Party System Change in Italy: Politicising the EU and the Rise of Eccentric Parties

Abstract

Using expert survey data covering the 2001–13 period, this article investigates the changing shape and structure of the Italian policy space over time, as well as parties’ shifting policy positions. Our results show the emergence of a distinct pro-/anti-EU (European Union) dimension structuring party competition and a change in the meaning of EU related attitudes. The increasing importance of the pro-/anti-EU dimension is due to the entry of brand new parties in 2013. Our findings also speak to the Euroscepticism literature, as they question the hypothesis according to which Euroscepticism is confined to peripheral parties located at the extremes on the left–right scale.

Keywords:

• Policy space
• party system change
• expert surveys
• EU dimension
• Euroscepticism
• Italy

Party system change has long been one of the most studied topics within the comparative politics literature. Since the seminal work of Sartori (1976, p. 44), a party system has been conceptualised as ‘the system of interactions resulting from interparty competition’. According to this definition, party systems can be distinguished by looking at specific features of the relationships between political parties and at the pattern of party competition. These features include the number of relevant parties, parties’ relative size and strength, the key policy dimensions shaping the space of political competition, the matrix of distances between parties’ ideal positions, the patterns of government formation, and parties’ ability to cooperate with each other after elections (Wolinetz 2006).

In an attempt to capture significant differences among the most stable liberal democracies, comparative scholars have built classifications and typologies of party systems focusing on one or more of the relational features described above: counting the number of relevant parties (Duverger 1954); combining the number of parties with information about their relative size and strength (Blondel 1968; Siaroff 2000); looking at patterns of government formation (Dahl 1966; Rokkan 1970; Mair 1996, 2002); or examining both the number of parties and their ideological polarisation (Sartori 1976). However, while these criteria have proven useful in categorising party systems ‘into distinct classes or types, such as two-party systems, systems of moderate pluralism, multiparty systems, or whatever’ (Mair 2006, p. 63), they are quite ineffective in capturing party system change over time. Only on rare occasions (e.g. Italy after 1993) have changes in the defining features of party systems led to a reclassification (Ieraci 2006; Morlino 1996; Pasquino 2009). This failure to effectively measure party system change has had the effect of biasing analyses in favour of stability (Mair 1997, 2006).

In this work we address the study of party systems and party system change from the perspective of the spatial approach to elections and party competition. The spatial approach is not meant to provide classification schemes. Rather it introduces a set of parameters capturing changes both in the structure of the party competition space and in parties’ policy positions on the main dimensions defining that space. Within such an approach, party system change relates to the changing shape and structure of the policy space over time, which eventually define the nature of parties’ mutual interactions. Change is characterised in terms of two features: the relative salience of key policy dimensions and the relationships between party positions on these different dimensions.

In the early 1990s Italy adopted a new electoral system and this was followed by a partial renovation of Italian politics.¹ Most of the parties that had marked Italian post-war political life disappeared. In the last two decades several fusions and fissions have resulted in a shift in the identity, size, and organisation of parties. Moreover, brand new parties emerged and obtained considerable electoral success (Cotta & Verzichelli 2007). Have these developments altered in a significant way the spatial properties of the Italian party system? How is such evolution related to the changing shape and structure of the party system over time? We aim to answer these questions by examining changes that occurred in Italy between 2001 and 2013 with regard to (1) the salience of the dimensions structuring the space of political competition, (2) the number of fundamental axes that can be used to describe party competition, and (3) shifts in the policy positions of the most significant political parties.

We investigate party system change in Italy by means of consistent expert survey data for each national election that occurred in the period under observation, collected using a common format developed by Benoit and Laver (2006). We follow the approach adopted by Laver and Benoit (2007), who measured changes in spatial attributes in a set of West European countries using data on party positions from two directly comparable expert surveys.² In so doing, we provide not only a single country case study of party system change, but also a basis for comparative analysis. Our results highlight the emergence of a distinct pro-/anti-EU (European Union) dimension. In addition, they show that the meaning of EU-related issues has shifted from economic and immigration policies to more basic pro-/anti-EU attitudes. Such a change cannot be attributed to policy shifts of existing parties, but can be attributed to the entry of brand new parties such as the Five Star Movement (Movimento cinque stelle – M5S) into the Italian party system in 2013. These results are consistent with previous studies conducted with different methodologies and highlighting, on the one hand, the increasing importance of EU issues in defining the Italian policy space and, on the other hand, the role of new parties in structuring political competition in Italy (Conti & Memoli 2010; Conti 2014). Moreover, our analysis suggests that Euroscepticism in Italy does not seem confined to peripheral parties located at the extremes on the left–right scale. These findings question Hooghe, Marks and Wilson’s (2002) hypothesis (inverted U-curve) that has been central to the way the literature has looked at the role of Euroscepticism in party systems over the past decade.

The argument presented in this article is structured as follows. In the next section we briefly introduce the spatial approach to party competition. Section three illustrates the expert survey methodology, assessing the quality of measurement across issues and parties. Section four describes the data and the research strategy adopted in our study, discussing how to make feasible a comparison of the positions of Italian parties across the 2001–13 period. In the subsequent two sections, we present evidence about party system change in Italy between 2001 and 2013. In particular, section five examines changes in the first of the abovementioned features underpinning the spatial approach, i.e. the structure of policy space. Section six deals with the second feature, highlighting changes in parties’ policy positions along the main axes of political competition. Concluding remarks follow in the final section.

The spatial approach to party competition

The spatial approach to party competition relies on the assumption that the preferences and choices of political actors take place in some kind of policy space. The dimensions defining such a space derive from actors’ preferences on clusters of related issues. For instance, if preferences over issues dealing with tax levels, workers’ rights, and public expenditure are highly correlated, we may think of these issues as being part of an underlying economic policy dimension. On each policy dimension, political actors have different locations based on their preferences on related issues. For example, a political party advocating a greater role for the state in the economy can be considered to hold a left-wing position on the economic dimension, while a party favouring a limited state intervention in the economy can be considered as right-wing on the same dimension. Analysing party competition using the spatial approach requires assessing empirically the distances between party positions as well as their movement in the policy space. For this purpose, an overarching left–right dimension has been used to map the positions and movements of political parties in Western democracies (Castles & Mair 1984). Moreover, two-dimensional maps of policy spaces have been estimated for most European countries by Benoit and Laver (2006).

The spatial approach can improve our understanding of party system change, as it enables us to assess (a) how the parameters defining the structure of the policy space change over time, and (b) how the policy positions of political actors evolve. This article considers both aspects of the Italian party system in the last 15 years. Changes may arise from exogenous events such as policy and agenda shocks, or endogenously as an output of party competition.

In the first case, the distribution of party positions on major policy dimensions can be altered by some external policy shock. Parties in fact can be conceived of as agents of their constituencies: when events influencing markets, international relations, and many other aspects of social, economic, and political life modify constituency preferences, then parties’ ideal points should change accordingly (Lupia & Strøm 1995). In a similar way, agenda shocks may change policy programmes: even though they do not affect parties’ policy preferences, such shocks can influence the relative salience of key policy issues through changing the structure of the policy space within which all parties interact (Laver & Shepsle 1998).

When party system change is endogenous, modifications may result from actions undertaken by politicians themselves. On the one hand, parties may adjust policy positions in response to changed incentives faced by their leadership or as a reaction to a strategic repositioning of one of their main competitors (Enelow & Hinich 1984). On the other hand, party positions and the salience of issue dimensions may shift following the entry of new actors into the electoral arena (Meguid 2005). Finally, the dimensions structuring the policy space may be strategically manipulated by political actors in order to influence political outcomes (Riker 1982).

Expert survey methodology

Testing spatial models requires accurate estimates of the policy positions of political actors. Several techniques can be used to measure party positions. Some of them rely on mass surveys, while others use elite or expert surveys (Mair 2001). One prominent source of data is the Manifesto Research Group/Comparative Manifestos Project (MRG/CMP), which has been manually coding the electoral platforms of parties in 56 countries since 1945 (Budge et al. 2001). Other estimates are based on data derived from computerised content analysis of party manifestos (Laver, Benoit & Garry 2003). Finally, data about legislative voting behaviour have been used to infer party positions, especially in the United States (US) context (Poole & Rosenthal 1997). All of the abovementioned techniques have strengths and weaknesses, and it is therefore necessary to make a trade-off depending on the research question being addressed. For a critical review of different techniques see Benoit and Laver (2006).

In this paper, we examine party system change in Italy using expert survey data. The expert survey methodology adopts an a priori approach: policy dimensions or scales are predefined and parties are given a position on these scales by country experts. The scores used to estimate party positions are therefore the aggregated results of expert judgements. A fixed set of dimensions is included in Benoit and Laver’s expert survey (see Appendix 1 for the exact phrasing of the survey’s questions): Taxes vs. spending (measuring parties’ support for public spending vis-à-vis lower taxes), Deregulation (degree of state regulation of the market), Social policy (liberal vs. conservative policies on matters such as abortion, gay rights, and euthanasia), Immigration (support for integration of immigrants into society), Environment (environmental protection vs. economic growth), and Decentralisation (territorial decentralisation of decision-making). The survey also includes three policy scales assessing party positions on specific aspects of European politics: EU authority (scope of EU intervention), EU security (about the EU’s role in peacekeeping), and EU accountability (role of the European Parliament and national governments as democratic accountability mechanisms).³ For each of these nine policy dimensions, experts were also asked to locate each party on a scale measuring the importance or salience of the dimension for that party. This scale ranges from ‘1’ (not important at all) to ‘20’ (very important).

Compared with other methods, expert surveys are a relatively quick and costless way of collecting data on parties. Moreover, they provide scores for individual parties even when they contest the elections as members of pre-electoral coalitions.⁴ Finally, this methodology offers the researcher several possibilities to assess confidence in the accuracy of estimates.⁵ Following Lindstädt, Proksch and Slapin (2015), in order to check the variability in quality of experts’ responses across parties and dimensions between 2001 and 2013, we employ an agreement score borrowed from the organisational psychology and medical fields, which measures the extent to which experts give essentially the same evaluation.

The three plots displayed in Figure 1 represent the distribution of the agreement scores by party across dimensions (a) and by dimension across parties. In the latter case, we show agreement scores relating to both policy positions (b) and issue salience (c). According to the literature, a median score (the white line in the box plots) exceeding 0.7 indicates strong agreement, while a median value below 0.5 signals weak agreement (both cut-off values are indicated by the dashed vertical lines in the figure). Figure 1 shows that the distribution of median scores relating to policy positions across both parties and dimensions is always higher than the 0.5 threshold; and this indicates moderate to strong agreement between the experts’ evaluations.

Figure 1. Agreement score by party and dimension, Italy 2001–13.

In contrast, the scores relating to issue salience are somewhat lower, with few dimensions lying in the poor agreement area. Given the high number of parties and dimensions that experts were asked to judge, we consider these results a good indicator of the quality of measurement associated with our data. However, since not all parties and dimensions are equally well measured, in the following pages we will use, where appropriate, bootstrapped standard errors to account for uncertainty around expert placements (for more details see Lindstädt, Proksch & Slapin 2015).⁶

Research strategy

Expert surveys are generally conducted at specific time points and usually do not provide a basis for creating time series datasets (MacDonald & Mendes 2001). However, in the context of the present research this limitation is not relevant, as expert survey data on the most significant Italian parties have been gathered since 2003 using exactly the same format, thereby allowing comparison across time. Starting from 2003, four expert surveys have been administered to Italian experts following the methodology developed by Benoit and Laver (2006). The first was fielded in 2003. Since it covered parties competing in the Italian 2001 general election, hereafter we will refer to it using the label ‘2001’. The following three expert surveys were held immediately after the 2006, 2008, and 2013 elections, respectively.⁷ Thus, the four surveys together allow us to trace party system change over a relatively long period of time. In each study, experts were asked to locate the most significant political parties competing in the elections (i.e. those obtaining at least one per cent of the popular vote) on nine substantive policy dimensions as well as on the general left–right dimension, using a 20-point scale.

Assessing party system change in Italy over four elections is undertaken here in two steps. First, we examine the changing structure of the policy space, characterised in terms of the relative salience of key issue dimensions and the underlying dimensional structure of the space of competition. Second, we map shifts in the positions of the main political parties in the policy space defined by the most salient dimensions resulting from our analysis. We track such changes by comparing data from the four expert surveys described above, focusing in particular on the first (2001) and last (2013) ones.

To compare parties’ policy positions across time we build upon Laver and Benoit’s (2007) classification of ‘survivor’ or ‘defunct’ parties, and extend it to include parties involved in processes of fusion and fission. In particular, we categorise parties defined as politically significant in 2001 and 2013 into three broad types:

(1)	Those parties that contested both the 2001 and the 2013 elections under the same label. These are defined as ‘survivor’ parties.
(2)	Those parties that contested only the 2001 elections. Among these, parties that ceased to exist between 2001 and 2013 are labelled as ‘defunct’ parties, while those involved in a process of fusion in the years under study are labelled as ‘pre-merger’ parties.
(3)	Those parties that contested only the 2013 election. This category includes ‘brand new’ parties that came into being between 2001 and 2013, ‘merged’ parties resulting from a process of fusion between existing parties, and ‘splinter’ parties originating from a process of fission.

We are aware that the precision of our categorisation comes at the expense of parsimony. However, this seems to be a reasonable cost to bear for the purpose of a cross-time comparison. An alternative approach defining all parties that changed label as ‘defunct’ parties would make cross-time comparison impossible. Table 1 summarises all the parties included in our analysis.

Table 1. Classification of Italian parties, 2001–13.

Only in 2013 elections			Both in 2001 and 2013 elections	Only in 2001 elections
Brand new	Merged	Splinter	Survivor	Pre-merger	Defunct
M5S	PD	DX	LN	FI	MSFT
FiD	PDL	FLI	UDC	AN	RAD
SC	RIV	FdI		DS	SDI
		SEL		DL
				RC
				PDCI
				GR
				IDV

There are only two survivor parties – the regionalist Northern League (Lega Nord – LN) and the Christian Democrats of the Centre Union (Unione di centro – UDC) – that have preserved their original label throughout the period under investigation. Among the other ten parties considered politically significant in 2013, three – M5S, the conservative Civic Choice (Scelta civica – SC), and the economically liberal Act to Stop the Decline (Fare per fermare il declino – FiD) – can be identified as brand new parties; the others are the product of fusions and fissions of existing parties.

The ‘merged’ category includes three political parties: (a) the main party of the centre-left, i.e. the Democratic Party (Partito democratico – PD), originating from the fusion of the post-communist Left Democrats (Democratici di sinistra – DS) with the centre-left Daisy party (Margherita – DL) in 2007; (b) the main party of the centre-right, i.e. People of Freedom (Popolo delle libertà – PDL), originating from the fusion of right-wing Go Italy (Forza Italia – FI) and the National Alliance (Alleanza nazionale – AN) in 2009; (c) Civil Revolution (Rivoluzione civile – RIV), an extreme-left electoral list born in 2012 and composed of Communist Refoundation (Rifondazione comunista – RC), the Party of the Italian Communists (Partito dei comunisti italiani – PDCI), the Greens (Verdi – GR), and corruption-bashing Italy of Values (Italia dei valori – IDV).⁸

The ‘splinter’ category includes four parties: two offshoots of the PDL, i.e. the more centrist Future and Freedom (Futuro e libertà – FLI) and the more right-wing Brothers of Italy (Fratelli d’Italia – FdI); a splinter of AN named The Right (La Destra – DX); and Left Ecology Freedom (Sinistra ecologia libertà – SEL), which is an extreme-left party originating from a fission within the PD in 2009.

Finally, we count three parties that ceased to exist as politically significant actors between 2001 and 2013: the extreme-right Tricolor Flame Social Movement (Movimento sociale fiamma tricolore – MSFT), the left-wing Italian Social Democrats (Socialisti democratici italiani – SDI), and lastly the Radicals (Radicali – RAD). Table 1 is illustrative of the fluidity of the Italian party system over the last two decades (Di Virgilio, Giannetti & Pinto 2012; Pinto 2015; Verzichelli 1996).

In what follows, we evaluate how the turnover of party labels and processes of fission and fusion brought about significant changes in the space of political competition. In this regard, the study of brand new parties may be particularly interesting. It has been shown that new parties can have an impact in redefining the structure of the space of policy competition, as they may introduce and prioritise issues previously ignored by established parties. As a consequence, issue salience and party positions can vary according to the combination of dismissive, accommodating, or adversarial strategies developed by mainstream parties in response to the entry of new actors into the political arena (Meguid 2005, 2010).

Changing policy space

How has the importance of policy domains for Italian political actors evolved over time? And what are the implications for the underlying structure of the policy space? To answer these questions, we analyse the full set of substantive issues included in the expert surveys in order to describe the spatial structure of party competition in Italy between 2001 and 2013. We undertake this task in two steps. First, we use parties’ importance scores to identify the more politically salient policy dimensions in Italy in the four time points examined. Second, we analyse the patterns of correlation between party positions on different policy dimensions in order to identify the underlying axes of political competition in the period under investigation.

As noted above, the expert survey methodology is based on a priori decisions that identify the policy issues of potential importance in a given context. Parties attach different degrees of importance to each of these issues, and party saliency scores enable us to understand which dimensions are the most relevant. For each of the four time points, we measure the overall importance score of a policy dimension by computing the mean of the party-specific saliency scores, weighting it by the vote share received by each party in the election immediately preceding the survey. In order to guarantee comparability over time, we compute the relative salience of a policy domain as the ratio between the saliency of that domain and the average saliency of all domains at that time point. Table 2 reports the relative importance score of each dimension in each election. Saliency scores higher (or lower) than unity indicate above (or below) average importance.

Table 2. Relative salience of policy domains in Italy, 2001–13.

Policy domains	Elections
	2001	2006	2008	2013
Taxes vs. spending	1.11	1.12	1.15	1.11
Social policy	0.97	1.09	0.98	0.93
Deregulation	1.08	1.09	1.15	1.05
Environment	0.81	0.83	0.91	0.97
Decentralisation	0.93	0.91	0.93	0.91
Immigration	1.08	1.09	1.16	1.00
EU authority	1.05	0.99	0.98	1.17
EU accountability	0.96	0.86	0.80	0.96
EU security	1.00	1.01	0.92	0.91

Notes:

Relative saliency is computed as the ratio between saliency of each domain and the average saliency of all domains for each time point. Bold = issues with salience higher than the average salience in that election.

As Table 2 shows, we can identify three domains that retain their importance for all four elections examined in this study. These are the two dimensions dealing with economic policy (Taxes vs. spending and Deregulation) and Immigration. Let us also underline the relevance of EU authority, referring to parties’ propensity to increase or reduce the areas dealt with at the European level. This dimension is particularly salient in the years 2001 and 2013 – following the introduction of the euro and the economic and financial crisis in the eurozone, respectively – and has an average saliency in the period in between.

Taxes vs. spending has been judged by our sample of experts to be the most important dimension in Italian politics between 2001 and 2006. Contestation on this dimension has indeed been considered predominant in most Western countries throughout the post-war period (Bartolini & Mair 1990; Mair & Bartolini 1990). However, the policies of Immigration in 2008 and EU authority in 2013 received on average higher saliency scores than those domains dealing with economic policy. This indicates a significant change in the relative weight given by Italian parties to issue selection in recent election campaigns. According to Figure 1, all the dimensions mentioned above have an expert agreement score higher than 0.5, indicating moderate or strong congruence among experts’ placements.

Changes in the relative saliency of policy domains were particularly important during the 2013 electoral campaign, when the involvement of European institutions in Italian domestic affairs was a dominant theme (Bosco & Verney 2012; Di Virgilio & Giannetti 2014). The policies adopted by the Monti government during 2012 created increasing tensions both between and within parties. This is because of the austerity measures demanded by the EU, which made pro-/anti-EU attitudes the most important dimension of competition between parties (Di Virgilio et al. 2015a). With regard to the other domains, environment policy and the EU’s military role are particularly interesting. EU security has progressively lost importance for Italian parties: this is probably due to the gradual disengagement of the Italian armed forces from Iraq and Afghanistan. Although the environment was the least salient issue in 2001, it emerged as one of the top five in 2013.

The increased salience of environmental protection can be better understood by looking at Figure 2, which complements the results presented earlier in Table 2. Figure 2 plots 2013 saliency scores (with 95 per cent bootstrapped confidence intervals) by dimensions and parties, identified according to the categorisation introduced in the previous section. Brand new parties entering the Italian party system between 2001 and 2013 attach significantly greater importance to environmental protection than do other political groups. This is undoubtedly due to the presence of M5S, whose electoral manifesto emphasised the protection of common goods and the environment, sustainable development, and support of eco-friendly lifestyles (Bordignon & Ceccarini 2013; Pedrazzani & Pinto 2013, 2015).

Figure 2. Saliency scores in Italy 2013, by dimension and party classification.

Conversely, new parties tend to attach less importance to some of the issues that are among the top ranked for other parties, such as economic or immigration policy. These results support the idea that brand new parties, and in particular M5S, are trying to develop a reputation for competence in specific policy domains in order to differentiate themselves from extant political parties seen as representative of the ‘old’ politics (Pedrazzani & Pinto 2015).⁹

Having identified the most salient policy domains in the period under study, we now turn to providing a more synthetic interpretation of the underlying dimensional structure of Italian party competition between 2001 and 2013, checking whether issue dimensions are orthogonal to each other or load on one or more latent factors. Following Benoit and Laver (2006), we assess the complexity of political competition in Italy using principal component analysis (PCA, also known as ‘factor analysis’): a statistical ‘data reduction’ technique that allows us to describe the variability among a large set of observed variables in terms of few unobserved underlying factors.¹⁰ Each of the extracted factors can then be interpreted substantively by looking at those original variables that correlate (or ‘load’) on the factor. The main results of PCA applied to the experts’ placements of Italian parties’ policy positions are illustrated in Table 3, which reports the factor loadings of each domain on the latent dimensions with eigenvalues higher than unity.¹¹

Table 3. Dimensional analysis of party positions in Italy, 2001–13.

Policy domains	Elections
	2001			2006			2008		2013
	1	2	3	1	2	3	1	2	1	2
Taxes vs. spending	0.78	–	–	0.62	–	–	0.75	–	0.83	–
Social policy	0.70	–	–	0.78	–	–	0.81	–	0.82	–
Environment	0.85	–	–	0.79	–	–	0.76	–	0.81	–
Decentralisation	–	−0.83	–	–	−0.88	–	–	−0.61	–	–
EU security	–	–	0.95	–	–	−0.94	–	−0.80	–	0.84
Immigration	0.93	–	–	0.91	–	–	0.91	–	0.76	–
Deregulation	0.58	0.69	–	–	0.68	–	–	0.71	0.80	–
EU accountability	0.89	–	–	0.89	–	–	0.85	–	0.53	0.54
EU authority	0.89	–	–	0.87	–	–	0.89	–	–	0.84

Notes:

Only factor loadings higher than 0.5 are reported. See Appendix 3 for complete data.

The results of the PCA indicate that the full set of the nine policy dimensions identified for Italy can be reduced to three primary axes of competition between 2001 and 2006, and to just two axes starting from 2008. As Table 3 shows, in 2001 and 2006 the eigenvalue is greater than one for three latent factors, implying that the competition among Italian parties can be described in a three-dimensional political space. Conversely, starting from 2008 we observe a simplification of the structure of the party competition, with only two components passing the threshold of unity.

As reported in Table 3, the first and most important factor emerging from the 2001 analysis is associated with a wide range of issues: one of the two dimensions dealing with economic policy (Taxes vs. spending), the two dimensions tapping social liberalism vs. conservatism (Social policy and Immigration), as well as environmental policy. Also, two of the three dimensions related to European politics (EU authority and EU accountability) are associated with the first component emerging from our analysis, which we can interpret as a general ‘socio-economic left–right’. The second latent factor may be interpreted as a dimension expressing attitudes in favour or against ‘autonomy’ from the central government, as it is associated with Decentralisation and Deregulation. The latter domain, which clearly deals with economic policy, also loads on the first latent factor. The third latent factor is only associated with the EU security domain, measuring party attitudes towards Italy’s involvement in European military security and peacekeeping activities. The Italian policy space looks very similar during the 2006 electoral campaign, when party competition can be still reduced to the same three latent factors.

Starting from 2008 we observe a reduction of the dimensional structure of the Italian policy space. Consistent with the decreasing relevance of EU security, which we emphasised above in our analysis of the evolution of the salience of issue dimensions, the PCA shows that in 2008 the domain dealing with EU peacekeeping operations can be subsumed into the second extracted factor. Besides this, we do not observe any other significant changes in the association between policy domains and latent dimensions described above.

In contrast, the results of the 2013 analysis represent a major change from the past, as they reveal the emergence of a distinct pro-/anti-EU axis of political competition. Before 2013, policy issues concerning the EU were subsumed into other components (see Marks & Steenbergen 2002). However, in 2013 Italian party stances towards the EU formed a primary political axis on their own, orthogonal to socio-economic policy. The second factor reported by Table 3 is in fact strongly associated for the first time with all three policy domains relating to European politics. This component can thus be easily interpreted as capturing the attitudes of Italian parties towards the EU and can be labelled as a ‘pro-/anti-EU dimension’. Although this is not surprising if we consider the recent financial and economic crisis in the Eurozone and its dramatic consequences for Italian politics, the growing relevance of pro-/anti-EU policy for party competition in 2013 constitutes the most important evidence of a change in the structure of the Italian policy space during the last 15 years.

Such a finding may be particularly relevant for a growing research focus on the increasing role of EU-related issues in shaping political competition in EU member states. A number of works on the positions of political elites and voters have shown that European integration can be considered to be part of a cultural (Kriesi et al. 2008) or green/alternative/libertarian vs. traditional/authoritarian/nationalist (GAL-TAN) dimension (Hooghe, Marks & Wilson 2002), or even a dimension on its own (Thomassen & Schmitt 1999). Our results resonate well with the latter position. Moreover, our account largely confirms the findings provided by Van der Brug and Van Spanje (2009), who analysed 82 party positions in 14 countries and found that they were structured by one dominant dimension, with the exception of EU integration.

Research on the politicisation of EU-related issues has emphasised the strong relationship that linked economic motivations with public opinion’s attitudes towards Europe in the past, as European integration used to be associated with material benefits (Bellucci, Sanders & Serricchio 2012). This link has probably been broken by the euro crisis and by the related spread of Eurosceptical sentiments in Italy. The fact that in 2013 we find strong anti-EU stances on both the centre-left and the centre-right of the Italian party system indicates that the well-known ‘inverted U-curve’ pattern, in which opposition to Europe was confined to peripheral parties on the extreme left and right, is no longer completely accurate (see Hooghe, Marks & Wilson 2002). In the following section we discuss party positions on the EU authority dimension and on other salient domains emerging from our analysis.

Change of Italian parties’ policy positions

Having described and interpreted the structure of the policy space in Italy during the 2001–13 period, we now turn to analysing change in parties’ policy positions to see if they have shifted over time in a significant way. This work will cast some light on (1) the pattern of changes experienced by the Italian policy space, (2) the shifts in the positions of existing parties, and (3) the impact of new parties entering Italy’s party system in 2013. To evaluate whether shifts in parties’ placements occurred between 2001 and 2013, we first plot changes in policy positions on the nine issue domains included in our analysis for those parties for which this comparison makes sense: survivor, pre-merger, and merged parties. For the first category, data are available for the 2001 and the 2013 elections. For the second and third ones, we can compare merged parties’ positions with those of pre-merger parties, i.e. those that have been involved in a fusion starting in 2001.

To examine the policy positions of parties for which changes cannot be measured, we plot the locations of all the parties included in our study in a two-dimensional policy space formed by the most salient dimensions highlighted in our previous analysis: economic policy (Taxes vs. spending and Deregulation) and attitudes towards the EU (EU authority). In other words, as suggested by Benoit and Laver (2012), we build policy maps using the dimensions judged to be the most salient by experts rather than the representation of policy positions based on the latent dimensions extracted by factor analysis. Although useful to understand how complex the policy space is and which dimensions are orthogonal to each other, the latter can indeed produce inconsistent placements of parties.

Figure 3 reports the differences, with 95 per cent bootstrapped confidence intervals, in policy positions of survivors (LN and UDC) and parties that merged during the 2001–13 period. For the latter group, we compare the 2013 locations of PD, PDL, and RIV with the 2001 joint positions of the corresponding constituent parties (those included in the ‘pre-merger’ group; see Table 1). Joint positions are calculated by weighting the scores of constituent parties by their vote shares. A downward movement corresponds to a shift to more pro-spending, socially liberal, pro-environment, pro-decentralisation, pro-immigration, anti-deregulation, and pro-European positions, respectively. Upward movements imply shifts in the opposite directions. To detect if shifts in parties’ locations represent real movements (i.e. the difference is statistically significant), we calculated 95 per cent bootstrapped confidence intervals around estimated means. When the confidence intervals’ bars are both above or below the zero line the differences are statistically significant – that is, there is a reasonable probability that the movements observed are real. On the contrary, if the bars cross the zero line, changes in policy positions are more likely to be due to chance.

Figure 3. Changes in Italian party positions, 2013–2001.

There are three key patterns evident in Figure 3. First, the figure reveals the stability of the policy positions of the two main parties that have dominated Italian political life over the last two decades. Between 2001 and 2013, both PDL and PD roughly maintained their positions on most of the issues included in our analysis, with the exception of Decentralisation. On this issue, the PDL has become more in favour of decentralising powers, while the PD has moved in the opposite direction. We also observe that the PD is significantly more liberal regarding abortion, homosexuality, and euthanasia when compared with the joint position of its constituent parties, DS and DL. Second, Figure 3 shows a noticeable movement of the LN. Possibly as a consequence of its repeated participation in coalition governments, LN has swung more to the centre, adopting less rightist stances towards tax cuts and state intervention in the economy. Thirdly, Figure 3 reveals that UDC and RIV have changed their position on EU-related domains, especially on the EU authority issue. The Christian Democrats are more in favour of increasing EU influence than in the past, while RIV holds a more anti-EU stance than its constituent parties. The stability in the locations of the other parties on EU authority, which has become the most salient domain in 2013, suggests that the changes in the Italian policy space described above might be attributed to the entry of brand new parties in the 2013 election rather than to ideological shifts of existing political parties.

Figure 4 plots the policy location of all the parties investigated in our study in a common two-dimensional space identified by the combination of economic policy and the EU authority dimension.¹² This allows us to highlight several pieces of information useful for the purpose of this work. For the sake of clarification, we divide each policy space into different quadrants, marked with the Roman numerals ‘I’, ‘II’ and ‘III’. ‘I’ corresponds to the region of the space ranging from 0 to 5 and from 15 to 20 on both policy scales and identifies parties with ‘extreme’ policy positions. ‘II’ indicates parties with an ‘eccentric’ position, i.e. those that are located in the 5–15 area on one dimension and in the 0–5 or 15–20 area on the other dimension. Finally, the third quadrant denoted by ‘III’ refers to the area ranging from 5 to 15 on both domains, describing parties with a ‘central’ policy position.

Figure 4. Party policy locations in Italy, 2001–13.

Figure 4 highlights the location of both brand new and defunct parties. Those parties that came into being between 2001 and 2013 can be defined as ‘eccentric’, as they have an extreme policy position either on economic policy or on the EU authority dimension. In particular, Beppe Grillo’s M5S and Mario Monti’s SC have opposite but extreme positions on the EU authority domain, while occupying a quite central position on the scales measuring economic policy.¹³ This result leads us to conjecture that it is the birth of new parties that has contributed most to the changing structure of the Italian policy space, and specifically to the emergence of a distinct pro-/anti-EU axis of competition. The parties that disappeared between 2001 and 2013, with the only exception of the SDI, had extreme or eccentric locations on most of the dimensions of the policy spaces represented in Figure 4. Emblematic is the case of the MSFT, whose extreme position on both EU authority and Deregulation characterised it as a kind of anti-system party.

Turning to the parties for which we can compare policy positions in 2001 and 2013, we observe that the LN’s leftward shift on economic policy allowed the party to move from an extreme to an eccentric location. In 2013, the party remains strongly against the EU, but has more moderate stances concerning market regulation and the trade-off between taxation and public service provisions. This has contributed to reducing overall party system polarisation on this fundamental axis of political competition. Another factor that has slightly decreased polarisation on economic policy domains relates to one of the main merged parties in our study: the PDL. While in 2001 one of its constituent parties, FI, was eccentric on economic policy, in 2013 the PDL was located inside the borders of the central area of the issue space. As for the other large merged party, the process of fusion experienced by PD brought about virtually no change on the economic dimensions, while producing a shift toward the centre of the space on the EU authority dimension (where DL was slightly eccentric in 2001).

Analysing mass surveys data, Morlino (1996) defined the party system of the Italian Second Republic (1994 onwards) as a ‘neo-polarised’ one. Ieraci (2006), using a content analysis of parliamentary debates on the investiture votes of governments between 1996 and 2001, reached a similar conclusion when talking about a ‘polarised bipolarism’. These definitions catch three particular features of the Italian party system emerging from the transition from the First Republic (1948–93) to the Second Republic: a bipolar coalitional configuration, high polarisation, and the absence of any large party occupying the centre of the policy space. In contrast, other studies conducted with different data and statistical techniques emphasised a gradual change towards a centripetal pattern of party competition in Italy (e.g. Conti 2008; Curini & Iacus 2008). Our results show the relative stability throughout the 2001–13 period in the positions of the main political parties characterising Italian political life during the Second Republic, with just minor centripetal dynamics.

Figure 4 also suggests that the PD was born on a more solid basis than the PDL, given that the ideological distance of PD’s constituent parties (DL and DS) on our core dimensions is lower than that registered for the PDL. As a result, the policy position of the PD is almost a perfect synthesis of those of the DS and the DL, while the position of the PDL is more skewed towards FI. Policy distances between PDL’s constituent parties have undoubtedly contributed to the fissiparous nature of Silvio Berlusconi’s political party, which since its birth has had three different fissions characterised by either centripetal or centrifugal movements. During Legislature XVI, Gianfranco Fini openly challenged Berlusconi’s leadership, demanding more intra-party democracy and a change in government policies. As a result, in July 2010 a new parliamentary party – FLI – was created. In December 2012, the PDL lost another piece when former members of AN founded FdI. Finally, in Legislature XVII, which is not covered by our analysis, the Secretary of the PDL initiated a new split by forming the New Centre-Right (Nuovo centrodestra – NCD) party.¹⁴ Overall, during the 2001–13 period the distance between the main parties of the left (PD) and right (PDL) does not seem to have increased. The merger of FI into the PDL has indeed led to the disappearance of a large party with rather extreme stances concerning the tax–services trade-off.

Our analysis, then, shows that fusions and fissions have certainly contributed to the transformation of the Italian party system through modifying fundamental parameters such as parties’ relative strength and their policy positions. However, the most crucial modifications in the structure of the Italian party competition space seem more attributable to the emergence of new parties, than to processes of fusion and fission involving existing parties or to their ideological realignments.

Conclusions

The purpose of this article has been to examine party system change in Italy between 2001 and 2013 from the perspective of the spatial model of party competition. This led us to analyse the evolution of the policy space and parties’ changing policy positions. To perform this task we used data from four expert surveys in 2001, 2006, 2008, and 2013, which provide comparable information about the salience of several issue dimensions and party policy positions for each election in the period under investigation, as well as evidence on the quality of measurement. Overall, our empirical findings indicate that the Italian policy space between 2001 and 2013 has been characterised by four salient issues relating mainly to economic policy, immigration, and attitudes towards EU influence on domestic affairs.

However, in spite of this stability a number of important changes can be observed: economic policy and immigration have become less important in comparison with the other core policy dimensions (albeit not in comparison to all policy dimensions), while issues relating to EU authority have become more salient. This trend has been confirmed by an analysis of the correlation patterns in party positions on issue dimensions. While in the past EU-related factors used to be associated with economic and immigration policy, since 2013 they have become more distinctly aligned with a new dimension of political competition tapping pro-/anti-EU attitudes. This constitutes a major change in the structure of the Italian policy space.

Our analysis of changing party policy positions suggests that the emergence of a distinct pro-/anti-EU axis may be attributed not to shifts of existing parties, but to the entry of brand new parties such as M5S and SC in 2013. Such parties have an extreme although opposite location on the dimension measuring parties’ attitudes towards the EU, while they occupy a quite central position on the domains relating to economic and immigration policies. In addition, the astonishing success of Grillo’s M5S – which in the 2013 elections won 25 per cent of the popular vote for the Chamber of Deputies and around 23 per cent for the Senate – contributed to undermining the bipolar competition that started with the coalescence of the main centre-left and centre-right parties. This represents clear evidence of party system change.

Our findings also speak to the literature about Euroscepticism. The extreme anti-EU attitudes of the M5S challenge the way the literature has looked at Euroscepticism over the past decade, questioning the hypothesis according to which the anti-EU stance was typical of parties located at the periphery of the policy space (Hooghe, Marks & Wilson 2002).

Our work does not simply assess the dimensionality of policy space in Italy at a particular time point. On the contrary, we trace its evolution through a reasonable period of time, showing how EU integration, which was embedded in a general left–right dimension, became an axis of competition on its own. In so doing, we contribute to the vast literature dealing with emerging cleavages and the dimensionality of party systems in Western European countries. Here we focus mainly on parties, touching only marginally on the role and attitudes of voters. Future research should therefore concentrate on the latter, exploring for example whether changes in the structure of the policy spaces and parties’ policy positions are matched by analogous shifts in voters’ ideological preferences and polarisation. This work could contribute to a better understanding of the whole process of representation and democratic competition, thus constituting a fruitful direction for future research.

Notes on contributors

Daniela Giannetti is Professor of Political Science at the University of Bologna. Previously she has been a research associate in the Department of Political Science, Trinity College, Dublin. Her research interests and publications focus on rational choice theory of political behaviour and institutions.

Andrea Pedrazzani is a research assistant in the Department of Political and Social Sciences at the University of Bologna. His research interests include parliaments, legislative behaviour, executive–legislative relations, and intra-coalitional politics. He has published articles in journals such as European Journal of Political Research, Legislative Studies Quarterly, Government and Opposition, Political Studies, and Italian Political Science Review.

Luca Pinto is a research assistant in the Institute of Human and Social Sciences, Scuola Normale Superiore, Italy. His research interests include party competition, legislative studies, and coalition governments. He has published in several journals such as Party Politics, Journal of Legislative Studies, Government and Opposition, International Political Science Review, Political Studies, Acta Politica, and Italian Political Science Review.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. In 1993, the open-list proportional representation (PR) system that had been used in Italy since 1948 was replaced by a mixed electoral system (75 per cent plurality, 25 per cent PR). In 2005 a new electoral reform was enacted establishing a closed-list PR rule with a majority bonus.

2. Laver and Benoit (2007) used expert survey data collected in 1989 and 2003. Italy was partially excluded from their study because there was too much party turnover between the two time points to make useful pairwise comparisons. Our investigation therefore complements their work, although focusing on a different time period (2001–13).

3. Let us note that, although the term ‘EU security’ explicitly refers to the EU’s participation in peacekeeping activities, this domain captures the overall attitude of parties towards military operations.

4. This is especially important for the Italian case. Indeed, since the 1994 election the MRG/CMP has coded not the programmes of Italian parties but rather the pre-electoral coalition platforms, as the latter were the only programmes available. Hence, using MRG/CMP data would not allow us to analyse the policy positions of the main Italian parties in 2001 and subsequent elections.

5. For this characteristic, expert surveys are often used to validate party positions obtained through other techniques.

6. As experts are not drawn from a random sample, uncertainty simply derived from the standard deviation of the sampling distribution of the mean expert placement of parties’ positions could be problematic (Lindstädt, Proksch & Slapin 2015).

7. Experts are usually selected from members of the Italian Political Science Association (SISP). For further details, see Benoit and Laver (2006) on the 2003 expert survey, Curini and Iacus (2008) on the 2008 survey, and Di Virgilio et al. (2015a) on the 2013 survey. We thank Kenneth Benoit for sharing with us expert survey data for the 2006 national elections. Appendix 2 summarises all the relevant information about the four expert surveys taken into consideration in this study.

8. RIV was a joint list formed to run in the 2013 elections, headed by the anti-mafia prosecutor Antonio Ingroia, who entered the race with the support of some of the parties that in 2008 had formed the Rainbow Left coalition. Since we do not have data on the groups joining the electoral list, we treat it as a single party.

9. This cleavage between ‘new’ and ‘old’ politics is also reflected in the electoral base of the M5S, which is particularly popular among young voters (Maraffi, Pedrazzani & Pinto 2013; Pedrazzani & Pinto 2015).

10. Factor analysis has been applied to extract primary dimensions of party competition both from the policy issues identified by expert surveys (Laver & Hunt 1992; Benoit & Laver 2006) and from the coding categories of party manifestos (Budge, Robertson & Hearl 1987; Gabel & Huber 2000). The method we use for factor extraction is PCA, whereby the input variables are modelled as linear combinations of a smaller set of factors in order to account for the maximum possible variance in the data. To obtain factors that are more easily interpretable, we used varimax rotation, which is the most commonly employed rotation option. Detailed analysis for each time point is reported in Appendix 3.

11. Eigenvalues measure the amount of variation accounted for by each factor. In the PCA technique, only factors with eigenvalues greater than unity are conventionally retained.

12. Party positions on Deregulation and Taxes vs. spending are highly correlated (r = 0.97, p < 0.05).

13. The expert survey data used in this paper characterise the M5S as a party with extreme views on the EU, but not on the economic left–right spectrum. Similar results are reached using voters’ and candidates’ self-placements, whose aggregated positions result in a quite centrist location of the M5S, between PD and SC (see Pedrazzani & Pinto 2015, p. 93; Di Virgilio et al. 2015b). In contrast, recent analyses of the M5S’s 2013 electoral platform locate it on the extreme left of the economic scale (Conti 2014). This can be explained by the fact that the M5S manifesto was created through a deliberative bottom-up process involving members and followers, which resulted in a leftist party platform. Indeed, Italian experts assigned policy scores to the M5S taking into account not only the electoral platform, but also the views expressed in public by the M5S leader, who has often taken positions that appeal to right-wing voters (Conti 2014, pp. 6–7). It should be noted, however, that M5S is among the parties for which experts’ placements show the highest level of disagreement (see Figure 1).

14. After this split, the PDL re-labelled itself as FI, the political party active between 1994 and 2009.

Appendix 1. Italian expert survey policy dimensions

TAXES VS. SPENDING [1]

Promotes raising taxes to increase public services. (1)

Promotes cutting public services to cut taxes. (20)

DEREGULATION [2]

Favours high levels of state regulation and control of the market. (1)

Favours deregulation of markets at every opportunity. (20)

SOCIAL POLICY [3]

Favours liberal policies on matters such as abortion, homosexuality, and euthanasia. (1)

Opposes liberal policies on matters such as abortion, homosexuality, and euthanasia. (20)

IMMIGRATION [4]

Favours policies designed to help asylum-seekers and immigrants integrate into Italy’s society. (1)

Favours policies designed to help asylum-seekers and immigrants return to their country of origin. (20)

ENVIRONMENT [5]

Supports protection of the environment, even at the cost of economic growth. (1)

Supports economic growth, even at the cost of damage to the environment. (20)

DECENTRALISATION [6]

Promotes decentralisation of all administration and decision-making. (1)

Opposes any decentralisation of administration and decision-making. (20)

EU AUTHORITY [7]

Favours increasing the range of areas in which the EU can set policy. (1)

Favours reducing the range of areas in which the EU can set policy. (20)

EU SECURITY [8]

Favours Italy’s involvement in European security and peacekeeping missions. (1)

Opposes any Italy’s involvement in European military affairs. (20)

EU ACCOUNTABILITY [9]

Promotes the direct accountability of the EU to citizens via institutions such as the European Parliament. (1)

Promotes the indirect accountability of the EU to citizens via their own national governments. (20)

THE GENERAL LEFT–RIGHT DIMENSION [10]

Please locate each party on a general left–right dimension, taking all aspects of party policy into account.

Left. (1) Right. (20)

Appendix 2.

Some details of the Italian expert surveys, 2001–13

Table

Criteria	2001	2006	2008	2013
No. of relevant parties	13	16	8	12
No. of respondents	54	40	54	95
No. of dimensions	10	10	10	10
Survey type	Web	Web	Web	Web
Election date	May 2001	April 2006	April 2008	February 2013
Field	September–December 2003	March–May 2006	July–August 2008	February–March 2013
Main reference	Benoit & Laver 2006	–	Curini & Iacus 2008	Di Virgilio et al. 2015a

Appendix 3. Principal components analysis, 2001–13

Table

2001 expert survey		Factors
Policy domains	1	2	3
Taxes vs. spending	0.78	0.48	−0.14
Social policy	0.70	−0.26	−0.39
Environment	0.85	0.27	−0.13
Immigration	0.93	−0.03	−0.05
EU authority	0.89	0.01	0.26
EU accountability	0.89	−0.08	0.16
Deregulation	0.58	0.69	−0.14
Decentralisation	0.19	−0.83	0.04
EU security	0.05	−0.10	0.95
Eigenvalue	4.76	1.55	1.08
Variance explained (proportion)	0.53	0.17	0.12
Variance explained (cumulative)	0.53	0.70	0.82

Notes: Principal component factor analysis weighted by the vote share received by each party. N = 419; parameters = 24. Bold values identify the dimensions that load on each factor.

Table

2006 expert survey		Factors
Policy domains	1	2	3
Taxes vs. spending	0.62	0.39	0.45
Social policy	0.78	−0.10	0.29
Environment	0.79	0.34	0.28
Immigration	0.91	0.11	0.17
EU authority	0.87	0.19	−0.22
EU accountability	0.89	0.11	−0.06
Deregulation	0.41	0.68	0.33
Decentralisation	−0.04	−0.88	0.08
EU security	−0.01	−0.02	−0.94
Eigenvalue	4.80	1.28	1.11
Variance explained (proportion)	0.53	0.14	0.12
Variance explained (cumulative)	0.53	0.68	0.80

Notes: Principal component factor analysis weighted by the vote share received by each party. N = 300; parameters = 24. Bold values identify the dimensions that load on each factor.

Table

2008 expert survey	Factors
Policy domains	1	2
Taxes vs. spending	0.75	0.52
Social policy	0.81	0.31
Environment	0.76	0.46
Immigration	0.91	0.25
EU authority	0.89	−0.16
EU accountability	0.85	−0.08
Deregulation	0.50	0.71
Decentralisation	−0.25	−0.61
EU security	0.19	−0.80
Eigenvalue	5.12	1.54
Variance explained (proportion)	0.57	0.17
Variance explained (cumulative)	0.57	0.74

Notes: Principal component factor analysis weighted by the vote share received by each party. N = 190; parameters = 17. Bold values identify the dimensions that load on each factor.

Table

2013 expert survey	Factors
Policy domains	1	2
Taxes vs. spending	0.83	0.05
Social policy	0.82	0.13
Deregulation	0.80	−0.03
Environment	0.81	−0.29
Immigration	0.76	0.47
EU authority	0.22	0.88
EU accountability	0.53	0.54
EU security	−0.31	0.84
Decentralisation	−0.08	−0.38
Eigenvalue	3.84	2.09
Variance explained (proportion)	0.43	0.23
Variance explained (cumulative)	0.43	0.66

Notes: Principal component factor analysis weighted by the vote share received by each party. N = 684; parameters = 17. Bold values identify the dimensions that load on each factor.