the electoral payoff of immigration flows for anti-immigration parties: the case of Italy’s Lega Nord

Abstract

The aim of this article is to examine and quantify the relationship between regular immigration and voting for anti-immigration parties in Italy’s eight Northern regions – from 1992 to 2015 – and in Italy’s forty-five Northern provinces from 2004 to 2015. Firstly, we identified the Lega Nord as Italy’s main xenophobic party, then we used a fixed effects regression model for panel data to test our hypothesis, while controlling for a series of social, political and economic variables. Our results suggest that there is a positive and significant effect: an increase of 1 per cent in the number of new regular immigrants is associated with an increase of more than 2 per cent in the votes given to the Lega Nord; voter-turnout is significantly and negatively associated with Lega Nord’s vote-share, and a high perception of crime – but not actual crime – is positively associated with anti-immigration votes. At the provincial level, we also found a positive association between Muslim foreign population and Lega Nord’s vote-share. These results ought to be considered along with the fact that Northern Italy is highly representative of the wealthiest and most densely populated regions of Europe, suggesting the possibility that similar phenomena could be witnessed in other comparable contexts.

INTRODUCTION

The current international scenario includes a long list of critical areas, many of which are located in the Middle East and Northern Africa (MENA) region and Central Africa. As expected, such political, social, ethnic and religious instability has led to a sharp increase in asylum seekers, the so-called ‘economic migrants’ and regular migrants, the three categories that constitute the migration flows directed towards Italy and Europe, as reported by the United Nations High Commissioner for Refugees (UNHCR Italy, 2015) and the European Union (European Parliament, 2015). The EU’s response has so far fluctuated, being contested both internally and externally, while the current communitarian approach emphasises ‘securitisation over humanitarian responses’ (Massey and Coluccello, 2015). This article focuses on the third group, regular migrants, with the aim of examining and quantifying the relationship between the number of regular immigrants and vote for xenophobic parties in Northern Italy, in a timeframe of slightly more than twenty years, from 1992 to 2015.

In defining Italy’s xenophobic parties, we decided to consider the Lega Nord (Northern League, abbreviated as LN) alone, as not only it is the main Italian political party that has consistently both opposed welcoming asylum seekers and asked for stricter limits for regular immigrants, but also because it is the only current Italian party that can, historically, be defined as xenophobic (Cento Bull, 2010), a position that will be explained in depth in the following sections. With regard to the chosen timeframe, we had to take into account both the Italian political scenario in which the LN was being formed and the availability of data in sufficient quantity to conduct the analysis. Consequently, we decided to choose 1992 as the starting year for this article, because in that year the Lega Nord, officially established one year before, participated for the first time in a general election.

As regards the geographic delimitation of our analysis, lastly, we choose Northern Italy for three reasons: it is the macro-region in which the LN first developed; it is where the majority of regular immigrants are located (Colombo and Sciortino, 2004), and Northern Italy is highly representative of the wealthiest and most densely populated regions of Europe. According to the analyses of a number of political and economic geographers (Brunet, 1989; Hospers, 2003), later confirmed by the OECD, Northern Italy is a structural component of the so-called ‘Blue Banana’, also known as the ‘Industrial Pentagon’ or ‘Corridor 5’, that is a discontinuous and curvilinear corridor which includes European regions sharing important socio-economic characteristics, such as the highest population density and the highest GDP per capita of the continent (OECD, 2006).

In the light of these preliminary considerations, we hypothesised that there could be a positive association between the percentage of regular immigrants over the total resident population of Northern Italy – rising from 1992 to the present day – and the electoral success of the Lega Nord, a political party with an increasing weight both in government coalitions and in the opposition. Firstly, we assessed if Lega Nord’s voters are more inclined to oppose immigration than other voters. Then we used a fixed effects regression model to test our hypothesis of an association between regular immigration and the Lega Nord’s results, both on a regional and a provincial level, while controlling for a series of political, economic and social variables.

This article proceeds as follows: In the next section, a brief literature review is presented; this is followed by the theoretical and analytical framework of this study, a presentation of data and methods, a discussion of the results of our analysis and conclusion.

REVIEW OF THE LITERATURE

There is abundant literature investigating possible explanations for the electoral success of xenophobic parties, which are often represented through the ‘market metaphor’, implying that political parties are in constant competition over their ‘market share’. Accordingly, such explanations are generally categorised in terms of demand (Ramet, 1999) and supply (Mudde, 2010). In the words of Rydgren (2007), ‘the most common family of explanations focuses on the demand-side of politics: factors that have changed the interests, emotions, attitudes, and preferences of West European voters’, whereas supply-side factors are related to xenophobic parties’ political programmes and organisation, national political environments and media activity. Drawing from this dual approach, this section highlights some supply-related issues, while the following section introduces a number of demand-side theories relating to anti-immigration voting.

Two of the main supply-side factors are the message of xenophobic parties and their dealignment and realignment with major parties. The Lega Nord provides interesting insights on both these elements, since its anti-immigration political programme has been supported by the alliance with Berlusconi’s leading centre-right party, resulting in the LN being in government several times. In this respect, the relationship between success of xenophobic parties and changes in immigration law has often been considered a potential concern, due to the influence that anti-immigration parties could exert on a country’s migration policies.

A quantitative analysis by Ackerman (2012) has displayed that such an influence is somewhat mitigated by the difficulties of radical right parties in adapting to public office, while a paper by Perlmutter (2002) has found that there is no sufficient theoretical or empirical evidence to demonstrate that repercussions of this sort have occurred in the first ten years of Lega Nord’s activity in Italian politics. An article by Tarchi (2008), moreover, has concluded that ‘some of the issues which were held dear by the populist electorate were tackled, but in much more moderate terms than suggested by the parties’ manifestos, especially that of the Lega Nord’. It therefore seems that, while this party has been in government several times, it has not been able or willing to fully implement the policies that it advocated during the electoral campaigns. Nevertheless, Lega Nord’s presence in many of the governments led by Berlusconi is the result of a committed political alliance with him, one that allowed a secondary political party to be in charge of a number of ministries over time.

A third supply-related issue is that of a party’s structure and behaviour, strongly linked to what has been mentioned above and concerning, in this case, the reasons for the discrepancy between anti-immigration electoral rhetoric and political outcomes in Italy. It has been pointed out that the country’s fluctuating attitude towards the phenomenon of immigration is not necessarily the by-product of a weak policy apparatus but rather the result of unrealistic political goals, structural internal constraints, fragile party coalitions and contrasting international pressures (Finotelli and Sciortino, 2009). Further, other authors have suggested that while some of Lega Nord’s approved laws are formally in line with the party’s rhetoric, the actual policy outcomes of the government coalitions of which it has been part contradict such rhetoric; hence describing the LN as a case of ‘simulative politics’ (Cento Bull, 2009, 2010) and as a party in which ‘inputs and outputs do not correspond’ (Geddes, 2008).

There is, however, one unambiguous case in which the electoral breakthrough of the Lega Nord has had a strong effect on the political agenda of the Italian government regarding immigration policy, than that of the fourth Berlusconi government of 2008 (Albertazzi and McDonnell, 2010). Not including Italy’s controversial migration policy with Gaddafi’s Libya, whose much-debated and complex consequences (Abbondanza, 2015) are beyond the scope of this work, the influence of the country’s main xenophobic party was profound even on a national level. In the words of Albertazzi and McDonnell (2010: 1326):

Since returning to government, the party has championed a barrage of hard-line measures on immigration such as the reassignment (‘rejections’ of boatloads of mainly African migrants in the run-up to the 2009 EP elections), the census and fingerprinting of some members of the Romany community, making immigrants pay large fees for residence permits and citizenship requests and, finally, the ‘ronde’ (local patrols regarded by many as legalised vigilante groups).

With this conspicuous exception, then, during the years Lega Nord’s policies have been somewhat distant from its political programmes and rhetoric, and hence the description ‘simulative politics’.

A fourth supply-related element that is relevant to the relationship between immigration and party/electoral outcomes is connected to the inner mechanisms of the political systems in which radical right parties operate. It is in this context that some authors have studied the possibility of a ‘contagion effect’ on non-xenophobic parties, due to electoral pressures from anti-immigration parties such as the Lega Nord. A comparative study by van Spanje (2010) shows that such a process actually takes place when the issue of immigration becomes part of a common political agenda, a phenomenon that is further enhanced by the widespread Eurosceptic feelings of recent years (Meijers, 2015). A multi-level analysis of twenty-six European countries also indicates that ‘left-leaning individuals are particularly influenced when parties belonging to the political left raise immigration issues’ (Bohman, 2011), thus potentially favouring xenophobic parties of which the Lega Nord is an example. To add to the complexity of European political systems, Zaslove (2006) argued that in Europe – more specifically in Italy, the object of his case study – the distinction between racist and anti-racist parties, as well as their location in the political spectrum, is not as clear as it is often perceived, thus creating a political dilemma for both voters and political forces.

In conclusion, the supply-side factors that have been presented, contextualised by the existing literature, show some of the possible explanations for the electoral success of the LN. First, they suggest that it might capitalise on the fears of excessive immigration, given that its rhetoric is generally more sharp than the policies it actually implements; second, they show that its alliances have effectively served the purpose of leading the party into government; and third, they imply that the party might have benefited from the increasing importance of immigration on the Italian political agenda. In order to fully understand the possible explanations for the success of the Lega Nord as a xenophobic party, as previously mentioned, the next section introduces the demand-side theories that are relevant to anti-immigration voting.

THEORETICAL EXPLANATIONS FOR THE ELECTORAL SUCCESS OF XENOPHOBIC PARTIES

The electoral success of radical right parties has long been the subject of intense debate among political scientists and sociologists, who have conceived a number of relevant theories. Drawing from the market metaphor, demand-side theories focus on subjective discontent that radical right parties could exploit, such as anti-immigration feelings, economic or political crises, ultra-conservative ideologies and/or perceived insecurity (Roushas, 2014).

Realistic Conflict Theory (RCT) is based on the assumption that the competition over insufficient resources and services results in antagonistic attitudes between different social groups, even when the perception of intra-group competition is stronger than the actual competition (Blalock, 1967). Social Identity Theory (SIT), on the other hand, focuses on the instinctive perception of one’s own social group as superior to other groups (Tajfel, 1982), a psychological feature that could, arguably, be reinforced under competitive conditions such as those of prolonged immigration. The combined effects of these two theories are often referred to as the Ethnic Competition Theory (ECT), the central argument of which is that ‘competition … may reinforce the mechanisms of social (contra-) identification, the eventual outcome of which is referred to as ethnic exclusionism’ (Scheepers et al, 2002). Arguably, this may be the case of Lega Nord’s electoral success, since the party is known for its hostility towards foreign workers who ‘steal jobs from Italians’, a belief that is in line with RCT and that implies, at the same time, that priority should be given to Italian workers, ‘our own people’, who are experiencing ‘counter-racism’ in an open labour market (Fella and Ruzza, 2009), thus fitting the principles of SIT and, more generally, those of Ethnic Competition Theory.

A fourth theoretical tenet relating to the study of radical right parties is that of the Defended Neighbourhood Theory (DNT), based on the core assumption that the arrival of new social groups, differing from the dominant group, prompts a violent response from the latter against the former. Unlike RCT, however, the Defended Neighbourhood model does not infer that limited resources are the cause for social conflict, which are to be found instead in the threat to a local common social identity (Green et al, 1998). Another approach is that of the Relative Deprivation Theory (RDT), which focuses on the frustration resulting from feelings of ‘relative deprivation’, derived from a negative comparison with a dominant social group (Runciman, 1966). A sixth theoretical strand is represented by the Modernisation Losers Theory (MLT), which posits that the views on traditional societies upheld by radical right parties might appeal to those who do not accept or cannot cope with the social changes associated with modernisation (Minkenberg, 2003).

There is a certain degree of intersection between some of the demand-side theories, a condition that can be noticed between the previously mentioned Social Identity Theory and the Defended Neighbourhood Theory and between the Realistic Conflict Theory and the Relative Deprivation Theory. These could, at least in part, help explain how xenophobia has supported the LN from an electoral point of view, given that ‘the outsider is perceived as an undesirable figure representing a perpetual threat to Padania’¹ (Huysseune, 2006), in line with DNT. Moreover, Lega Nord’s rhetoric has long targeted outsiders, on the grounds that ‘discrimination exists in favour of immigrants’ and also because it is often stated that ‘any immigrant has higher points than one of our families’ in social-housing lists (Fella and Ruzza, 2009). With regard to Modernisation Losers Theory, however, there is no sufficient evidence to suggest that those who vote for the Lega Nord are citizens who cannot cope with the changes of modern society; as Gianfranco Miglio, Lega Nord’s most valued intellectual, was the first to note, the party possesses a very diverse social base, composed by ‘workers, artisans, businessmen, small and medium entrepreneurs, professionals, teachers, intellectuals’ (Kurth and Petras, 1993), thus making it difficult to claim that MLT alone is able to explain the success of the Lega Nord as a xenophobic party, without further research to substantiate this.

Two elements are clearly defined by this review: first, theoretical approaches relating to radical right parties focus on the exclusionary social patterns that relate to them; second, social marginalisation appears to be the core explanation for the emergence (or resurgence) of such parties (Rydgren and Ruth, 2011). According to what has been presented, we believe that the sharp increase in the amount of regular immigration to Italy in the past twenty years, along with the growing role of the LN as both a government and an opposition party and the rise of xenophobic parties elsewhere in Europe, all fit properly into such a theoretical context, leading us to hypothesise that there could be a degree of convergence between the levels of immigration and anti-immigration voting in Northern Italy.

Although a paper by Dinas and van Spanje (2011) reports that ‘some scholars have found that mass immigration fuels the success of anti-immigration parties, whereas others have found that it does not’, there are specific studies that support our assumption. While Mudde’s (1999) tepid conclusion is that ‘at best, immigration has been a catalyst for most ERPs in certain periods of time’, the psychological threat of immigration has been identified and described as a ‘halo effect’ of French communities living close to large groups of foreigners, resulting in a larger vote-share for the anti-immigration party Front National in those communities (Perrineau, 1985). Further, a more recent article has found a more direct and less symbolic link between immigration and xenophobic vote (Arzheimer, 2009), thus potentially supporting our hypotheses.

The existing literature is therefore rich and diverse; however, it does not unequivocally clarify whether a relationship between immigration and anti-immigration voting actually exists. Drawing from this, the next sections present the analytical framework of this study, along with data, methods and the results of the analysis employed to evaluate such a relationship.

ANALYTICAL FRAMEWORK

Starting from these premises, this article aims to investigate and quantify the relation between the percentage of regular immigrants and its demographic components (gender and religion) over the total resident population of Northern Italy and the electoral results of the Lega Nord, from 1992 to 2015, in Northern Italy. We assumed that a positive association could exist, on the basis that such a party might capitalise on the increasing immigration flow in those regions by underlining the phenomenon in electoral campaigns. Thus, our hypothesis is that the LN exploits anxieties relating to safety and security which immigration might contribute to strengthen, in order to obtain a higher electoral consensus. A socially conservative electorate, in turn, might look favourably at a party whose rhetoric opposes the socio-economic changes brought by immigrants. The existing literature outlined above shows that immigration can be perceived as a threat on at least three levels: culturally, by introducing foreign habits; socially, by altering the ethnic and religious fabric and/or increasing criminal behaviours; and economically, by saturating the labour market.

The electoral programmes of the Lega Nord have systematically related immigration to crime and the additional economic costs sustained by Italians and have implied a degradation of cultural and religious identity. A clear example is that of the 1994 programme, whose section titled ‘Jobs and extra-communitarian immigration’, stated that ‘immigrants are often prone to criminality’, ‘immigrants are favoured over Italians by social policies’ and that ‘a multiracial mentality’ causes ‘a loss of the sense of belonging to the ethnic, cultural and religious identity of the Italian people’ (Volkens et al, 2016). We therefore expect to find that the LN – in line with other anti-immigration parties – benefits the most from an increase in male Muslim immigration, which might be seen as threatening on all three levels.

In doing so, we first decided to assess if Lega Nord’s voters could be defined as more inclined to oppose immigration compared to other voters, which involved data collected in five electoral surveys conducted in Italy by the ITANES (Italian National Election Studies, see ‘Results’ section). Having concluded that the Lega Nord is, among Italian parties, the one that is most consistently focused on immigration, we tested our hypothesis.

When dealing with panel data, we avoided treating all unit-periods as independent and equally distributed random variables and decided to pool them. Given the peculiar historical and social background of each Italian region – which is mostly visible in strong differences in voting behaviour – we assumed a time-constant unit effect that remained unobserved and, consequently, subjected the data to a within transformation² to eliminate the unobserved effect from the equation (Wooldridge, 2010). All data ranging from 1992 to 2015 were collected in standardised study spreadsheets and entered into Excel 2010 (Microsoft Corporation, Redmond, Washington), while the fixed effects regression was done with R (R Development Core Team, Vienna, Austria) with the percentage of votes for the LN as the dependent variable and the percentage of regular immigrants in Northern Italy as the main independent variable. Nationality and gender data of the immigrating population were also used for the analysis at the provincial level, for the 2004–2015 timeframe.

In order to better implement our analysis, a number of relevant control variables – which we assume might have influenced voting – have been included as well: voter-turnout; per capita GDP; whether the Lega Nord is in the regional government coalition or not; unemployment rate; perceived crime rate and the officially reported number of robberies, thefts and homicides per 100,000 people; and the percentage of population over sixty-five years of age. These variables have been selected due to the fact that such socio-economic issues are found to be closely related to anti-immigration feelings, as shown by a well-known cross-country investigation of individual attitudes towards immigration (Mayda, 2004). These control variables are also in line with the theoretical explanations for xenophobic voting outlined in the previous sections, given that two of the former take into account potential reasons for economic discontent and four contemplate factors that effectively represent perceived social threat and competition. With regard to the LN being in the regional government coalition or not, moreover, we assumed that the party’s presence in the regional government could have influenced voting patterns due to higher media exposure, while the choice of considering only its presence in regional (not national) governments was dictated by the structure of our model, which could not employ a variable that remained fixed for all the units involved in the analysis.

Electoral data came from the official publications of the Italian Ministry of the Interior,³ survey data employed to assess the Lega Nord’s position on immigration in relation to other parties came from the ITANES and all data relating to both the independent variable and the control variables came from the ISTAT.⁴ In precisely outlining Northern Italy, we adhered to the official definitions of both Italy’s national statistical institute (ISTAT) and the EU, which include eight regions (Aosta Valley (Valle d’Aosta); Piedmont (Piemonte); Liguria; Lombardy (Lombardia); Emilia-Romagna; Veneto; Friuli-Venezia Giulia; and Trentino-Alto Adige) and forty-five provinces (Figure 1).

Figure 1

Northern Italy’s 8 regions and 45 provinces within the country. Key: 1 Aosta Valley, 2 Piedmont, 3 Liguria, 4 Lombardy, 5 Emilia-Romagna, 6 Trentino-Alto Adige, 7 Veneto and 8 Friuli-Venezia Giulia.

DATA AND METHODS

We used data collected in five electoral surveys conducted in Italy by the ITANES⁵ after the general elections of 1996, 2001, 2006, 2008 and 2013 (Tables 1, 2, 3, 4, 5). The responses to the electoral surveys are used to assess the position of Lega Nord’s voters. Whether or not anti-immigration feelings are deeply rooted in the Italian electorate, does the Lega Nord intercept these relatively more than other parties? Can the LN more effectively capitalise on the worsening of anti-immigration feelings, from an electoral point of view? Although electoral surveys were conducted after every general election since 1968, opinion towards immigration started to be tested only in 1996, once again showing the increasing scale of this phenomenon since the mid-1990s. Despite the fact that questions on immigration were not phrased consistently over time (see Appendix for survey questions), our interest is not focused on tracking voters’ feelings across time but rather in estimating the relative importance of the issue to voters of different parties in the same election.

Table 1 Importance of Immigration Issue in 1996

Table 2 Importance of Immigration Issue in 2001

Table 3 Importance of Immigration Issue in 2006

Table 4 Importance of Immigration Issue in 2008

Table 5 Importance of Immigration Issue in 2013

We operationalised the assessment of the opinion of Lega Nord’s voters towards immigration with two research questions: first, among centre-right and right-wing voters, are voters of the Lega Nord significantly more inclined to oppose immigration? Second, based on the opinion expressed by its voters, is the Lega Nord the party with the strongest feelings against immigration in the country? The first research question is assessed with a Chi-squared test for association between two binary variables: votes for the LN and immigration as the first- or second-most important problem (for years 1996 and 2001), while, for the other years, a difference of means test (t test) between a categorical variable (vote to the Lega Nord) and a continuous variable (intensity of the opinion) was employed. On the other hand, the position of Lega Nord’s voters relative to the position of voters for other parties is assessed by ranking each party, according to the percentage of voters indicating immigration as the most important or second-most important issue in 1996 and 2001, while, in 2006, 2008 and 2013 such a position was based on the average answer on the respective continuous variable.

In order to estimate the effect of the variation in non-Italian resident population on Lega Nord’s electoral trajectory, we considered two core variables: votes _rt , the votes the LN received as percentage of the total number of valid votes, and foreignPop _rt , the percentage of non-Italian resident population over total resident population,⁶ where r are the eight regions of Northern Italy and t are the election years (1992, 1993, 1994, 1996, 1998, 2000, 2001, 2004, 2005, 2006, 2008, 2009, 2010, 2013). Based on the data, we made an unbalanced⁷ data panel. The years that we have chosen for the analysis are those in which an election (general, regional or European) took place in these eight Italian regions between 1992 and 2013. Table 6 shows the votes obtained by the Lega Nord as a percentage of valid votes cast in the same region (votes _rt ).

Table 6 Electoral Results of the Lega Nord

Data on non-Italian residents in each region, measured as a percentage of the total resident population of the region (foreignPop _rt ), is based on regional figures published by ISTAT on its website³ for the years 1981, 1991 and from 2001 to 2014. In each region, the percentage of the non-Italian population in the missing years between 1991 and 2001 was approximated with interpolated values based on a constant growth estimate (Figure 2).

Figure 2

Estimate of trend of regular immigration in Northern Italy, as percentage of total population.

We controlled for eight additional variables that we assume might have influenced voting. Voter turnout (tournout _rt ) captures the level of participation in the election while a binary variable, LegaGoverningRegion _rt , indicates whether the Lega Nord was in the government of one specific region (alone or in coalition with other parties) at the time of the election. To take into account the evolution of the economic environment in the different regions, we controlled for the GDP per person (procapiteGDP _rt )⁸ and the unemployment rate (unemp _rt ). Lastly, we controlled for the effect of crime (with official and perceived crime data) on voting behaviour by adding four variables indicating: the number of robberies per 100,000 residents (robberies _rt ); the number of house thefts per 100,000 residents (thefts _rt ); the number of homicides per 100,000 residents (homicides _rt ) and a variable indicating the percentage of families declaring that the ‘criminality risk in their area is very or quite present’, as measured by a recurrent survey conducted by the ISTAT (crimePerception _rt ).⁹

Due to the historical peculiarities of each Italian region, which are reflected not only in visible economic, social and cultural differences but also in electoral patterns, we assumed an unobserved fixed effect for each region in the panel, which is constant in time. We proceeded to formalise a linear unobserved effects model:

$$votes_{rt} = foreingPop_{rt} \beta_{1} + tornout_{rt} \beta_{2} + LegaGoverningRegion_{rt} \beta_{3} + procapiteGDP_{rt} \beta_{4} + unemp_{rt} \beta_{5} + thefts_{rt} \beta_{6} + robberies_{rt} \beta_{7} + thefts_{rt} \beta_{8} + crimePerception_{rt} \beta_{9} + c_{r} + u_{rt} ,$$

where c _r is an unobserved random variable for the region r that we assume, for historical reasons, constant in time. Given our panel data, we established the fixed effects regression model with the plm package for R (R Core Team, 2015; Croissant and Millo, 2008).

To check the results obtained from the analysis of the 1992–2013 timeframe, we applied a very similar fixed effect model to immigration data at the provincial level (available only for the 2004–2015 timeframe):

$$votes_{pt} = voters_{pt} + turnout_{pt} + unemp_{pt} + over65Pop_{pt} + neighForeignPop_{pt} + c_{t} + u_{pt} ,$$

where over65Pop is the population over 65 in the province p at the time t and neighForeignPop is the percentage of foreign population in the province p and its neighbouring provinces over the number of eligible voters in that province. The choice to include data on neighbouring provinces is justified by the fact that an Italian province is a relatively small unit so, arguably, voters have a high probability of moving across provincial boundaries in the course of their routines. Since the data at the provincial level included information on the nationality and gender of the immigrating population as well, we were also able to add two gender variables (neighMaleForeignPop _pt and neighFemaleForeignPop _pt ) and a religion variable (neighMuslimForeignPop _pt ) on Muslim immigration, assuming that the population emigrating from any country would present the same percentage of Muslims as found in the population of the country of origin (Tables 7, 8, 9, 10).

Table 7 Coefficients and Significance of Variables in the Fixed Effect and Pooled Models (1992–2013, Regional Level)

Table 8 Coefficients and Significance of Variables in the Fixed Effect Models (2004–2015, Provincial Level) – with Muslim Foreign Population

Table 9 Coefficients and Significance of Variables in the Fixed Effect Models (2004–2015, Provincial Level) – with Muslim Foreign Population and Region of Provenance and Gender of Foreign Residents

Table 10 Coefficients and Significance of Variables in the Fixed Effect Models (2004-2015, Provincial Level) – with Muslim a Non-Muslim Foreign Population in Neighbouring Provinces

RESULTS

Based on the ITANES electoral survey conducted in 1996 and 2001, respectively, 25.48 and 27.78 per cent of Lega Nord’s voters considered immigration as the most or second-most important issue, in comparison to, respectively, 13.02 and 13.57 per cent of voters of other centre-right and right-wing parties. In 2006, the opinion of Lega Nord’s voters was still significantly imbued with anti-immigration sentiment – when compared to that of voters of other centre-right and right-wing parties – but not in 2008 and 2013, years of high irregular immigration in which such a difference disappears. To summarise, Lega Nord’s voters appear to consider immigration as a fundamental issue and/or have anti-immigration views more frequently than any other centre-right and right-wing voters in three out of five surveyed years.

More importantly, however, if we exclude from the analysis of the electoral survey data all parties with less than ten declared voters, we observe that, based on its voters’ opinions, the LN is the strongest party to have anti-immigration views in four out of five years and that, in 2008 and 2013, its position was very similar to that of Silvio Berlusconi’s Popolo della Libertá (see Figure 3 and tables in Appendix). Based on surveys conducted over seventeen years, we can conclude that the Lega Nord is the Italian political party that is most consistently focused on immigration. Therefore, an increase in anti-immigration feelings among the country’s electorate, potentially triggered by an increase in the non-Italian population, might benefit the Lega Nord from an electoral point of view more than other centre-right and right-wing parties.

Figure 3

Relative opposition to immigration, based on voters’ opinions in ITANES surveys.

Moreover, the fixed effects regression model on the panel data of Northern Italy’s eight regions (see Appendix for the regression table) provides evidence for a positive, strong and significant effect of the inflow of new regular immigrants on the votes received by the LN. The model estimates that, after having controlled for the unobserved but constant characteristics of each region and for the various observed political, economic and crime variables, an increase of 1 per cent in the number of non-Italian regular residents results in an increase of more than 2 per cent in the votes given to the Lega Nord. That is, an increase of one percentage point in regular immigration results in an increase of more than twice that in Lega Nord’s votes, a condition that is consistent with our theory of immigration as a significant contributing factor to xenophobic voting in Northern Italy.

Interestingly, the effect of regular immigration on voting for the Lega Nord is insignificant when the panel data are pooled, that is, when the peculiarities of each region, constant in time, are not taken into account. This is also clear by charting the evolution of voting and regular immigration flows for the eight regions (see Figure 4). Of the three regions with the highest percentage of non-Italian regular residents, Emilia-Romagna, for historical reasons, has always shown a preference for left-wing parties (Sengenberger and Campbell, 1994), and thus the LN was not able to capitalise on the increase in numbers of immigrants as fully as it probably did in Veneto and Lombardy. This could be due to the socio-political peculiarities of Italian regions, each with a deeply rooted and autonomous history, which results in well-documented political preferences, such as those for left-wing parties in Emilia-Romagna and right-wing parties in Veneto and Lombardy, autonomy movements in the Aosta Valley and minority parties in Trentino-Alto Adige and Friuli-Venezia Giulia (Spektorowski, 2003). Such historical characteristics might have influenced the electoral patterns of those regions, which underlines the importance of carefully considering them in terms of the structure of the regression model (Figures 5, 6, 7).

Figure 4

Votes for Lega Nord and number of regular immigrants as percentage of total resident population in Northern Italy.

Figure 5

Percentage of female immigrants from different macro-regions of the world.

Figure 6

Scatterplot matrix, provincial level.

Figure 7

Scatterplot matrix, regional level.

Control variables provided other interesting estimates. Voter-turnout is significantly and negatively associated with Lega Nord’s votes. Such data might be explained by Apaydin’s (2016) study, whose findings confirm that loyalty to a supranational entity (EU) enhances citizen participation in national elections and, perhaps in this case, vice versa. Another interpretation is that some of the votes for the Lega Nord are protest votes in ‘minor’ elections (European Parliament, regional). A third interpretation could be drawn from a recent study by Immerzeel and Pickup (2015), who reported that ‘the Western European results are consistent with theories that it is those who strongly oppose populist radical right parties that will be motivated to turnout as a result of the emergence of a successful PRR’, potentially suggesting an opposite effect in this case, a condition that would require further research to be substantiated.

Not surprisingly, in regions that are governed by the Lega Nord, the party is estimated to receive 7 per cent more votes compared to other regions, thus indicating strong local support for the party. Crime variables indicate a significant association between the perception of crime and votes for the LN, along with a non-significant association between those votes and the number of thefts, robberies or homicides. Since actual crime – specifically, thefts and robberies but not homicides, which are quite exceptional events in Italy – does significantly correlate with the perception of crime,¹⁰ we also modelled factual crime variables without including the perception of crime variable (see Appendix); however, the effect of crime variables on Lega Nord’s votes remained non-significant. This strongly suggests that the Lega Nord capitalised on the perception of crime-related insecurity but not on actual crime.

This could be due to the fact that, to some centre-right and right-wing voters, the perception of immigration and crime appear to pose ‘a threat to the security, identity and economic wellbeing of the North’, thus resulting in an electoral benefit for the political party that has been the most consistently focused on anti-immigration policies (Zaslove, 2011). In numbers, according to our model, an increase of 4 per cent in the number of families that feel unsafe will result in an increase of 1.1 per cent in votes for the Lega Nord. The findings relating to the regional level could therefore be well contextualised in the current research on rising anti-party and anti-immigration populism (Brading, 2015), as well as in the debate of populist and radical right parties and their role in centre-right coalitions, an ongoing subject of study in European politics, due to the resurgence of the latter and the increasing relevance of the former (Akkerman et al, 2016).

The results from the fixed effect regression at the provincial level support the results of the analysis at the regional level, although the inferring power is tempered by the shorter time span of the data currently available. Moreover, the provincial analysis provides a more detailed picture on the association between votes for the LN and the characteristics of immigration in Northern Italy. Having controlled for Muslim and non-Muslim immigrating population, it appears that only Muslim immigration (slightly more than a third of the total immigrating population) is positively associated with Lega Nord’s votes, while non-Muslim immigration had no effect.

This association is consistent with the party’s political rhetoric, which over time has noticeably targeted Muslim immigration, as well as publicly opposing the construction of new mosques. However, there could be a different explanation: global immigration from Muslim regions is predominantly a male immigration. This can be perceived as more threatening on a cultural and demographic level, as well as from a perspective of public order and labour market (Scheepers et al, 2002). On the other hand, immigration from Eastern Europe, by far the main source of immigration in Northern Italy, mainly consists of female workers who are generally employed as housekeepers or carers for the elderly, and who, in turn, are generally not perceived as competitors in the labour market.

Conclusion

The aim of this article was to test our hypothesis of a potential positive association between the percentage of regular immigrants in Northern Italy and the electoral success of the Lega Nord. Firstly, we identified the latter as Italy’s main xenophobic party, by assessing that Lega Nord’s voters are consistently more inclined to oppose immigration than other voters. Then we used a fixed effects regression model to test our hypothesis at both the regional and the provincial level, while controlling for a series of political, economic and social variables that we assumed might have influenced voting.

The results of our analyses¹¹suggest that there is a positive and significant effect of the number of new regular immigrants on the votes received by the LN. Specifically, an increase of 1 per cent in the number of non-Italian regular residents results in an increase of more than 2 per cent in votes given to the Lega Nord, while voter-turnout is significantly and negatively associated with Lega Nord’s votes. Other relevant estimates show that, in Northern Italy, a high perception of crime – but not actual crime – is positively associated with votes for the Lega Nord, thus showing both the relevance of crime in the xenophobic rhetoric of the LN and the importance of crime perception in anti-immigration voting patterns in Northern Italy. At the provincial level, moreover, we also found a positive association between Muslim foreign population and Lega Nord’s votes.

These findings are consistent with the majority of common explanations for xenophobic voting, particularly those that refer to the demand-side of politics. Realistic Conflict Theory and Relative Deprivation Theory appear to be corroborated by the perceived threat of immigrants within the labour market, the priorities of social welfare and the economy in general, as outlined previously, while the principles of Social Identity Theory and Defended Neighbourhood Theory seem to be substantiated by hostility to the ‘otherness’ represented by the immigrant, a condition that encompasses social and identity characteristics. The analysis at the provincial level, moreover, suggests that Ethnic Competition Theory could be an appropriate explanation for the electoral success of the Lega Nord, as mechanisms of ethnic (overlapping religious, in this case) exclusionism have been triggered by an increase in the specifically Muslim foreign population. Modernisation Losers Theory, however, does not appear to be a suitable explanatory model, given that the social base of Lega Nord’s electorate is so diverse that its composition could not justify the lack of adaptation to modernity as a stand-alone reason.

In conclusion, this article provides evidence for a positive correlation between the number of regular immigrants and voting for anti-immigration parties, in a timeframe of slightly more than 20 years, in Northern Italy. Along with these findings is a note of caution. There is always the possibility of some inherent limitations to this study, such as potential mistakes in the conceptualisation of our research, insincere answers in the ITANES surveys or excessive confidence in the considerable amount of data included in our calculations. However, we conducted our analysis with the greatest attention to variables that could have influenced its results and with the aim of avoiding any miscalculation, a process that led us to the conclusion that this article could be a useful tool for the study of the relationship between immigration and anti-immigration vote. Moreover, the fact that Northern Italy is highly representative of the wealthiest and most densely populated regions of Europe, suggests the possibility that similar results could be witnessed in other comparable contexts, a prospect that inevitably calls for further research.

Notes

• 1 In political terms, Padania is the name of a proposed country in Northern Italy and parts of Central Italy, including Northern Italy’s eight regions plus Tuscany, Umbria and Marche. Padania’s secession from Italy has long been a central theme of Lega Nord’s political programme.

• 2 It is a specific econometric method to control for the fact that different units might have time-invariant characteristics that are unobservable. The within transformation eliminates the unobserved effect.

• 3 Data available on: http://elezionistorico.interno.it.

• 4 Data available on: www.istat.it/en.

• 5 Data are freely accessible on: www.itanes.org/dati.

• 6 Data on non-Italian residents were compiled from different sources: for the years 1981, 1991 and 2001 figures were obtained from results of the national census, for the period 2002–13 from annual probes conducted by ISTAT.

• 7 Each region followed a different election cycle.

• 8 Regional data on per capita GDP preceding the year 1995 were not present on the ISTAT website at the time of this study. Data were accessed via a cached version of the site.

• 9 In 2004, the survey was not conducted. The missing values for that year are interpolated from the value of the previous and following years.

• 10 An increase of one robbery for every 100,000 people results in an increase in the number of families worried about crime of 0.24 percentage points (0.22 when controlling for regional fixed effects) and an increase of 100 thefts in an increase of 0.4 percentage points (0.3 with fixed effects).

• 11 The research material for this article can be accessed at http://dx.doi.org/10.7910/DVN/YM7IWP .

APPENDIX

ITANES survey questions

• 1996: Is immigration the most or the second-most serious problem in Italy? (binary).

• 2001: Do you consider immigration the most or the second-most important problem in Italy? (binary).

• 2006: How much do you agree with the opinion that immigrants are a threat to our culture and our identity? (1 to 4 scale).

• 2008: How much do you agree with the opinion that immigrants are a threat to our culture? (1 to 4 scale).

• 2013: How much trust do you have in immigrants? (1 to 4 scale).