IPIs) often mandate proportional representation in the delegation based on seats in national legislatures, a rule generally adhered to, albeit to varying extents (Lipps, 2020). Consequently, if sufficiently large, illiberal opposition parties can secure their place in the delegation. But what motivates their behavior as international delegates?

Previous research on the behavioral implications of illiberalism within LIO institutions has primarily centered on the United Nations human rights regime, focusing on the conduct of governmental representatives exclusively (see Hug, 2016; Voss, 2019). As argued earlier, the legitimizing advantages of endorsing liberal policies for governments manifest at the international level, particularly for congruent illiberal governments. Furthermore, governments may face the consequences of non-compliance, such as international isolation.

Illiberal opposition parties lack the same incentives to adhere to international norms. Crucially, opposition parties exert less influence over the design of domestic political institutions, the operational principles of the state apparatus, and foreign policy—typically a domain dominated by a strong executive (Raunio & Wagner, 2019). Consequently, they are typically not the primary targets of blame for illiberal practices within the international community. Additionally, due to their restricted access to international resources suitable for domestic co-optation strategies, they have far less to gain from adopting face-saving behaviors and seeking international recognition. Thus, we would expect that illiberal opposition can afford to consistently contest liberal politics more freely than illiberal governments.

Hypothesis 2 Illiberal parties in opposition contest international liberal policies more frequently than illiberal parties in government (government moderation hypothesis).

Besides testing the government moderation hypothesis across parties, we furthermore examine whether illiberal government parties moderate their contestation behavior depending on their domestic regime environment. That is, we study whether government moderation is similar or different between illiberal and liberal regime environments.

3 Evidence from the Parliamentary Assembly of the Council of Europe

PACE was founded in 1949 and today consists of 324 parliamentarians delegated to the Assembly from the national parliaments of its 47 member countries. The Assembly elects the judges of the European Court of Human Rights, the Commissioner for Human Rights, and the Secretary-General of the Council of Europe. It can summon Heads of State and Government to answer its questions, conduct election observations, and inquire to collect evidence on human rights violations. It may furthermore recommend actions to the intergovernmental Committee of Ministers. The Council is obliged to respond to PACE requests, which can go as far as recommending the expulsion of a country from the Council of Europe.Footnote 4 Monitoring procedures, which entail regular visits by two rapporteurs to meet civil society and institutional actors in the country in question, are also part of the Assembly’s area of competence.

We select the Parliamentary Assembly of the Council of Europe (PACE) for four main reasons. First, despite the Council of Europe being an international organization wielding relatively weak hard authority (Zürn et al., 2021), it possesses normative authority. Much like other inclusive institutions such as the United Nations, it has the potential to exert a significant influence on international politics by articulating shared values on matters such as human rights, thereby shaping perceptions that either legitimize or delegitimize state conduct (Binder & Payton, 2019). The European human rights regime is the most developed in terms of institutional powers in the world, and PACE plays an essential role in it, holding more far-reaching powers compared to other international parliamentary institutions (Rocabert, 2020).

Second, the declared goal of the Assembly is to uphold human rights, democracy, and the rule of law, an objective that illiberal actors frequently denounce as interference with domestic affairs. PACE thus offers a unique environment to examine political actors’ repeated and comparable position-taking on democracy and human rights at the international level. In the Assembly, voting is non-binding, and roll-call votes can be viewed as an individual delegate’s choice to publicly endorse or oppose a proposal. Therefore, a delegate has the flexibility to deviate from the position of her party group or country delegation.

Third, PACE membership is diverse with regard to democratic conduct and support for democratic norms, providing us with the largest possible variation of liberal and illiberal parties, liberal and illiberal regimes, and their governmental status. Fourth, to ensure transparency and accountability, the Assembly records votes electronically since 2007, offering us a rich data source for nearly all decisions made in PACE.Footnote 5

3.1 Data and methods

The Assembly holds four sessions annually and debates about 10 to 15 topics per session. Roll calls may include resolutions, recommendations to the Council of Ministers, or oral and written amendments tabled during the plenary debate. Resolutions are decided by a simple majority, and recommendations require a two-thirds majority to pass. For an amendment to be tabled, it must be sponsored by at least five parliamentarians.Footnote 6 The latter makes up the most significant portion of PACE votes. PACE publishes the detailed voting records of its electronic voting system online on its official website, accessible to the public.Footnote 7

Our analysis rests on an original roll-call dataset on PACE legislators’ complete voting records from 2007 to 2021. Each parliamentarian who cast a vote is listed with vote choice, full name, country, and PACE party group. Using the hyperlinks on the PACE webpage, we additionally scraped information from the profile pages of all (voting) parliamentarians, such as their national party affiliation. We also gathered information on the title and the assigned committee’s name for the draft of each legislative text and linked it with the voting data. In addition, starting with sessions in 2012, we scraped the text proposals for amendments along with the names of the tabling signatories. This scraping process yielded approximately 2000 amendments.Footnote 8 In total, our dataset covers 689 legislative proposals and 5,554 roll calls cast by 1987 individual parliamentarians representing 278 different national parties.

3.1.1 Dependent variables

Voting behavior in international organizations is a central way to openly contest the promotion of liberal values (Meyerrose and Nooruddin, this issue, see also Bailey et al. (2017); Bailey and Voeten (2018); Binder and Payton (2019); Hug and Lukacs (2014)). Our definition of “liberal” relies on the multi-dimensional conception of liberal democracy as formulated in Engler (2023): the separation of powers and constitutional jurisdiction, the rule of law, the independence of the judiciary, and equality before the law, freedom of the press, freedom of speech, and freedom of assembly, as well as individual liberties and fundamental rights, such as physical integrity, property rights, human rights, or the freedom of religion. As we elaborate below, we develop two measures to capture whether a vote cast in PACE is in favor of or against liberal democratic values.

Contestation in final votes First, we operationalize contestation as a delegate’s vote against final texts in PACE. We do not consider abstentions in the analysis as the reasons for not participating in roll call votes can be manifold. However, we report models with abstentions coded as no votes in Appendix Table G.4. We validated our measure with human coders who assessed all recommendations and resolutions for their (il)liberal content. This validation exercise indicated that all but four final texts can be categorized as liberal, confirming that no votes contest the liberal content of final texts.

Contestation in amendment votes In contrast to the majority position on final votes (i.e., resolutions and recommendations), amendments provide legislators with an opportunity to propose changes to legislative texts, which may run counter to the advancement of liberal principles, but occasionally be adopted by a PACE majority. Several examples highlight the diverse intentions behind amendments, demonstrating that they are not merely corrections but are driven by distinct political agendas. In 2013, Hungary came under pressure for threatening media freedom, leading PACE to consider placing the country under a monitoring procedure. However, a group of conservative MPs from several Western European countries sponsored an amendment aimed at deleting the critical call for monitoring from the draft resolution.Footnote 9 The amendment was adopted with 135 votes in favor, 88 votes against, and 6 abstentions. The resolution passed with even more votes in favor.

Conversely, a group of Italian MPs from different parties advocated for the inclusion of a phrase in a resolution on media freedom that read: “The Assembly invites Member States to promote the dissemination of information about criminal proceedings against organized crime and to discourage all those who try to reduce the visibility of the above proceedings.”Footnote 10 This amendment was unanimously adopted.

To address the issue that some adopted amendments may pursue illiberal goals, we opted for a machine-learning approach to classify amendment texts as either in support of or against liberal values or as neutral. We tasked two human coders to classify a training dataset (17 percent of the overall number of amendments) with 346 amendments. With this training dataset, we implemented a multinomial machine learning model, with the predictors being words mentioned in the amendment texts. This approach leads to a strong prediction performance: the model correctly classified 99% of liberal amendments and 100% of illiberal training amendments. We provide details on the machine learning approach in Appendix E. We code all votes in favor of illiberal amendments and against liberal amendments as contesting votes.

3.1.2 Independent variables

To examine which members of parliament (MPs) are most likely to contest the majority in PACE roll calls, we match each PACE legislator’s national party affiliation with information about their party’s political orientation in the V-Party dataset (Lührmann et al., 2020), an expert survey consisting of various measures to document parties’ policy stances and ideological characteristics worldwide over time. Moreover, we rely on the Varieties of Democracy (V-Dem) dataset to measure countries’ levels of liberal democracy (Coppedge, 2020). We focus on the following variables.

Party (il)liberalism We use V-Party’s illiberalism index as an indicator of the extent to which the national party shows anti-democratic rhetoric and conduct.Footnote 11 This index considers severe attacks against political opponents, rejecting political pluralism and minority rights and refusing to condemn political violence (cf. Medzihorsky & Lindberg, 2023).

Country (il)liberalism We use V-Dem’s liberal democracy index to measure to what extent an MP is a delegate of a liberal democratic regime. The liberal democracy score relies on the degree to which the regime can be considered an electoral democracy (in the spirit of Dahl’s polyarchy concept) and the extent to which liberal democratic norms, e.g., constraints on the executive and minority rights, are institutionalized and complied with in practice. The liberal component adopts a critical stance toward power, assessing the quality of democracy based on the constraints imposed on the government. It underscores the importance of safeguarding rights against both the tyranny of the state and the tyranny of the majority.Footnote 12

Drawing on these variables, we categorized parties into four types that align with our theory. All parties with a party and country liberalism score of at least 8 (on a scale from 0 to 10) are coded as congruent liberals, which we use as a baseline in our regression models. Substantially, this category refers to all liberal parties that operate in a liberal political regime. We chose the threshold of 8 because it is commonly used in V-Dem’s categorization scheme for liberal democracies. Moreover, as shown in Appendix B, applying this threshold has face value: all parties coded as liberal baseline are commonly held to commit to liberal values and operate in comparatively strong liberal democracies.

In line with our theory, we distinguish between two types of illiberal parties. For one, we refer to a party as dissenting illiberal when it runs in a liberal political regime but rejects liberal principles. Appendix B provides a full list of party classifications from 2007 to 2019.

Domestic government status We consider whether the national party has been involved in national government formation. Drawing on the V-Party data, we classify all parties to fulfill those criteria if i) it serves as a senior partner (head of government is affiliated with that party) or ii) is a junior partner (head of government not affiliated with this party, but at least one cabinet minister is). All other parties are coded as opposition parties.

3.1.3 Control variables

In addition to these three variables, we control for several other factors that might be associated with the likelihood of contesting PACE decisions. These refer to other national party orientations and their share of seats in the national parliament.

Economic left-right orientation We incorporate V-Party’s economic left-right indicator as another explanatory variable. This indicator refers to the extent to which the party prioritizes a government actively intervening in economic affairs or prefers a limited role of the government in economic affairs.

Cultural liberal-conservative orientation The cultural dimension differentiates parties with respect to their position on non-economic issues, typically questions of minority and civil rights. To capture the extent to which parties favor conservative over liberal cultural values or vice versa, we draw on a variable indicating to what degree a party supports the equal participation of women in the labor market.Footnote 13

Seat share in national legislature Lastly, the size of a party is an indicator of how influential a party is domestically and how many representatives it can send to PACE. We control for the share of seats held in the national legislature.

All variables except government participation and seat share are divided by two standard deviations to make them more comparable. A higher value indicates a more liberal, economic right, or culturally liberal orientation.

3.2 Results

Our empirical analysis proceeds in the following steps. First, we analyze the degree to which legislative texts proposed to PACE are met with contestation over time. This descriptive analysis allows us to assess the extent to which decision-making in PACE has faced contestation from parliamentarians over the last decades. Second, we investigate the strength of illiberal parties and how often they contest proposals made in PACE. Lastly, we test our hypotheses for a) final decisions and b) amendment proposals.

3.2.1 Contestation in PACE over time

First, we analyze the development of contestation in PACE over time. The inquiry into the overall divisiveness of PACE decisions allows us to assess the extent to which liberal policies have been met with increasing backlash in recent years. We consider contestation to be the extent to which MPs are divided over a policy issue. We capture the level of cross-voting among all legislators of PACE for each of the 5554 roll-call votes we record, including amendments. The Rice Index is a commonly used measure to determine voting unity among legislators (Carey, 2007).Footnote 14 The Index ranges from zero (i.e., parliament is split equally between MPs who voted in favor and MPs who voted against a proposal) to a value of 1 (i.e., all MPs voted in unanimity.)Footnote 15

We choose a threshold of RI < 0.8Footnote 16 for contested roll-calls, equivalent to the minority side representing ten percent of the votes. Subsequently, we calculate the share of contested roll-calls over all roll-calls in each PACE session.Footnote 17 Figure 3 reveals that between 2007 and 2012, only around 50 percent of roll-calls can be considered contested. Starting in 2017, the share of contested roll calls has sharply increased. In the most recent year of our analysis, 2021, a significant share of 70 percent of roll-calls faces opposition from at least 10 percent of voting MPs. These patterns provide the first important insight into contestation dynamics in PACE: decision-making in parliamentary votes has become less consensual, and a larger share of MPs contests votes in PACE.

As we discuss in the next section, we suggest that this change is driven by the emergence of a larger group of illiberal parties. However, it is also possible that a shift in PACE’s legislative agenda is contributing to the increase in contestation. To investigate this possibility, we plotted the number of draft texts proposed by each of the committees in Fig. 3. No obvious relationship emerges: contestation has been similarly low, whether the Assembly debated many political and social affairs (2009-2011) or not (2012-2016), and has not risen significantly with migration dominating the agenda in 2016. Still, contestation increases with more legal affairs tabled in 2018. Yet, the number of legal affairs proposals was also relatively high in other years (2007, 2010, 2011), while contestation was not as high. We conclude that agenda change in PACE is not the main driver of contestation levels.

3.2.2 The strength of illiberal parties in PACE

Building on our typology of illiberal parties, we distinguish between types of illiberal parties and evaluate which party groups may drive our first finding that disagreement has been increasing in PACE in recent years.Footnote 18 Figure 4(a) shows the percentage of votes that were cast by each of the four party groups. While members affiliated with a liberal party operating in a liberal regime (congruent liberals) had made the lion’s share of votes up until 2014, their weight in PACE has been decreasing since then. At the same time, the number of dissenting illiberals increased considerably, although they remain below 10 percent. This is also mirrored by the standard deviation of party illiberalism increasing over time (see Appendix Fig. C.2).

Fig. 3
figure 3

Share of contested over all roll-calls over time and number of legislative texts by committee over time

Full size image

Congruent illiberals represent the largest group of illiberal parties. Hence, congruent illiberals possess considerable power in PACE regarding votes, but dissenting illiberals are on the rise and may play an increasingly important role in PACE decisions if their upward trend continues.

Next, we examine how forcefully illiberal parties make use of their right to vote in final PACE decisions. Figure 4(b) plots the share of contesting votes cast in final decisions (recommendations and resolutions). In most years, congruent illiberals have contested the majority more than dissenting illiberals. The recent uptick in contesting votes from dissenting illiberals furthermore suggests that this party group is increasing its efforts to build a coherent opposition to liberal forces in PACE. In 2018 and 2019, dissenting illiberals voted even more often against the majority than all other party groups, including congruent illiberals, although representing the smallest party group in PACE. Despite these trends, liberal parties continue to dominate final decisions in PACE.

3.2.3 Predictors of contestation voting behavior

We now turn to analyzing the determinants of contestation behavior, allowing us to test our hypotheses in a regression framework. The unit of analysis is the individual MP casting a vote. We employ a mixed-effects modeling approach that specifies random effects of PACE party, MP nested in national party nested in countries, and document ID.Footnote 19 While our article primarily focuses on the results obtained through our regression framework, Appendix D implements an ideal point estimation approach. Although key results can be replicated using ideal point estimation, we favor the regression framework. Ideal point estimation is useful to identify a latent conflict dimension and learn about opposing political camps. However, we study the relative frequency of contestation by pre-defined actor groups on a known dimension—liberalism. Additionally, the frequent turnover of delegates poses challenges for reliably estimating individual legislators’ ideal points.

Fig. 4
figure 4

Votes cast by illiberal and liberal parties in PACE, 2007-2019

Full size image

While we restrict attention to mixed-effects regression models, our findings are robust to numerous different model specifications, including using continuous measures of party and country liberalism (Appendix F.2), controlling for overtime effects of democratic backsliding (Appendix G.2), fixed effects specifications with standard errors clustered at the document level (Appendix G.3), and recoding abstentions to no votes (Appendix G.4).

Contestation in final votes We begin our analysis with recommendations and resolutions, which represent the final votes after amendments have either been adopted or rejected. We designate a vote against a final text as a contesting vote, serving as our dependent variable. Table 1 reports the results of this analysis, and Fig. 5 plots the predicted probabilities based on Model 2 in Table 1. As expected, both congruent and dissenting illiberal parties are more likely to contest liberal final legislative proposals. However, we do not find support for our hypothesis that dissenting illiberal governments contest more often than congruent illiberal governments (H1). Similarly, our results do not lend support to the hypothesis that illiberal government parties contest less than their illiberal opposition counterparts (H2). This finding holds for both dissenting and congruent illiberal parties: Opposition parties do not have a higher propensity to contest liberal final texts among either of the two party groups

Table 1 Generalized probit regression for voting against final liberal legislative texts
Full size table
Fig. 5
figure 5

Predicted probabilities of voting against liberal final legislative texts in PACE (Model 2 in Table 1)

Full size image
3.2.3.1 Contestation in amendment votes

Next, we turn to contesting behavior in amendment votes. We differentiate between two types of contesting behavior: a) voting against liberal proposals, and b) voting in favor of illiberal proposals. Table 2 reports the results for these two types of contesting behavior, and Fig. 6(a) displays the corresponding predicted probabilities. For voting against liberal amendments, we find that illiberal parties are not more likely to contest liberal proposals than liberal parties. Instead, on average, dissenting and congruent liberal parties are more likely to reject liberal amendments than illiberal parties. The results furthermore do not provide evidence that dissenting illiberal governments contest more frequently than congruent illiberal governments (H1).

While we do not find support for a government moderation effect (H2) across illiberal parties, differentiating between dissenting and congruent illiberals reveals an intriguing pattern: while congruent illiberal government parties are more likely to contest liberal amendments than their counterparts in opposition, we find a reverse relationship among dissenting illiberals. That is, dissenting illiberal government parties considerably moderate their contesting behavior compared to their opposition counterparts. Hence, our results confirm the government moderation hypothesis (H2) for dissenting but not for congruent illiberals.

Table 2 Generalized probit regression for voting against liberal amendments (Models 1-2) and in favor of illiberal amendments (Models 3-4)
Full size table

As for voting in favor of illiberal proposals, a different picture emerges. In contrast to rejecting liberal proposals, congruent illiberals are the most likely to vote for illiberal amendments, followed by dissenting illiberals, dissenting liberals, and congruent liberals. We find that dissenting illiberals in government are slightly more likely to contest than congruent illiberal parties in government (H1). As for the government moderation hypothesis (H2), similar to rejecting liberal amendments, our results do not suggest that government parties consistently contest less than opposition parties across illiberal parties. Instead, we find that dissenting illiberal parties in government are more likely to support illiberal amendments than their counterparts in opposition. We find a reverse relationship among congruent illiberal parties: among this party group, government parties tend to be in favor of illiberal proposals less frequently than their opposition counterparts.

To summarize, across legislative texts (i.e., final and amendment stages), our findings do not provide consistent evidence for our hypothesis that dissenting illiberal government parties contest liberal proposals more than congruent illiberal government parties. Congruent illiberals are slightly more likely to support illiberal amendments than dissenting illiberals. By contrast, we find a substantial government moderation effect for dissenting illiberal parties in voting against liberal amendments and a weaker government moderation effect among congruent illiberal parties in voting in favor of illiberal amendments. Illiberal opposition parties’ contestation behavior is also contingent on regime type: dissenting illiberal opposition parties tend to oppose liberal amendments as much as congruent illiberal governments but are not more likely to support illiberal amendments than liberal parties. Congruent illiberal opposition parties, in contrast, do not focus on opposing liberal amendments but are the strongest supporters of illiberal amendments compared to all other party types.

Fig. 6
figure 6

Predicted probabilities of voting in amendment decisions

Full size image

4 Conclusions

Over several decades, liberal international institutions included members that did not align with the fundamental values they sought to promote. At the same time, non-compliance did not necessarily translate into open conflict with other members of the institution, which speaks in favor of norm robustness (Simmons & Jo, 2019). Recently, however, international governance has come under increasing attack from parliamentarians both on the left and right (e.g., Hooghe et al. 2006). The implications are far-reaching. Importantly, public contestation of liberal values is likely to weaken norm robustness in addition to, and far more than, widespread non-compliance (Deitelhoff & Zimmermann, 2018; Hurd, 2013; Simmons & Jo, 2019). After all, forcing lower liberal international standards through contesting ambitious proposals is not punishable under any formal rule and cannot be sanctioned as non-compliance could.

Our study speaks to the ongoing debate on the consequences of the rise of illiberal actors by showing that if the democratic basis of liberal international institutions erodes, international institutions will face increasing levels of open conflict over the values they are founded on. Drawing on new roll-call data from the Parliamentary Assembly of the Council of Europe (PACE), our results suggest that the level of contestation in floor votes is on the rise and that, simultaneously, the number of delegates from illiberal parties has been growing. Nevertheless, liberal parties have retained a majority in the Assembly and continue to support liberal policies. The behavior of liberal parties’ MPs is not conditioned by the government status of their party or their home country’s level of liberal democracy, suggesting that they are still largely aligned with PACE’s liberal agenda, independent of domestic democratic developments. By contrast, we find divergent behavior among illiberal parties, which suggests that their willingness to contest the LIO is shaped by domestic calculations. In line with our expectations, being in government moderates illiberal parties’ contestation behavior, but only in liberal regime environments (i.e., dissenting illiberal parties) and at the amendment stage.

Contra to expectations, we do not find consistent evidence that dissenting illiberal government parties contest the LIO more frequently than their illiberal counterparts in illiberal regime environments. The strategies of illiberal parties vary also with regard to the (il)liberal content of proposals: dissenting illiberal governments prioritize supporting illiberal content, but the dissenting illiberal opposition focuses on voting against liberal content. This divergent strategy can be interpreted in light of domestic power struggles, with the dissenting illiberal opposition primarily competing against liberal mainstream parties while dissenting illiberal governments seek to consolidate their domestic attempts to undermine liberal democratic institutions. Congruent illiberals pursue both strategies.

Our findings carry significant implications for PACE’s ability to curb human rights violations. The standing of illiberal parties in the Assembly has evolved, with their group expanding due to parties whose illiberal positions have gained legitimacy through free elections in liberal democracies. These parties have increasingly succeeded in amassing a critical mass of votes on numerous proposals, altering the content of decisions made in PACE. This shift appears to have prompted them to abandon their strategy of mimicry and coexistence in favor of active resistance. They can now argue that their challenger position stems not from a fear of losing power but rather from the appeal of their competing ideals to citizens.

These findings carry implications for how liberal international policymakers might address the challenge posed by emerging illiberal actors. There is reason to assert that the Council of Europe can only wield influence upon a country’s accession. If illiberal actors are invited in or the commitment to liberal values changes during membership, the liberal actors need to unite across all factions in order to prevent lower standards or norm decay. A prerequisite for this mechanism to be effective, however, is the critical mass of members remaining liberal and coordinating their response. Membership also binds countries to the rulings of the European Court of Human Rights (ECHR) and provides support in legal reform and constitutional matters through expert and training missions (Stivachtis & Habegger, 2011). Exclusion from membership deprives the opposition, civil society actors, and individuals of access to the ECHR. In addition, exclusion precludes the representation of liberal opposition parties from illiberal regimes in PACE. Working through inclusion channels is potentially most fruitful when liberal domestic actors remain significant political competitors in the domestic arena. At the same time, to better fulfill its liberal goals amid a more diverse membership, PACE could engage more critically with democratic spin rhetoric to retain its prerogative of interpretation over acceptable conduct in a liberal-democratic community of states. Additionally, PACE could prioritize resolving the backlog of cases at the ECHR, which may bolster liberal opposition to backsliding governments.

However, a prerequisite for this mechanism to work is that PACE—as well as other international parliaments—establishes formal ties to liberal opposition actors in illiberal member states. If LIO regimes do not enable international policymakers to connect with and empower liberal domestic actors beyond the government, the benefits of expelling illiberal member states may be greater, allowing the organization to continue to promote liberal international policies.