The question of citizenship – and who may access it – arguably remains nowhere more pertinent than in Western Europe in recent years. Over the course of several decades, many European countries have undergone a demographic, socioeconomic and political transformation from relatively homogenous countries of emigration in the immediate post-war period to culturally diverse countries of immigration today, prompting governments to regulate the acquisition of citizenship in ways that define and redefine political membership. Recent pressures associated with responding to the hundreds of thousands of asylum seekers only bring questions of citizenship to new levels of urgency. Indeed, from the introduction of formal language requirements in countries as disparate as Portugal and Finland, to robust civic integration tests in the once avowedly multicultural Netherlands, to the introduction of partial jus soli in a once ethnocultural Germany, few states in Western Europe have left citizenship policy unaltered in recent years. But while some countries have recently introduced new prerequisites for citizenship, others have abolished them. While some have restricted access to naturalization, others have gradually made it more accessible. What explains this variation in the direction and timing of these citizenship policy changes?
Despite a burgeoning literature explaining immigration policy and the spatial positions parties take on the issue of immigration generally, the question of citizenship policy change has remained relatively unexplored, and scholarship on the subject remains in its relative infancy. On the one hand, immigration scholars often examine such policies as the result of pressures and politicization from far right parties (Givens and Luedtke Reference Givens and Luedtke2005; Howard Reference Howard2009; Howard Reference Howard2010) or generic left party power (Janoski Reference Janoski2010). On the other hand, scholars of spatial competition examine parties' manifesto positions on immigration through similar electoral mechanisms, and the rise of far right parties that have owned the issue in recent decades has, not surprisingly, attracted significant attention (Abou-Chadi Reference Abou-Chadi2016a; Abou-Chadi Reference Abou-Chadi2016b; Abou-Chadi Reference Abou-Chadi2018; Akkerman Reference Akkerman2015; Alonso and da Fonseca (Reference Alonso and da Fonseca2011); Bale et al. Reference Bale2010; Downs Reference Downs2001; Van Spanje Reference Van Spanje2010). However, in the midst of increased political fragmentation in the European party system and the rise of an electorally viable far left in recent years, it is increasingly necessary that analyses of immigration policy change move beyond this singular focus and begin to incorporate parties on both the left and right.
In this article, I adapt theorizing from the immigration, citizenship, and party competition literatures in order to argue that citizenship policy changes in either an accessible or restrictive direction result from electoral competition on both sides of the political spectrum in conjunction with the ideological orientation of governments in power. This research makes two distinct contributions to the literatures on comparative immigration policy, party policy positions and electoral competition.
First, I provide robust evidence of a citizenship policy framework that incorporates both left and right, and both liberalization and restriction, into a single empirically tested causal mechanism rather than focusing exclusively on one end of the spectrum or one direction of change. Using a new dataset of citizenship policies across sixteen European countries from 1975 to the present, I demonstrate that left-of-center governments facing increasing levels of leftist party competition are associated with liberalizing policy changes, while increasing levels of party competition from the far right yield more restrictive citizenship policy changes under not only right-of-center governments, but also centrist and even left-of-center governments as well. In other words, both the right and left have substantive effects on citizenship policy change in similar yet distinct ways. This expanded focus helps explain not only the content and direction of policy change in both a liberalizing or restrictive direction, but also its timing.
Secondly, by examining the interplay between electoral competition and government orientation, the framework advanced here also helps shift our focus beyond party positions and onto the policy-making outcomes themselves. In doing so, the empirical analysis establishes a clear link between party competition, government orientation, and actual policy outputs produced by governments across countries and over time.
The article proceeds as follows. I first situate my framework within the literature on immigration policy making and party competition, and generate several testable hypotheses that follow from these literatures. I then advance and discuss my mechanism for understanding how party competition and government orientation jointly affect citizenship policy change. Next, I empirically analyze the direction and timing of citizenship policy change using this citizenship policy data and a novel measure of party competition. The final section concludes with a summary discussion of the findings and their relevance for future comparative analysis.
Electoral Competition, Government Orientation and Citizenship Policy Change
A growing body of scholarly work examines the politics of citizenship. In the subfield of comparative immigration and citizenship, parties’ ideological orientations play a central role in explaining policy outputs and shifts. Leftist parties, which tend to favor social egalitarianism and solidarity among the working class, of which many immigrants are part and which tend to regard immigrants as constituents or potential constituents, should generally favor more accessible citizenship policies when in government compared to their right-of-center counterparts (Joppke Reference Joppke2003, 436). As Messina (Reference Messina2007) notes, ‘the left's traditional policy agenda tends to dovetail with the perceived interests of immigrants’ (208). They also tend to favor citizenship legislation that grants immigrants access to political, economic or social structures to stimulate their integration into society (Ersanilli and Koopmans Reference Ersanilli and Koopmans2010; Joppke and Morawska Reference Joppke and Morawska2003, 430–431). Thus left party power in government may correlate strongly with policy liberalization (Janoski Reference Janoski2010).
Right-of-center parties are more likely to favor more restrictive citizenship policies, in part due to concerns about the negative impact of immigration on social services, crime and terrorism, or national identity (Bale Reference Bale2003; Ireland Reference Ireland2004), and in part because eased access, such as permitting dual citizenship, may undermine the incentive to integrate (Schuck Reference Schuck1998). Conservative parties also often play on the restrictive sentiments of the median voter and highlight immigration issues in order to win votes (Cornelius et al. Reference Cornelius2004; Ivarsflaten Reference Ivarsflaten2005).Footnote 1 This rationale leads to Hypothesis 1:
Hypothesis 1: Citizenship policies should liberalize under ideologically left-of-center governments, and become more restrictive under ideologically right-of-center governments.
An extensive body of scholarship also examines spatial competition as a mechanism for explaining party positions and strategies. Building on Downs (Reference Downs1957), parties compete along policy dimensions by staking out different positions in hopes of attracting voters who are closest to them on those issues. Following this logic, open space on any policy dimension allows new niche parties to emerge and compete for electoral support with established parties, which may then either reposition themselves closer to the position of the niche party, adopt a different policy or dismiss the niche party position entirely (Meguid Reference Meguid2005; Van Spanje Reference Van Spanje2010). The rise of far right parties in particular has attracted widespread attention from immigration scholars using this perspective (Givens Reference Givens2005; Givens and Luedtke Reference Givens and Luedtke2005; Howard Reference Howard2009; Schain Reference Schain2006), as well as from scholars of party politics and electoral competition examining the ‘contagion from the right’ thesis (Abou-Chadi Reference Abou-Chadi2016b; Abou-Chadi Reference Abou-Chadi2018; Akkerman Reference Akkerman2015; Alonso and da Fonseca Reference Alonso and da Fonseca2011; Bale et al. Reference Bale2010; Downs Reference Downs2001; Van Spanje Reference Van Spanje2010). Here, right-of-center parties that may prefer to take a neutral stance on immigration can risk losing electoral support to parties on the far right over the issue, and may move to accommodate the more extreme position while increasing their competitiveness with the mainstream left on the issue (Alonso and da Fonseca Reference Alonso and da Fonseca2011; Bale Reference Bale2003).
Similarly, because left-of-center parties find ideological as well as electoral support among the native-born working class, citizens with immigration backgrounds, and those who prioritize social egalitarianism and multicultural solidarity, any left-of-center party that is perceived to resist the politicization of the immigration issue by the far right and adopt what Meguid (Reference Meguid2005) calls an ‘adversarial strategy’ – in other words, to be less accommodating of restrictionism – risks losing resentful working-class supporters to far right parties and possibly empowering the formation of center-right governments (Alonso and da Fonseca Reference Alonso and da Fonseca2011; Bale et al. Reference Bale2010; Kitschelt and McGann Reference Kitschelt and McGann1995).
Empirical research on the contagion effects of niche parties in competition with mainstream parties has yielded promising, albeit mixed, results.Footnote 2 However, while far right competition and ‘contagion from the right’ have attracted significant scholarly attention, there has been little analysis of parallel dynamics on the left. Though some scholars have begun to examine green parties as examples of niche party competition over issues like the environment (Abou-Chadi Reference Abou-Chadi2016b; Abou-Chadi and Orlowski Reference Abou-Chadi and Orlowski2016; Meguid Reference Meguid2005), as Akkerman (Reference Akkerman2012) notes, ‘systematic research taking account of the dynamics of competition at the right and left sides of the political spectrum is still scarce’ (54). Yet while the radical left has been in general decline since the Cold War and other niche left parties have been relatively modest electoral players in many European countries (March and Mudde Reference March and Mudde2005), this is slowly changing. Support for the radical left is on the rise across Europe: most mainstream left parties now face credible challenges on the left and far left flank (Rooduijn and Akkerman Reference Rooduijn and Akkerman2017; Rooduijn et al. Reference Rooduijn2017). As the party system becomes more fragmented across Europe's political landscape, niche parties like the Greens and new radical left parties like the Dutch Socialist Party often enjoy levels of electoral success commensurate with their counterparts on the right (March Reference March2012; March and Mudde Reference March and Mudde2005).
Government participation by such parties has also become increasingly common, leading Dunphy and Bale (Reference Dunphy and Bale2011) to conclude that it is time to ‘stop treating radical left parties as “niche” players’ and ‘bring them in from the cold’ (489). Furthermore, on immigration issues, mainstream left parties are threatened electorally not only by far-right demands for restriction, but also by green and far-left advocacy of more liberal, egalitarian and multicultural policies that might encourage defections from the mainstream's left-libertarian constituents (Bale et al. Reference Bale2010). This competition, and attempts to accommodate right-wing positions on immigration while not ceding policy space to other leftist parties, have caused the Dutch Labor Party to lose core voters to the Green Left and Dutch Socialist Party throughout the 1990s and 2000s, while also fending off a growing threat from the right-wing Freedom Party (Bale et al. Reference Bale2010, 416–417). The Austrian Social Democrats and Danish Social Democrats have likewise been squeezed by the populist challenge of the Austrian Freedom Party and the Danish People's Party on the right and the Austrian Greens and Danish Social Liberals on the left (Bale et al. Reference Bale2010, 419–420). The Greens in particular have moved beyond single-issue advocacy of environmentalism in many countries and adopted often exceptionally pro-immigration policy stances (March and Mudde Reference March and Mudde2005). For example, as Howard (Reference Howard2008, 47) points out, it was the German Greens throughout the 1980s and 1990s who advocated most vocally for jus soli, dual citizenship and relaxed naturalization requirements, and whose participation in a Social Democrat-led government was instrumental in passing Germany's 1999 citizenship reform.
Thus we have strong theoretical and empirical reasons to include potential ‘contagion effects’ from the right and left in an analysis of immigration and citizenship policy change. Because spatial competition over immigration on both the right and left of the spectrum might affect the direction of citizenship policy change in either a liberal or restrictive direction, we can test two additional hypotheses.
Hypothesis 2: Increasing electoral success of radical right parties will be associated with citizenship policy restrictions.
Hypothesis 3: Increasing electoral success of radical left parties will be associated with citizenship policy liberalizations.
Of course, while spatial competition between emerging parties on the far right or far left may be sufficient to influence the stated immigration policy positions of mainstream parties within party manifestos or expert surveys,Footnote 3 such dynamics may be insufficient to explain how such parties influence actual government policy outputs across countries or over time. Furthermore, it may also be insufficient to assume that the government's ideological orientation and preferences directly shape policy outputs, since governments operate within an electoral and institutional context that incentivizes certain positions and not others (Abou-Chadi Reference Abou-Chadi2016a, 2091). I thus adapt theorizing from the comparative immigration and citizenship literature and the party competition literature to argue that incentives from electoral competition on both sides of the spectrum drive governments’ citizenship policy making.Footnote 4
On the left of the political spectrum, we might expect a left-of-center party to compete with a right-of-center party over a typically immigration-skeptical median voter. In such a context, mainstream left-of-center parties have little electoral incentive to deviate too far in a liberalizing direction and away from this median voter, and may even succumb to the so-called ‘contagion from the right’ (Alonso and da Fonseca Reference Alonso and da Fonseca2011). Yet left parties also must cater to constituencies that may hold more pro-immigrant positions, or risk losing support to other niche left or far left parties (Alonso and da Fonseca Reference Alonso and da Fonseca2011; Bale et al. Reference Bale2010). As the number of left-of-center parties increases to fill open electoral space on their side of the political spectrum, these parties will be forced to compete among themselves with more explicitly pro-immigrant positions, including citizenship policy reform. In these contexts of heightened party density and competition on the left, left-oriented governments in office should find themselves in a more conducive and incentivizing position to liberalize policy in an attempt to accommodate pro-immigration positions to the left and to maintain leftist electoral support rather than cede electoral space to other left parties that may benefit from a more pro-immigrant position on citizenship. The addition of a green or radical left party in government should also be associated with liberalized policies. But because far left parties do not usually compete for support with other conservative parties, left party competition should have little discernible ‘contagion’ effect on right-of-center governments' policy making.
At the opposite end of the political spectrum, we can assume that right-of-center parties maintain a default position that is more consistently skeptical of immigration and restrictive in their application of citizenship law than their left-of-center counterparts. Similarly, as the number of right-of-center parties increases to fill the electoral space on their side of the political spectrum, these parties may be forced to compete among themselves and to accommodate more explicit anti-immigrant positions. Consequently, within an electoral atmosphere dominated by electoral competition on the right, I expect right-of-center governments to be most likely to enact policies that restrict access to citizenship and increase the requirements to acquire it. Where far right parties join the government, thus serving to confirm the salience of the issue within the electorate, citizenship policy is highly likely to be restricted.
Competition, in conjunction with a government's ideological orientation, is thus key to understanding the conditions under which a government might seek to change its citizenship policy in a restrictive or more accessible direction. Empirically, then, this logic leads us to the following final set of testable hypotheses.
Hypothesis 4: Right-of-center governments should be more likely than other governments to restrict their citizenship policies as the number of right-of-center parties in competition with one another increases.
Hypothesis 5: Left-of-center governments should be more likely than other governments to liberalize their citizenship policies as the number of left-of-center parties in competition with one another increases.
Hypothesis 6: The inclusion of a radical left or radical right party in a government coalition should be associated with citizenship policy liberalizations and restrictions, respectively.
Testing the Electoral Competition Model of Citizenship Policy Change
To analyze changes in citizenship policy as a dependent variable, it is necessary to quantify citizenship policies over a sufficient number of years.Footnote 5 To this end, I adapted a citizenship policy framework that builds upon that of Goodman (Reference Goodman2010) using formal state legislation and legal texts as well as scholarly reports on citizenship policies collected by EUDO (2013).Footnote 6 I then coded citizenship policies passed by the legislatures of sixteen European states from 1975 to 2014 on the basis of nine criteria regulating citizenship acquisition.Footnote 7 One set of components includes the more performance-based requirements for naturalization, such as linguistic, economic, civic, legal and familial criteria. The linguistic component captures the degree of linguistic proficiency required, based primarily on the Common European Framework of Reference for Languages, while the economic component gauges financial prerequisites and restrictions on the use of public assistance. The civic component measures the extent of country or civic knowledge required for naturalization, as well as mandatory oath and ceremony requirements, while the legal component scores the extent to which good character or criminal records are considered for naturalization. Finally, a familial criterion captures whether the state confers citizenship automatically on spouses, or requires spouses to fulfill some degree of naturalization requirements. The indicators on this dimension are generally scored from 0 to indicate the absence of such criteria up to a score of 2, which represents more restrictive stipulations.
A second set of requirements reflects the more latent and ascriptive membership criteria. One's length of residence in the country, as a marker of social integration, was coded as the average number of years in residence required for naturalization, up to a total of two points. Full acceptance of plural citizenship was coded 0 and a ban on dual citizenship coded 2, with gradations for partial toleration. Provisions on jus soli and jus sanguinis measure descent-based criteria, with unconditional jus soli policy coded 0 as being least restrictive and strict jus sanguinis coded 2 as most restrictive. A final ethnic component captures the extent of state accommodations for co-ethnic immigrants from abroad, on a similar 0–2 scale.
For my main dependent variable, I normalized the scores of all components onto a 0 to 1 scale and aggregated them, and then calculated the magnitude of annual policy changes for each government-year. Policy liberalizations thus yield negative values and increased restrictiveness yields positive values. For example, a full policy shift toward tolerating dual citizenship would be reflected by a negative 1.0-unit change compared to the previous year, while an increase in the language proficiency requirement or financial self-sufficiency requirement might result in a positive 0.2-unit change.Footnote 8 Figure 1 plots these aggregated scores for the countries in the sample over the period of analysis.
Figure 1. Citizenship policy scores by country, 1975–2014
The central explanatory variables derive from electoral data from the ParlGov database (Döring and Manow Reference Döring and Manow2015). For the relevant data for my models – relative vote shares, degree of left and right party competition in the electoral system, and the orientation of individual cabinets in a given year – I coded parties on a left–right scale in line with previous literature showing that the this scale remains the most salient dimension of party competition in Europe (Budge et al. Reference Budge2001).Footnote 9 To assign parties along this scale, I largely adopted the recognized party family names from the ParlGov database,Footnote 10 which classifies Conservative or Christian Democratic parties on the right and Labour, Social Democratic and Socialist parties on the left. Communist, Green, and other niche left or new left parties on the far left and populist parties of the far right were designated as Green/Far Left or Radical Right, accordingly (see also March and Mudde Reference March and Mudde2005; Mudde Reference Mudde2007).
Lastly, because elections are held at different times in different countries, and cabinets form and fall sporadically throughout the year, I had to rely on coding rules that aligned election and coalition data with the appropriate government-year. Here I assigned a government-year at time t in which an election was held prior to August in year t with those election results, while I assigned a government-year in which a late-year (post-August) election was held with results from t − 1, and those election results were lagged until the following year at time t + 1. Similarly, a new cabinet was coded upon any change of parties with cabinet membership or any change of prime minister.Footnote 11 I assigned cabinets in office before August to that year, and lagged them for those assuming office after August, unless the coalition dissolved early in that subsequent year, in which case that cabinet is largely excluded from the analysis. Generally, cabinets of less than three months were not included. Caretaker cabinets lasting a majority of months in a given year were assigned as centrist coalitions for that year.
I then used this data to create electoral measures that capture the respective electoral strength of different parties and the degree of competition among parties on either side of the political spectrum. To test Hypotheses 1, 2 and 3, I first tabulated the vote shares and seat shares of the mainstream right-of-center or left-of-center parties in each country that emerged in the post-war period, as well as those for radical right and radical left/green parties in each election. As a second measure of electoral strength, I calculated the averaged difference in vote share from the current election against the previous two elections. I would expect increases in these vote shares on the extremes of the political spectrum to be associated with contagion effects on the policy-making outcomes of respective governments.Footnote 12
To test Hypotheses 1 and 6, I coded several variables to capture the ideological orientation of a government. First, I coded a government comprised of any combination of left, extreme left and centrist parties as Left, any combination of right-of-center, conservative, radical right and centrist parties as Right, and any combination of liberal, centrist, left and conservative parties, or a grand coalition as the Centrist reference category.Footnote 13 In the subsequent logit models, Left is a dichotomous variable scored 1 if a government lacked a right-of-center party, and 0 otherwise; Right is scored similarly as 1 or 0.Footnote 14 To test for the direct effects of extreme parties in government in Hypothesis 6 – perhaps the most direct indication of ideological contagion from within a government – I coded Radical Right in Government and Green/Far Left in Government as 1 if a radical right or left party was in the governing coalition, and 0 otherwise.
To assess the competition hypotheses (Hypotheses 4 and 5), the main measure of right and left party competition is an adaptation of the effective number of parties measure by Laakso and Taagepera (Reference Laakso and Taagepera1979), arguably the most widely used measure in comparative research that represents the number of parties in competition.Footnote 15 Here, I adapt the standard formula 
$(1/\sum Vote_i^2 )$ to calculate an effective number of right parties (ENRP) and an effective number of left parties (ENLP) where Votei refers to the vote share of each right or left party in a particular election, respectively.Footnote 16 As ENRP (ENLP) increases, I would expect more electoral competition on the right (left) side of the political spectrum, and thus an associated increase (decrease) in the citizenship score under the hypothesized conditions.
Finally, I include a number of control variables that might also affect citizenship policy making. First, I control for foreign-born stock as a percentage of the total population drawn from national statistical offices, Eurostat (2015b), the OECD (2015), and the United Nations (2009, 2013), since the size of the foreign population would likely cause policy makers to consider adjustments to citizenship policy. Because the flows of migrants across national borders might also spur citizenship policy making, I include a control for the annual adjusted net migration rate, lagged by one year (Eurostat 2015a). To capture economic conditions in the country that might affect policy makers’ propensity to adjust citizenship policy, I include a variable for annual GDP Growth (World Bank 2016). Finally, a control variable, Liberal, captures whether the country's policy was more liberal and accessible than the sample mean in 1970 compared to the reference category of more restrictive than that year's mean.Footnote 17
Figure 1 illustrates that citizenship policy is quite stable over time and change is infrequent: only 31 government-years experienced a net liberalization in citizenship score and 83 government-years experienced a net restriction coded across the total 640 government-years in the dataset. Consequently, the magnitude of any regression estimates will be predictably small.Footnote 18 In the first series of models I present below in Table 1, I run a pooled ordinary least squares (OLS) regression with heteroskedastic robust standard errors clustered by national election. Clustering standard errors by election cycle controls for possible correlations across different national election cycles and is consistent with the practice of other scholars examining the effects of electoral dynamics on policy change (Somer-Topcu Reference Somer-Topcu2009).Footnote 19 In a second set of analyses in Table 2, I examine the robustness of these findings with a series of logistic regression models using a Firth penalized likelihood estimation approach that addresses potential small-sample bias caused by the rarity of events (King and Zeng Reference King and Zeng2001).
Table 1. Estimated effects on net citizenship policy change (OLS)
Note: standard errors in parentheses. *p < 0.10, **p < 0.05, ***p < 0.01
Table 2. Estimated effects of party competition on policy liberalizations and restrictions (rare events logistic regression)
Note: exponentiated coefficients. Standard errors in parentheses. *p < 0.10, **p < 0.05, ***p < 0.01
Analysis and Results
Table 1 presents the initial set of statistical tests of our hypotheses. Across all models, positive estimates indicate more restrictive change, while negative values indicate more liberalized change. Models 1 and 2 test the initial hypothesis (Hypothesis 1) regarding the effect of government ideological orientation on policy change. The categorical variable for government ideological orientation fails to achieve statistical significance in Model 1, although the estimates are signed in the expected direction compared to centrist coalition governments. When a government's ideological orientation is coded as a continuous weighted average (Cabinet Left/Right Score), the left–right orientation of a government is statistically significant: a one-unit increase in rightward orientation is associated with a 0.02-unit increase in the restrictiveness of citizenship. Although these results represent an improvement over the categorical ideology variable, they suggest that government orientation alone is not the predominant factor behind policy change.
Models 3, 4 and 5 test our next set of hypotheses (Hypotheses 2 and 3) about the potential contagion effects from far right and far left electoral strength. Here we find that, consistent with our hypothesized relationships, the vote share of radical right parties in a given election appears to have a modest restricting effect on citizenship policy change, while radical left party vote has no statistically significant effect on citizenship policy change.Footnote 20 This also seems to be the case in Model 4, where the same general dynamics hold, even if they are not statistically significant. Conversely, the electoral share of votes for a far left or green party is not statistically associated with policy liberalization on its own, although increases in the average change in vote share across elections are a statistically significant predictor of liberalizing change. While this change may also appear modest, we should remember that the policy index has been normalized from 0 to 1, and the mean change in the sample is only 0.02 units in any given year. Thus a strong increase in far left or niche party support from one election to another could account for a sizable liberalization in one component of citizenship policy, such as the weakening of a language requirement. Finally, Models 5 and 8 assess the hypothesized influence of contagion effects from general party competition among parties on the left and right of the political spectrum. In Model 5, while government orientation is statistically insignificant, ENRP is statistically significant, with an estimated effect comparable to that of the Radical Right Vote variable in Model 3. Although the corresponding ENLP variable does not achieve statistical significance,Footnote 21 this ENRP finding offers further confirmatory evidence in favor of our contagion from the right hypothesis (Hypothesis 2). Model 8 reveals that the inclusion of a radical right party in a governing coalition has a robust restrictive effect on citizenship compared to governments without such a party, consistent with Hypothesis 6, but again the commensurate inclusion of a green or far left party in a governing coalition does not have such a statistically observable effect.
Taken together, then, these models imply that, in terms of policy restriction, it is not the electoral strength of right-of-center parties or simply the ideologically right orientation of a government, but rather the electoral presence of radical right parties in the system that accounts for policy change. For policy liberalization, conversely, these results suggest that, regardless of the government in office, it is not the strength of the left, the simple presence of the far left, or the orientation of the government that matters, but the changing electoral prospects of radical left parties over time at the expense of mainstream left-of-center parties that may account for policy change. This represents limited statistical evidence in favor of our hypothesized contagion effects (Hypotheses 2 and 3), independent of the government in office: electoral success on the spatial extremes of the political spectrum may influence government policy making.
In Models 6 and 7, I test the hypothesized effects of my interactive hypotheses (Hypotheses 4 and 5): right-of-center governments should be more likely than other governments to restrict their citizenship policies as the number of right-of-center parties in competition with one another increases, and vice versa for left-of-center governments facing increasing levels of competition on the left. Thus the interaction terms between the competition variables and government type are of particular interest. In Model 6, where Left Governments serves as the reference category, ENRP is not a significant predictor of policy restriction in the case of left-oriented governments, as we might expect. But in the case of both right and centrist governments, however, the effective number of right parties in an election cycle appears to have a robust effect on policy restriction under those governments. This is especially notable since the orientation of the government had no statistically discernible effect on policy change in any of the previous non-interacted models. Likewise, in Model 7, with Right Governments as the reference category, we observe that the effective number of left parties in an election cycle is robustly associated with policy liberalization under left-oriented governments, but not under others.
Figure 2 illustrates these interactive dynamics graphically, revealing the predicted changes in citizenship policy as a result of the ENP interactions as well as the equivalent interaction between governments and radical right and green/far left party vote shares. Again, two top panels show that increases in the effective number of parties on the left of the political spectrum have a statistically significant liberalizing effect on the citizenship policy making of left-of-center governments, whereas increases in the effective number of parties on the right of the political spectrum has the opposite effect on both right-of-center and centrist governments. The lower two graphs also illustrate the contagion effect of the radical right, with restrictive changes associated with both right-of-center and centrist governments at higher levels of radical right party strength. However, the equivalent contagion effects are wholly absent on the left: increases in green or far left vote share have an observed liberalizing effect on policy change under left-of-center governments, but the predicted changes are statistically insignificant at all levels of vote share. Overall, then, these interactions lend strong support to Hypotheses 4 and 5. During election cycles with more left parties competing electorally, a left-of-center government can be expected to enact rather robust policy liberalization compared to non-leftist governments, while at times of greater right party competition – and especially radical right party competition – both right-of-center and centrist governments are associated with greater restrictive policy changes. Thus the restrictive influence of right party competition appears to extend beyond right-of-center governments to centrist coalitions as well.
Figure 2. Predicted citizenship policy changes, by government orientation
In case changes within each category and across categories might not have comparable magnitudes, I next analyze two sets of rare-events logit models in Table 2 with dichotomized dependent variables for any net Restriction (=1, 0 otherwise) or any net Liberalization (=1, 0 otherwise) using penalized maximum-likelihood estimation (Firth Reference Firth1993).Footnote 22 Models 1 and 2 both suggest that the mere presence of the radical right in an electoral system affects the odds of policy restriction, further confirming Hypothesis 2. Holding all else constant, for a one-unit increase in the radical right party vote, we should expect a roughly 7 per cent increase in the odds of policy restriction. Notably, the magnitude of mainstream right party vote share and seat share in parliament has no significant effect on policy restriction; nor does the government's ideological orientation, with one exception.
In Model 3, the inclusion of a radical right party in government increases the odds of restriction nearly fourfold. This is hardly surprising, but it does lend further evidence in support of Hypothesis 6 and suggests that the contagion effect of a radical right party may be especially forceful from within a government. Nonetheless, the general presence of right-wing competition in a system, as captured by the ENRP variable in Model 3, remains statistically significant, implying that a one-unit increase in the ENRP score increases the odds of policy restriction by 55 per cent, with all other variables held at their means.
In terms of policy liberalization, by comparison, we observe in Model 6 that it is mainstream left party strength, not green and far left party strength, that appears to significantly increase the odds of policy liberalization. Moreover, in Model 8, ENLP is not statistically associated with increased odds of liberalization, despite being signed in the expected direction, suggesting that party competition on the left does not exert a contagion influence on citizenship policy independent of the government. That said, whereas a government's ideological orientation has little predicted effect on policy restriction, having a left-of-center governing coalition (Model 8) or having a green or far left party included in that coalition (Model 7) are both associated with robust increases in the odds of policy liberalization. Together, these findings suggest that policy liberalization may occur in response to mainstream left party strength and the inclusion of left and far left parties in government, but not as an independent outcome separate from government, whereas policy restriction may operate through both an extrinsic contagion from the right as well as far right inclusion in government.
Finally, Models 4 and 5 on the restrictive side and Models 9 and 10 on the liberalization side of Table 2 provide strong evidence in favor of Hypotheses 4 and 5. In Model 4, the interaction between ENRP and right-of-center governments does not attain statistical significance, even though the direction of the estimated effect is as expected. This may suggest that the effect of stronger general electoral competition on the right dimension is not necessarily confined by government orientation, but, consistent with Model 3 and the broader findings in Table 1, may instead influence policy restriction across different governments in office. Figure 3, which plots the predicted probabilities of restriction and liberalization for these different interactions, supports this interpretation: increasing ENRP scores yield higher predicted probabilities of restriction for all governments in the sample. For left-of-center governments, the predicted probability across all levels of ENRP remains around 8 per cent. For centrist governments, the total effect is a 15.4 per cent increase in the probability of restriction; for right-of-center governments, the total effect is 28.2 per cent. The same restrictive dynamic applies when government orientation is interacted with the magnitude of the radical right's vote share. On the opposite side of the spectrum, we observe that the interaction between ENLP and left-of-center governments in Model 9 and in Figure 3 results in a sizable and significant effect on policy liberalization. For example, the predicted probability of policy liberalization under a leftist government increases from 2 per cent with only one effective left party in the system to 37 per cent when there are three effective left parties in the system, holding all other variables at their means. Under other governments, the increase in probability across the same range is negligible, a finding confirmed in Figure 3. Finally, the interaction between a left-of-center government and the magnitude of the green/far left vote share in Model 10 in the lower-left panel of Figure 3 is associated with slightly increased odds of liberalization (the predicted probability increases by 18.3 per cent under left-of-center governments), but this predicted effect is overall statistically and substantively weak. This suggests that it is not necessarily far left or green competition that matters for liberalization, but simply an increased number of parties in spatial competition on the left. Taken together, the data support the hypothesized interactive mechanisms (Hypotheses 4 and 5) between a government's ideological orientation and the electoral competition it faces on its side of the political spectrum. Yet the extent to which radical right parties exert restrictive policy effects across all government types also affirms the contagion-from-the-right hypothesis (Hypothesis 2), while the absence of such effects on the left suggests weak evidence in favor of a commensurate contagion-from-the-left hypothesis (Hypothesis 3).
Figure 3. Predicted probabilities of citizenship policy restrictions and liberalizations
Discussion and Conclusion
This article examines citizenship policy making in Europe as a result of the electoral dynamics involving parties on the left vs. right of the political spectrum. I tested a number of competing hypotheses about citizenship policy change as a product of the ideological orientation of a government (Hypotheses 1 and 6) and the increasing electoral strength of far right and far left parties (Hypotheses 2 and 3). I also advanced a theory of citizenship policy change as an outcome of an interactive relationship between party competition at both ends of the political spectrum and a government's corresponding ideological orientation (Hypotheses 4 and 5). The empirical analysis in Tables 1 and 2 and Figures 2 and 3 offers general confirmation of these hypothesized effects. Theoretically, the most important contribution of these findings is that we have robust evidence of a citizenship policy-making mechanism that incorporates both liberalization and restriction within the same model, rather than focusing exclusively on either liberalizations or restrictions as most previous work has done. By examining the interplay between electoral competition and government orientation, the framework advanced here also helps move our focus beyond immigration positions and into the realm of immigration policy making.
Secondly, the findings in this article demonstrate that electoral dynamics on both the right and left affect citizenship policy change, rather than simply the right and radical right, on which previous scholarship has focused almost exclusively. The inclusion of the political left and far left, and their relationship to more accessible policy change, is especially novel. These findings broadly suggest that policy liberalization is associated with mainstream left party strength and the inclusion of far-left parties in government, but these variables alone do not evince a substantive independent effect on policy change that is separate from the influence of government. Rather, the statistical evidence suggests that policy liberalizations are most likely when a left-of-center government faces a permissive electoral context in which multiple parties occupy and compete for space on the left, as captured by the ENLP variable. In such a context, other left parties – albeit not necessarily far left parties – appear to exert a policy influence on like-minded left cabinets, yet not on other governments of a centrist or right-of-center orientation. In other words, the presence of multiple left parties appears sufficient to influence liberalizing change under left-of-center governments, especially with a far-left party in government, but that presence does not extend across the political spectrum. Hence, we might conclude that a commensurate ‘contagion-from-the-left’ effect is not present beyond the mainstream left, a dynamic more akin to a distinct ‘party family effect’ found in previous research (Green-Pedersen and Mortensen Reference Green-Pedersen and Mortensen2015, 748). This is a novel and important contribution to the literature on citizenship policy, and electoral politics more generally.
Conversely, in terms of policy restriction, the presence of the radical right and the electoral competition variables on the right side of the political spectrum were also estimated to have a significant effect on citizenship. The statistical analysis reveals that policy restrictions are not simply associated with right-of-center governments alone; indeed, much of the statistical evidence presented here suggests that right-oriented governments must operate within some context of radical right electoral competition for such an outcome to occur. Thus restrictions appear most likely as a result of either an extrinsic ‘contagion from the right’ independent of the government's ideological orientation and from far right inclusion in government. Moreover, whereas higher left party competition increased the likelihood of liberalization only under left-of-center governments, in times of increasing right party competition – and especially radical right party competition – both right-of-center and centrist governments are associated with a higher likelihood of restrictive policy change, and even left-of-center governments to a lesser degree. In other words, stronger electoral competition on the right is not necessarily mediated exclusively by right-of-center governments, but may instead catalyze policy restrictions regardless of the government in office, a conclusion that dovetails with other ‘contagion from the right’ scholarship (Abou-Chadi and Krause Reference Abou-Chadi and Krause2020; Van Spanje Reference Van Spanje2010).
Altogether, the empirical results suggest that citizenship policy making in Europe operates not simply in response to the presence or absence of radical right parties, but according to some combination of electoral competition from parties on the left and right of the spectrum. While the empirical results presented here generally confirm the theorized dynamics, additional quantitative as well as qualitative case studies could help determine more precisely how these electoral competition mechanisms operate. Nonetheless, this analysis reveals that the far right is not the only force behind citizenship policy change in Europe in recent decades. Indeed, with current trends toward greater party system fragmentation and electoral realignment, considering parties from both ends of the political spectrum seems to be a promising avenue of future research.
Supplementary material
Data replication sets are available in Harvard Dataverse at: https://doi.org/10.7910/DVN/WL77WT and online appendices are available at https://doi.org/10.1017/S0007123420000368.
Acknowledgements
The author would like to thank Terri Givens, Gary Freeman, Robert Moser, Terry Chapman and Rahsaan Maxwell for their helpful comments, as well as the editor and three anonymous reviewers for their insights and constructive feedback.
