A conceptual definition of sovereignism

The concept of sovereignism is surrounded by a certain degree of uncertainty. Research is in fact still in its infancy and any attempt to study sovereignism requires a preliminary conceptualization of the object at stake. First, we agree with Linda Basile and Oscar Mazzoleni (Reference Basile and Mazzoleni2020) that sovereignism is not sovereignty. If it is true that conflicts concerning sovereignty are ‘as old as the concept itself’ (Bickerton et al. Reference Bickerton, Brack, Coman and Crespy2022) and that the two concepts of sovereignism and sovereignty are clearly connected, it is also worth clarifying that in our understanding sovereignism does not overlap with sovereignty, but is rather its ideological and social interpretation.

Sovereignty refers to the political authority of governing bodies within a bounded territory. In this perspective, the concept of sovereignty has mostly been articulated in relation to the Westphalian state, referring to a form of organization of political power which is exerted within a clearly defined territory and political community, and which does not recognize any kind of superior authority (Keating Reference Keating and Walker2003). In the context of liberal democracies, the concept of sovereignty has been then specified in terms of popular sovereignty (Sassen Reference Sassen2008) where ‘sovereignty … is an expression of the popular will; … such popular will is expressed through mechanisms of representative democracy, on which liberal democracies rely’ (Basile and Mazzoleni Reference Basile and Mazzoleni2020: 152).

Sovereignism, instead, can be understood as claiming political control of the sovereign people within the boundaries of the traditional nation state. As such, sovereignism emerges as a relatively recent phenomenon and as a reaction to processes of de-territorialization and dispersion of state sovereignty. Economic globalization, and political and institutional integration have progressively challenged the sovereignty of nation states (Grimm Reference Grimm2015; Sassen Reference Sassen2008). More intense fluxes of capital and persons have reduced the capacity of the state to control its borders; at the same time, a globally integrated economic and political system has often failed to be responsive to citizens' demands (Basile and Mazzoleni Reference Basile and Mazzoleni2020: 153). As a consequence, these processes have displaced the traditional authority of the nation state and have been perceived by large shares of citizens as detrimental.

All this has paved the way to sovereignism, which, as a claim, manifests itself as the desire to ‘get back control’ (Kallis Reference Kallis2018), that is to return to the traditional understanding of sovereignty based on the idea of mutually exclusive territories defined by the traditional form of the state (Basile and Mazzoleni Reference Basile and Mazzoleni2020: 154). From the point of view of parties, sovereignist claims are articulated through positions in favour of a restoration of the power of the nation state, vis-à-vis supranational institutions and economic, political and cultural integration.Footnote ¹ All these elements can co-exist and together they represent the most hard-line version of sovereignism but, as we show in the article, it is also possible that only some of these elements are present in one's stance, pointing to different nuances of sovereignism.

We conceive of sovereignism as a distinct concept compared to similar ones such as populism. Populist discourses are imbued with sovereignist claims; however, the latter can also be found among mainstream actors. The claim of restoring the national autonomy of the British Conservative Party in the context of the Brexit referendum is probably the most straightforward example of a sovereignist stance promoted by an established political party.

Finally, we define sovereignism as a multidimensional concept (Basile et al. Reference Basile, Borri and Verzichelli2020). While in its most emblematic form the sovereignist claim can be reduced to the idea of taking back control, it concerns different dimensions (and, as a consequence, different sovereignist reactions). Basile and Mazzoleni (Reference Basile and Mazzoleni2020) identified three dimensions of integration qualifying sovereignism which we borrow in our study: political, economic and cultural.

As a reaction to political integration, sovereignist claims are articulated on the very basic idea of restoring the control of the sovereign state over borders and decision-making processes. With respect to economic integration, sovereignism claims the restoration of the state authority over policy areas such as trade (promoting the return to a protectionist approach) and monetary policies. At the same time, it opposes economic globalization and integration in the international economic system. As a reaction to cultural integration and contamination, sovereignism appeals to the values of the native population which, with its tradition and culture, is the only one entitled to constitute the political community (and so, to be the legitimate holder of popular sovereignty and citizen rights). As we show in the article, the concept of sovereignism may combine claims of (reappropriation of) state control on the political, economic and cultural dimensions in various ways and it may create different sovereignist profiles.

The interplay between ideological predispositions and sovereignist party supply

Our party-level correlates of sovereignism focus on the ideological predispositions of parties and the competition between mainstream and radical parties. We agree with Giorgos Charalambous and colleagues (Reference Charalambous, Conti and Pedrazzani2021) that agency matters in issue mobilization by virtue of the relevance of party-level characteristics. In this respect, we consider ideology a first important factor as different party families exhibit distinct patterns in the evolution of their issue profiles. Firstly, we expect ideological positioning along the left–right spectrum to predict the adoption of sovereignist party claims.

The literature has consistently identified the greater predisposition of right-wing parties and, even more so, radical right parties (RRPs) to incorporate national identity and sociocultural issues into their profile (Halikiopoulou and Vlandas Reference Halikiopoulou and Vlandas2019). RRPs have pledged to reaffirm the sovereign authority of the nation state in fundamental policy domains, such as border controls and budgetary/economic prerogatives, rejecting at the same time the constraints imposed by globalization and European integration. In doing so, they have expressed their support for national identity against the cultural threats posed by multiculturalism (Basile et al. Reference Basile, Borri and Verzichelli2020). Moreover, RRPs have claimed to restore economic sovereignty, linking this stance to the prosperity of the national community, also adopting chauvinistic welfare stances to make the native population the only recipient of social services (Otjes et al. Reference Otjes, Ivaldi, Ravik Jupskås and Mazzoleni2018). But, although central to RRPs, sovereignist claims are not exclusive to them as, with different nuances, conservative parties may also decide to adopt such claims, often to narrow the programmatic distance to challengers to their right (Heinisch et al. Reference Heinisch, Werner and Habersack2020).

At the other end of the ideological spectrum, radical left parties (RLPs) are also likely to rally citizens on sovereignist issues. RLPs have conceived of the nation state as a hospitable environment for achieving their policy objectives, such as social protection and economic redistribution (Gerbaudo Reference Gerbaudo2021; Halikiopoulou et al. Reference Halikiopoulou, Nanou and Vasilopoulou2012). Their economic sovereignism stems from opposition to the neoliberal policies of contemporary capitalism (March Reference March2011). In particular, these parties have been committed to reasserting nation-state prerogatives on budgetary and fiscal policies, often blaming the processes of globalization and European integration for the growing levels of economic inequality at the domestic level (Basile et al. Reference Basile, Borri and Verzichelli2020; Carrieri and Vittori Reference Carrieri and Vittori2021).

It is worth noting that one of the corollaries of ideological sovereignism among RLPs and RRPs has been their common opposition towards the European integration processes (Braun et al. Reference Braun, Popa and Schmitt2019), although RLPs have often been divided on the EU (e.g. Charalambous Reference Charalambous2011).Footnote ² On the one hand, RRPs have mainly rooted their Euroscepticism in culture-based explanations, such as the defence of nation and the rejection of the free circulation of people within the EU (De Vries and Edwards Reference De Vries and Edwards2009). On the other hand, RLPs have ideologically grounded their Euroscepticism within the economic (neoliberal) embodiment of the EU project, which could be detrimental for the popular classes (Halikiopoulou et al. Reference Halikiopoulou, Nanou and Vasilopoulou2012).

In brief, the ideological position and level of radicalism along the left–right dimension appear to be conducive to a more sovereignist party supply. Indeed, we expect to find a higher degree of sovereignism at right-of-centre and at the extremes of the left–right dimension:

In the last few decades, political parties have increasingly located themselves along the so-called GAL–TAN dimension of political contestation (Hooghe et al. Reference Hooghe, Marks and Wilson2002). The issue of national sovereignty appears to be a fundamental parcel of this dimension, with those parties leaning towards the traditional, authoritarian and nationalist (TAN) pole being likely to oppose supranational integration (Jackson and Jolly Reference Jackson and Jolly2021). Indeed, one of the core ideological commitments of TAN parties has been the defence of the nation state against the transnational threats posed by immigration and European integration. On the contrary, green, alternative and libertarian (GAL) parties have endorsed more cosmopolitan worldviews, compatible with the free circulation of people and the deepening of European integration. Therefore, we predict finding higher levels of sovereignism on the TAN side of this dimension:

The interplay between structural factors, public opinion and the sovereignist party supply

Justification of sovereignist party supply has often focused on economic and sociocultural trends – namely, economic insecurity and cultural backlash – as drivers that predispose society against supranational integration while creating, at the same time, the demand for parties to act on the sources of social and economic distress and the logic of boundary construction (Hooghe and Marks Reference Hooghe and Marks2009; Kriesi et al. Reference Kriesi, Grande, Lachat, Dolezal, Bornschier and Frey2006; Norris and Inglehart Reference Norris and Inglehart2019).

Research shows that the perceived threat of outsiders, or outside powers, is often associated with perceptions about economic, cultural and physical security (De Spiegeleire et al. Reference De Spiegeleire, Skinner and Sweijs2017). As to economic security, the economic changes induced by interdependent economies and globalized markets and competition create a sense of insecurity among the people who have been left behind (Scheve and Slaughter Reference Scheve and Slaughter2004). In those countries with a high degree of economic interdependence (such as the EU countries), when the state of the economy worsens, public reaction against interdependence may also involve the broader population (Broz et al. Reference Broz, Frieden and Weymouth2021; Pappas and Kriesi Reference Pappas, Kriesi, Kriesi and Pappas2015). Indeed, failure to address the distributional consequences of economic shocks has been considered a key factor behind the ascendance of sovereignism in the political supply (Colantone and Stanig Reference Colantone and Stanig2019). In such cases, the economic and political elites who promoted supranational integration have become a main target for a widespread resentment (Colgan and Keohane Reference Colgan and Keohane2017).

The cultural backlash argument complements the economic explanations. Liberal values have become increasingly embedded within (and reinforced by) supranational institutions, such as the EU and other international organizations. But many authors have documented a cultural backlash among the Western public consisting of a reaction against the mainstreaming of liberal values (including free movement of goods and people) and the resurgence of national identity. Shrinking support for globalization and supranational integration, along with support for boundary construction and fear of immigrants, are considered a consequence of such a backlash (Norris and Inglehart Reference Norris and Inglehart2019; Oesch Reference Oesch2008). In particular, growing immigration rates have triggered concerns about the loss of national identity and negative attitudes towards European integration among citizens (Otjes and Katsanidou Reference Otjes and Katsanidou2017). Indeed, an increasing share of the native population has perceived migration (and multiculturalism) to be a detrimental process for the social cohesion and integrity of the national community (Tillman Reference Tillman2021). Terrorist attacks have aggravated the perception that foreigners threaten the physical security of natives (Böhmelt et al. Reference Böhmelt, Bove and Nussio2020) and the peaks in the migration flows of the past decade have certainly intensified the backlash effect.

Starting from the above premises, it is easy to understand that – given the impact of multiple crises on the EU in the past decade – cultural resentment towards cosmopolitan values, economic openness and immigration has become more widespread in European countries (Ausserladscheider Reference Ausserladscheider2019). This turn of public opinion has proved immensely important for the shaping of the European political landscape of the recent past. Indeed, public opinion has become more vocal on the activation of national identities and transfer of distrust from the national political system to supranational governance. Sovereignty-related matters have also become more relevant to people's electoral choices (Tillman Reference Tillman2021).

It is difficult to predict which of the different party types is closer to the public mood on sovereignty issues. Benjamin Ferland (Reference Ferland2020) maintains that parties adjust their position in line with changes in public opinion, depending on the party type and the direction of change of the public. The niche actors (smaller parties with an ‘extreme’ ideology – i.e. the RLPs and RRPs) should be more responsive to any policy shift of their partisan base (Adams et al. Reference Adams, Clark, Ezrow and Glasgow2006). Mainstream parties (larger organizations holding more centrist positions) should be more inclined to respond to the changing positions of the median voter (Ezrow et al. Reference Ezrow, De Vries, Steenbergen and Edwards2011). But the question is controversial and some authors maintain that European parties do not respond to public opinion on any issue dimension, and their policy commitments are not necessarily in tune with public mood (O'Grady and Abou-Chadi Reference O'Grady and Abou-Chadi2019) while policy outcomes can be more important for the public. But with specific reference to sovereignty issues, Luca Carrieri and Nicolò Conti (Reference Carrieri and Conti2023) show that parties are sensitive to public mood and responsive to their electoral bases. Oscar Mazzoleni and Gilles Ivaldi (Reference Mazzoleni and Ivaldi2023) prove that sovereignist attitudes have a significant effect on public support for the main advocates of sovereignist claims, namely RRPs.

We are also aware that the relationship between parties and voters is complex and reverse causation is possible. Parties may respond to the public, but they may also play an entrepreneurial role on different issues and influence the public. The problem of who cues whom between demand and supply is theoretically intricate and empirically difficult to demonstrate. In this work, rather than testing the causal effect of public stances on party supply, we simply generate expectations about their association.

The data

We test our hypotheses relying on the combination of three data sources based on party, public and country levels covering 27 European states at two time points: 2014 and 2019.Footnote ³ Our unit of analysis is at the party level, for which we derive our measures from two waves (2014 and 2019) of the Chapel Hill Expert Survey (CHES) data set. CHES data are used to position parties on multiple dimensions and they are adapted to new conflict lines as they emerge, with the experts relying on a multiple set of sources (such as speeches, roll-call votes, public debates, manifestos etc.) to classify parties' actual stances (Rohrschneider and Whitefield Reference Rohrschneider and Whitefield2012). This data set, in particular, nicely fits the purpose of our research, as it includes several items which separately tap into the three dimensions of our definition of party sovereignism, thus allowing us to develop a robust measure of the concept. Then, at the public level, we made use of two standard Eurobarometer studies surveyed in November 2014 (Eurobarometer 82.3) and in June–July 2019 (Eurobarometer 91.5), both including questions referring to sovereignty. Finally, at the country level, we relied on official statistics.

Dependent and independent variables

Our main dependent variable is an indicator of party sovereignism. Coherently with our definition, our index of party-based sovereignism is derived from the aggregation of several CHES items. Specifically, in 2019 we take into consideration four items which separately tap into our dimensions of sovereignism: party positions on the EU integration process,Footnote ⁴ nationalism, economic protectionism and multiculturalism. The item on EU positions relates to sovereignism defined as a main characteristic of Euroscepticism (Borriello and Brack Reference Borriello and Brack2019). The item on nationalism relates to sovereignism given that any call of ‘taking back control’ emphasizes lines of nation-statism (Kallis Reference Kallis2018). Economic protectionism clearly reflects hostility to the loss of control of the state over the economy and to the current trajectory of international trade (Jedinger and Burger Reference Jedinger and Burger2020). Finally, as to multiculturalism, assimilation of migrants has been considered a way through which states shape new members' values and identity and harmonize them with the state norms and principles (King et al. Reference King, Le Galès, Vitale, King and Le Galès2017). Thus, state assimilation can be seen as a tool through which the state seeks to (re)impose itself as the locus of control over the systems of values that build the fabric of society.

To assess whether these four items combine together in a conceptually consistent manner, we first standardized them (M = 0; SD = 1) and then we performed a confirmatory factor analysis. The results provided support for a one-factor solution, with all the four variables displaying a factor loading higher than 0.3, with nationalism and multiculturalism being the most relevant components of our latent factor (factor loadings being 0.96 and 0.90, respectively), followed by protectionism (0.50) and EU integration (0.32).Footnote ⁵ The Cronbach alpha scale reliability coefficient for the four items is equal to 0.74. On this basis, we then summed up the four items to create our additive index of sovereignism (our dependent variable).

Unfortunately, the 2014 CHES data do not include all the items used to build up our index in 2019. In particular, the economic protectionism item is missing. As a consequence, in order to be able to rely on a proxy of sovereignism also in 2014, we restricted our index to three items included in this wave (i.e. EU integration, multiculturalism and nationalism). To validate our 2014 proxy, we then correlated it with our index of sovereignism as it was calculated in 2019. To do so, we reshaped the CHES data from the usual long format into the wide format, with each column of the matrix now recording information for each party in each of the analysed years. The bivariate correlation coefficient is significantly strong (R = 0.75, p < 0.001), thus providing clear evidence of the fact that the two indexes capture the same phenomenon (with some deficiency that may well be due to the likely variations of party sovereignism over time). Furthermore, as a robustness test, we replicated all our analyses using a perfectly symmetric (across 2014 and 2019) index of sovereignism, combining only the three items which are included both in 2014 and 2019, and excluding the economic protectionism item in 2019.

We conducted further robustness tests to validate our measures of party sovereignism. Since we define sovereignism as a continuum between two distinct poles (i.e. sovereignism and anti-sovereignism), different nuances should be possible between the two most extreme scores. If the items that we selected can operationalize the latent concept of sovereignism, they should load on the same factor, particularly for those (sovereignist and anti-sovereignist) parties located at the two extremes of the continuum. Thus, we conducted a factor analysis considering only those parties that fell within the top and bottom deciles of the party sovereignism distribution (i.e. extreme sovereignist and anti-sovereignist parties). Then, we performed the same analysis on the remaining observations, which included parties falling between the top and bottom deciles of the distribution.

We observed that the results of the factor analyses were significantly stronger when we restricted our data to the most extreme cases of sovereignism and anti-sovereignism, compared to using the entire sample, or including only those observations falling between the two extreme poles. Specifically, for the asymmetric index in 2019 (which includes the economic protectionism item), we found support for a one-factor solution both when we considered only the most extreme parties and when we included all the other parties. However, the eigenvalue for the extracted factor was 3.5 in the former case, while it was 2.1 in the latter case. Additionally, all factor loadings consistently pointed in the same direction for the most extreme cases and were all above 0.6. When we considered parties falling between the top and bottom deciles, as we expected, the factor loadings became more uncertain (the EU item fell below the conventional threshold of 0.3).

We obtained similar results in the analysis of 2014. The factor analysis resulted in a one-factor solution when we considered both extreme and the rest of parties, but with different eigenvalues (2.8 in the former and 1.7 in the latter cases, respectively). Among extreme parties, all items loaded on the same factor, with all factor loadings above 0.9. However, the same was not true when we restricted our analysis to parties that were less extreme. In this case, only nationalism and multiculturalism loaded on the same factor with factor loadings above 0.9, while the EU item exhibited a weak loading (01.) on the same factor.Footnote ⁶

Overall, these analyses confirm the existence of a latent sovereignist dimension sustaining the most resolute stances (extreme sovereignism and anti-sovereignism) built on all our selected items. The two poles of sovereignism and anti-sovereignism can be effectively discriminated from the rest using the items we have chosen.

In addition to our index of sovereignism calculated in 2019 and 2014, we built a variable measuring the variation of party sovereignism over time. This is done by simply subtracting the party sovereignist score as measured in 2014 from the one measured in 2019. To make the values of this variable more immediately understandable, we first rescaled the two indexes of sovereignism in 2014 and 2019 to a range from 0 to 1 and then we calculated the difference between the two. This yields to a variable with positive values for those parties which moved towards more sovereignist positions in 2019 (compared to 2014) and negative values otherwise.

As for our main independent variables, we measured the ideological location of parties relying on the general left–right and the GAL–TAN position provided by the CHES data set both in 2014 and 2019 (both variables ranging from 0 to 10, where 0 = ‘Left/GAL’ and 10 = ‘Right/TAN’). We saw that in 2019 there is a stronger correlation between GAL–TAN (R = 0.82) and our index of party-based sovereignism, than between left–right (R = 0.49) and the same index (Figures C1 and C2 in the Supplementary Material).

As for public positions, we built an index of sovereignism in public opinion out of five items measured by the Eurobarometer both in 2014 and 2019. In particular, we took into consideration the variables measuring the image citizens have of the EU, globalization, free trade, protectionism and immigration. We selected these items based on the assumption that holding a more negative image in each item (and a positive one in protectionism) should favour support for border regimes, thus for sovereignism (see also Karstens Reference Karstens2020). While the EU item is measured on a five-point scale (ranging from 1 = ‘very positive’ to 5 ‘very negative’, with 3 indicating a ‘neutral’ position), all the others are measured on a four-point scale (ranging from 1 = ‘very positive’ to 4 = ‘very negative’). We thus rescaled the five items to range from 0 to 1 and we then summed them in an additive index (rescaled to a range from 0 to 1). As a final step, we averaged this indicator by country in order to get a measure of the scope of the public sovereignist mood in each country in each year.

Additionally, to test the differentiated relationship between public demands and sovereignist party supply, we included an interaction term between public opinion sovereignism and party type. The party type variable has been operationalized as a dichotomous variable (0 for RLPs/RRPs and 1 for mainstream parties – i.e. conservative, Christian democrats, liberal, social democrats), relying on the CHES variable that distinguishes between party families.

Finally, to measure the economic performance of countries and their exposure to immigration, we relied on the Eurostat measures of GDP per capita (expressed in purchasing power standards) and immigration rate (measured as the proportion, in percentage, of immigrants over the total population in each country in each reference year).

We also included as control variables the electoral strength of each party in the last general election and a variable accounting for regional differences across Eastern, Southern, and Western Europe. The electoral size of parties is included to control for the differential incentives that different-in-size parties might have in adopting sovereignist claims. Although we do not advance any formal hypothesis in this respect, we expect that smaller parties might have major incentives to introduce new issues into the political debate (as for them this should be relatively easier and less risky) (Meyer and Wagner Reference Meyer and Wagner2013). Along the same lines, we know that party losers (i.e. those experiencing an electoral defeat) are also more likely to adopt entrepreneurial issue strategies (Hobolt and De Vries Reference Hobolt and De Vries2015).

The variable for regional clusters is included to account for differences in the dynamics and content of party competition across regions. The literature on party competition has indeed shown that, while the logic of party competition in Eastern and Western Europe is somewhat similar, it differs radically in terms of content (Rovny and Edwards Reference Rovny and Edwards2012). At the same time, the recent literature has also shown that the structure of political conflict in Southern Europe (mostly dominated by the economy and the conflict between old and new politics) differs from the structure of conflict in North-West European countries (which is instead mostly dominated by the EU and immigration issues) (e.g. Charalambous et al. Reference Charalambous, Conti and Pedrazzani2021; Hutter et al. Reference Hutter, Kriesi and Vidal2018).

Modelling strategy

Our data are organized into a hierarchical structure. Our dependent variable is measured at the party level (i.e. our unit of analysis), with parties then clustered into 27 countries. At the same time, we use predictors which are measured both at party level (i.e. left–right, GAL–TAN) and country level (i.e. public opinion mood, economic performance and immigration rate). To account for this data structure, we opted for the estimation of a series of OLS regression models with clustered standard errors adjusted for the 27 European countries included in the analysis.Footnote ⁷

Party sovereignism in 2014 and 2019

To test our hypotheses, we begin regressing party sovereignism over our main predictors separately for 2014 and 2019, using for 2019 both the symmetric index of sovereignism and the asymmetric one. In so doing we should also gauge a dynamic perspective by assessing how party sovereignism (and its correlates) has changed over a five-year period. The results of these analyses are presented in Table 1.

Table 1. OLS Regression Analysis with Clustered Standard Errors

Notes: Dependent variable: Party sovereignism in 2014 and 2019. Standard errors (in parentheses) are robust, adjusted for the 27 European countries included in the analysis. Predictors are measured in the year of reference of the dependent variable. * p < 0.05, ** p < 0.01, *** p < 0.001.

Source: CHES 2014 and 2019.

In Models 1a and 1b, we assess the relationship between left–right party positions and party sovereignism in 2014 and in 2019, net of controls and using for 2019 the asymmetric index of sovereignism. In Model 1c we replicate the same analysis, but this time using the symmetric index of sovereignism for 2019. In all cases we observe a positive and statistically significant coefficient, suggesting that both in 2014 and in 2019 right-wing parties were more likely to adopt sovereignist positions. In Models 2a, 2b and 2c we dig deeper into this relation, plugging into our models a quadratic term for the left–right position of parties to account for nonlinearity (as we suggested in H1a). In line with our hypothesis, when using the asymmetric index of sovereignism, there is evidence of a statistically significant nonlinear relation between the ideological position of parties and their sovereignist stances in 2019. As is clearly observable in Figure 1, which reports the predictive margins of left–right position on party sovereignism (as estimated in Model 2b), sovereignism seems not to be ideologically one-sided. Although RRPs were certainly those displaying the highest degree of sovereignism (and centre-right parties more so than left-wing parties), in 2019 both RLPs and RRPs were more likely to display sovereignist positions than the more moderate political parties.

Figure 1. Predictive Margins of Left–Right Party Positions on Party Sovereignism (2019)

However, the same result does not hold in 2014 and it is only slightly significant in 2019 when using the symmetric index of sovereignism (Model 2c). In 2014, indeed, no significant curvilinear relation can be found between the left–right position of parties and their sovereignist positions (Model 2a). In 2019, instead, this result is fully confirmed only when the dependent variable incorporates party positions on economic protectionism, while it is weakened when the economic item is excluded. This seems to confirm that there exist different dimensions of sovereignism, and that different kinds of parties are differently positioned on these dimensions. While RRPs are relatively more sovereignist when considering the cultural, economic and institutional dimensions of sovereignism altogether, RLPs are more likely to adopt sovereignist positions when the economy is at stake. But even when using the symmetric index of sovereignism, there are some signs of a curvilinear effect of the left–right positions of parties on party sovereignism in 2019, a clue that, between 2014 and 2019, left-wing parties (more than right-wing parties) might have espoused more sovereignist positions.

One might hypothesize a contagion effect, with RRPs dragging RLPs to compete on a newly politicized dimension. This argument is in line with the thesis that left parties react to the surge of RRPs by adopting more restrictive positions on issues such as immigration and EU integration (Bale et al. Reference Bale, Green-Pedersen, Krouwel, Luther and Sitter2010). In this sense, among left parties, the Danish Social Democrats can be paradigmatic: when the immigration issue reached a peak in the Danish agenda, first the Social Democrats tried to defuse the issue, then as the issue remained highly politicized, they adopted more restrictive positions (McManus and Falkenbach Reference McManus and Falkenbach2022). The same kind of contagion effect has been documented in relation to the EU, with evidence showing that left parties are the most likely to be affected by a Eurosceptic contagion from radical parties (Meijers Reference Meijers2017). The counter-discourse against the hegemonic practices of the EU has become intertwined with an increased focus on national autonomy (Eklundh Reference Eklundh, García and Briziarelli2018). A notable (more radical) example is France Insoumise (FI), which has adopted a Eurosceptic populist stance to oppose the capitalist nature of the EU integration project. This party even pledged to take France out of the EU, while adopting more ambivalent positions on immigration (Ivaldi Reference Ivaldi2018). In Greece, SYRIZA is another case in point of a leftist party opposing dominant EU politics and policy (Serafis et al. Reference Serafis, Kitis and Assimakopoulos2022). According to Paolo Gerbaudo (Reference Gerbaudo2021), among leftist parties, sovereignism made significant inroads, especially among RLPs as a strategy to broaden their electoral support, notably among working-class constituents, to weaken the social bloc behind the RRPs. He argued that RLPs have developed a ‘social protectivism’ identity; for example, in the aftermath of the Great Recession, Podemos in Spain and FI in France claimed the supremacy of the state vis-à-vis international private market forces considered to threaten the well-being of the lower classes.

Contagion in the opposite direction is also possible. We should recall here that on issues such as those pertaining to trade liberalization and international agreements, the RLPs have often taken the lead (Tereszkiewicz Reference Tereszkiewicz2021) and RRPs have tried to adopt analyses and positions originally developed by the RLPs (Rone Reference Rone2018).

Our data lead to partial acceptance of our hypothesis H1a: if it is true that in 2019 sovereignist positions were more likely to be adopted by both RRPs and, to a lesser extent, by RLPs, this was not at all the case in 2014, when RRPs were clearly taking the lead in voicing sovereignist positions. Therefore, H1 appears to capture a process that is in the making and that may well document a fundamental pattern of realignment in contemporary party systems.

The spread of radical left sovereignism is documented in Figures 2A and 2B plotting the correlation between sovereignism variations between 2014 and 2019 and party ideology.Footnote ⁸ In particular, in Figure 2A we built the party sovereignism variations using, for 2019, the asymmetric index of sovereignism; in Figure 2B, instead, for 2019 we used the symmetric index. In both cases we observe a negative association, with RLPs (many of which scored above the 0 line) being more likely to shift towards sovereignist positions in 2019 compared to 2014. Furthermore, in line with our previous findings, it is clear that this association is stronger when using the asymmetric index of sovereignism (Figure 2A) rather than the symmetric one (Figure 2B). This confirms that although in general RLPs moved towards more sovereignist positions in 2019, this shift is more clearly apparent when the economic dimension is taken into consideration, as their sovereignism is mainly rooted in economic protectionism and social protectivism.

Figure 2A. Radical Left Party Positions and Variations in Party Sovereignism between 2019 and 2014

Note: Party sovereignism in 2019 includes the economic protectionism item.

Figure 2B. Radical Left Party Positions and Variations in Party Sovereignism between 2019 and 2014

Note: Party sovereignism in 2019 does not include the economic protectionism item.

In Models 3a, 3b and 3c we then assess the relationship between GAL–TAN and party sovereignism. Empirical results confirm our expectations condensed in H1b: both in 2014 and 2019, TAN parties were more likely than GAL parties to adopt sovereignist positions (independently of the construction of our dependent variable). Thus, unlike what we have observed for the left–right ideological position, over time sovereignism has remained a prerogative of TAN parties.

These results are even more interesting when the left–right and GAL–TAN positions are estimated simultaneously (Models 4a, 4b and 4c). Results show that in 2014, when controlled for the left–right position of parties, the positive effect of GAL–TAN remains statistically significant, proving, once again, that TAN parties were more likely to adopt sovereignist claims. However, the same does not apply to the left–right position. Specifically, in 2014 the effect of left–right loses statistical significance when controlled for the GAL–TAN position of parties. On the contrary, in 2019 (Models 4b and 4c), both the curvilinear relation between the left–right and party sovereignism and the linear relation between GAL–TAN and party sovereignism are confirmed (with, again, results more neatly defined when using our asymmetric index of sovereignism, Model 4b). This finding appears to suggest that, in the period of time studied here, party sovereignism underwent a relevant process of transformation: while in 2014 it was mostly subsumed under the GAL–TAN dimension, it became a more articulated and cross-cutting phenomenon in 2019, gathering (together with traditional TAN and radical right parties) also an increasing portion of radical left parties. These findings appear to confirm the expectation of an increasing orthogonality between GAL–TAN and left–right of which sovereignty issues represent an important parcel (Hooghe and Marks Reference Hooghe and Marks2018).

Finally, in Models 5a, 5b and 5c, besides all the other predictors already discussed, we include our indicator of public sovereignism and two indicators for the economic performance of the country (GDP per capita) and its exposure to immigration (immigration rate), respectively. In no case can we find any significant effect of these variables. More in detail, we find non-significant relationship of public sovereignism (both in 2014 and 2019), GDP and immigration rate with party-based sovereignism.Footnote ⁹ However, it is interesting to note that, even when including country-level predictors in our models, previous findings concerning the left–right and GAL–TAN party positions are fully confirmed.Footnote ¹⁰

Variations of party sovereignism over time

So far, we have analysed party sovereignism in 2014 and 2019 separately. In this section we take a step forwards and we analyse variations in party sovereignism over time. In particular, we take into consideration correlates of party shifts towards more or less sovereignist positions. To do so, here we use as our dependent variable the variation in party sovereignism obtained by the difference between party sovereignism in 2019 and in 2014 (i.e. Δ Party sovereignism). Accordingly, we use the variations between 2019 and 2014 of our main independent variables. Furthermore, to test the robustness of our results, we performed these analyses using different operationalizations of our dependent variable. In Models 1–4 depicted in Table 2, we used as dependent variable the difference between the asymmetric index of sovereignism in 2019 and our index of sovereignism in 2014; in Models 5–8, instead, we used the difference between the symmetric index of sovereignism in 2019 and the index of sovereignism in 2014.

Table 2. OLS Regression Analysis with Clustered Standard Errors

Notes: Dependent variable: Δ Party sovereignism (2019–2014). Standard errors (in parentheses) are robust, adjusted for the 27 European countries included in the analysis. * p < 0.05, ** p < 0.01, *** p < 0.001.

Source: CHES 2014 and 2019.

We first notice that neither the variations in the left–right nor the variations in the GAL–TAN party position had a significant relationship with party sovereignist shifts between 2014 and 2019 (Table 2, Model 1). Unsurprisingly, both the left–right and GAL–TAN positions of parties remained rather stable over time (it would be quite implausible that parties would cross these ideological boundaries in such a short period of time). Assuming that the left–right ideological position remained substantially stable over time, it is now safe to conclude that the changing effect of the left–right position on party sovereignism between 2014 and 2019 (with a curvilinear relation becoming statistically significant in 2019) was genuinely due to the fact that RLPs modified their sovereignist location within this span of time. This might lead us to conclude that although sovereignism was not a cross-ideological phenomenon until 2014, it became more so by 2019.

Even more relevantly, we find that changes of public opinion between 2014 and 2019 do have a positive and significant relationship with changes of party positions in the same period (Table 2, Model 2). Specifically, evidence suggests that in those countries where the public moved towards more sovereignist positions in 2019 (compared to 2014), political parties moved accordingly, thus demonstrating a connection between public opinion and political parties on this dimension. Furthermore, we find evidence of a significant and negative effect of country economic performance variations on the variations in party sovereignism (Table 2, Model 3). The negative coefficient for the GDP variation suggests that in those countries which experienced an improvement of their economic performance between 2014 and 2019, parties moved towards relatively less sovereignist positions (for GDP per capita the coefficients are statistically significant but very small due to the measurement of this variable and the fact that numbers are rounded down). On the contrary, where economic performances worsened, parties moved towards more sovereignist stances (H2a).Footnote ¹¹ Instead, and this time contrary to our expectations, we do not find evidence of any significant relationship between variations in immigration rates and in party sovereignist positions (H2b).

Finally, in Model 4 we included an interaction between party type and public opinion to test our H3. The results of this analysis are presented in Figure 3,Footnote ¹² which shows the predicted sovereignist positions of RLPs/RRPs and mainstream parties at different levels of public opinion sovereignism. Our results show that as public opinion becomes more sovereignist, RLPs/RRPs become significantly more sovereignist too. The same, however, does not apply to mainstream parties. In other words, public opinion shifts towards more sovereignist stances have a stronger relationship with the stance of radical parties. This finding demonstrates that the more sovereignist public opinion, the stronger the sovereignist claims of RRPs and RLPs compared to mainstream parties (H3).

Figure 3. Predicted Values of Party Sovereignism Variations between 2019 and 2014 among RLPs/RRPs and Mainstream Parties at Different Levels of Public Opinion Sovereignism (95% CIs)

Note: Panel A considers the symmetric index of party sovereignism; Panel B considers the asymmetric index of party sovereignism. In both cases, the dashed line indicates mainstream parties, while the solid line indicates RLPs/RRPs.

It is finally worth noting that all these findings are robust to the different operationalizations of our dependent variable. Indeed, in Models 5–8 we replicated the same analyses using the symmetric index of sovereignism and we arrived at estimations which are fully consistent with the ones discussed so far.Footnote ¹³

These results provide us with more information to assess the validity of our hypotheses. Taking everything into account, we can confirm our H1a and H1b, although with some caveats. While sovereignism has constantly been a prerogative of TAN and RRPs (and, although less so, of right-of-centre parties in general), now we know that it has also become an ideologically cross-cutting phenomenon, with RLPs recently displaying comparatively more sovereignist positions compared to ideologically centrist parties.

As for hypotheses H2a, H2b and H3, the evidence is more mixed. While we cannot establish a clear connection between our country-level variables (i.e. public opinion, economic performance, immigration rate) and party sovereignism in each specific year, there is enough evidence showing that, on average, changes in party sovereignist positions are significantly correlated with some of these variables. In particular, and in line with H2a, we found that better economic performances depress, on average, party sovereignism; and, in line with H3, the changing orientations of public opinion are strictly connected to changes in party positions of RLPs and RRPs, but not of mainstream parties.