Content uploaded by Giorgio Di Maio
Author content
All content in this area was uploaded by Giorgio Di Maio on Sep 14, 2022
Content may be subject to copyright.
Abstention and Populist Voting:
Evidence from the Italian 2018 Election
Lucia Dalla Pellegrina, Giorgio
Di Maio, Mario Gilli
No 503
SEPTEMBER 2022
1
Abstention and Populist Voting: Evidence from the Italian 2018 Election
Working paper version: 0.18
Date: September 13, 2022
Lucia DALLA PELLEGRINA
Department of Economics, Management, and Statistics (DEMS), University of Milano-Bicocca
Center for Interdisciplinary Studies on Economics, Psychology, and Social Sciences (CISEPS),
University of Milano-Bicocca
ORCiD: https://orcid.org/0000-0002-3648-7506
lucia.dallapellegrina@unimib.it
Giorgio DI MAIO *
Department of Economics, University of Insubria
Center for European Studies (CefES), University of Milano-Bicocca
ORCiD: https://orcid.org/0000-0002-2461-0925
giorgiodimaio@gmail.com
Mario GILLI
Department of Economics, Management, and Statistics (DEMS), University of Milano-Bicocca
Center for European Studies (CefES), University of Milano-Bicocca
ORCiD: https://orcid.org/0000-0002-9407-4021
mario.gilli@unimib.it
* Corresponding author.
We would like to thank the participants at the CefES Research Project "Norms and rules. Causes and effects of socio-
demographic, economic and political dynamics inside and outside Europe" and the participants to 4th International
Conference on “The Political Economy of Democracy and Dictatorship” (Münster, 2020), 3rd International Conference
on European Studies (Milano, 2021), and 11th Annual Conference of the Spanish Association of Law and Economics
(Barcelona, 2021). We are particularly grateful to Thomas Apolte, Roger Congleton, Mario Ferrero, Chiara Gigliarano,
Simon Hug, Samuli Leppala, Giulio Mellinato, Naci Mocan, Stefania Ottone, Martin Paldam, and Margherita Saraceno
for very useful comments.
2
Abstention and Populist Voting: Evidence from the Italian 2018 Election
ABSTRACT
This study aims to investigate the demographic, social, and economic drivers of rising abstention and
populist electoral success in Italy in 2018. The Italian case is unique in the euro area because, in the political
elections of 4 March 2018, two parties usually identified as left-wing and right-wing populists (Movimento 5
Stelle and Lega) obtained an absolute majority of valid votes. In reverse, the main established parties, the
center-left Partito Democratico and the center-right Forza Italia, which have alternated in government since
1994, reached their minimum electoral consent. We study the Italian case at the level of the 110 Italian
provinces (NUTS 3) by using a data set containing a wide set of demographic and socio-economic indicators,
in addition to the results of the political elections in 2008, 2013, and 2018. We regress the results of the political
elections of 2018, i.e., abstention and votes obtained by the main parties expressed as a share of citizens entitled
to vote, on nine factors obtained by applying an exploratory factor analysis on 41 demographic and socio-
economic variables. Results suggest that abstention is associated with the State’s failure in providing socio-
economic development and security and in repressing organized crime. Moreover, socio-economic well-being
is the main driver of voting behavior. In particular, the left-wing populist Movimento 5 Stelle has been
successful in the more backward Italian provinces and the right-wing populist Lega in the more developed
ones. These results indicate that in 2018 mainstream parties have fallen out of favor with both the most
backward and the most advanced provinces, suggesting that the notion of populism should be qualified for an
understanding of the observed varieties of non-mainstream parties and voting or abstention behavior.
KEYWORDS
Populism, voting, abstention, electoral turnout, Italy.
JEL CLASSIFICATION
D72, D78, H11, J68, P16
3
Contents
INTRODUCTION ........................................................................................................................................... 4
1. WHAT IS POPULISM? ............................................................................................................................. 7
2. THE INSTITUTIONAL SITUATION IN ITALY ........................................................................................... 10
2.1 The Italian Second Republic .......................................................................................... 10
2.2 The Changing Electoral Systems .................................................................................... 11
3. A SURVEY OF THE 2008, 2013, AND 2018 ELECTORAL RESULTS ........................................................ 12
3.1 2008 General Elections .................................................................................................. 12
3.2 2013 General Elections .................................................................................................. 12
3.3 2018 General Elections .................................................................................................. 13
4. DATA .................................................................................................................................................. 21
4.1 General Elections ........................................................................................................... 21
4.2 Demographic and Socio-economic Variables ................................................................ 21
5. METHODOLOGY .................................................................................................................................. 24
5.1 Analytical Framework .................................................................................................... 24
5.2 Correlation Analysis ....................................................................................................... 26
5.3 Cluster Analysis ............................................................................................................. 26
5.4 Factor Analysis ............................................................................................................... 27
5.5 Regressions Analysis on Factor Scores .......................................................................... 27
5.6 Supplemental Analysis ................................................................................................... 28
6. RESULTS ............................................................................................................................................. 28
6.1 Correlation Analysis ....................................................................................................... 28
6.2 Cluster Analysis ............................................................................................................. 30
6.3 Factor Analysis ............................................................................................................... 33
6.4 Regression Analysis on Factor Scores............................................................................ 38
7. DISCUSSION AND CONCLUSIONS ......................................................................................................... 40
8. REFERENCES ....................................................................................................................................... 41
4
INTRODUCTION
Italy is a laboratory for populism. The Italian party system has experienced crucial changes over the last
few decades. Key among these changes has been the general elections held on March 4, 2018, with the electoral
punishment of both the center-left and the center-right main incumbent parties, and the success of new or
significantly renovated parties that have been commonly considered as left-wing and right-wing populist and
opposed to European integration.
According to post-election surveys (e.g., Itanes, 2018) in the Italian political elections of 2018 a sizeable
proportion of voters (26.7%) made choices different from the ones they made in 2013. On the one hand, the
Partito Democratico (PD, Democratic Party), the mainstream center-left party, suffered heavy losses, getting
2.76 million fewer votes than in 2013, as well as Forza Italia (FI, Go Italy), the mainstream center-right party,
that got 2.81 million fewer votes than in 2013, both confirming the declining trend in consensus compared to
the previous electoral cycles (from 2008 to 2013). On the other hand, populist parties such as the Lega (League)
and the Movimento 5 Stelle (M5S, Five-Star Movement) gained 4.19 million votes and 1.55 million votes,
respectively (Table 1).
Thus, the 2018 Italian general elections were a crucial test to assess the resilience of mainstream parties
vis-à-vis the challenge provided by populist forces and the stabilization of the tripolar party system that
emerged in 2013.1 In particular, the 2018 election results demonstrated that the previous 2013 elections had
not been an anomaly, confirming the party-system transformation. Moreover, while this transformation has
been quite common in many European countries, such as France, Spain, and Germany, the Italian case is
unique in the Euro area because in the political elections of 4 March 2018, the two left-wing and right-wing
populist parties - M5S and Lega - obtained an absolute majority of valid votes, and formed a coalition
government that lasted from 1 June 2018 to 5 September 2019, when Lega decided to interrupt the government
experience. In reverse, PD and FI, the former main parties of the center-left and center-right coalition, both
reached their minimum electoral consent, being voted by only 12.6% and 9.6% of the citizens, respectively.
But Italy is a relevant setting to study populism, not only because of its electoral success but also because
of the varieties of populism2 represented in the party system. In particular, the electoral success of populist
parties spread across the usual left-right dimension, a very significant case in Europe, at least at this level. The
Italian case is unique in the EU also because of the economic decline experienced in the last decade, compared
1 In 2013 the bipolar party system that in previous elections was characterized by a confrontation between Center-left and
Center-right was substituted by a tripolar situation with the emergence of the M5S that claimed to be neither left nor right
wing but just for the people. See Section 2 for a review of Italian political situation from 2008 to 2018.
2 For a classification of European parties according to the degree of populism and the economic program see Inglehart
and Norris (2016).
5
to other Western countries.3 Despite some improvement in several macroeconomic indicators in the years
before the 2018 election, the lengthy economic crisis has left deep scars on Italian society.4
A final important aspect of the 2013 and 2018 Italian general elections is the high level of electoral
volatility, which is somewhat puzzling (Bobba & McDonnell, 2015), even if it reflects the large-scale processes
of partisan de-alignment and party change that occurred in many established democracies. Extensive literature
has investigated the reasons for this general increase in electoral volatility in representative democracies.5 The
Italian case clearly shows that new (or refurbished) parties can easily win support. But why is this so? The
answer lies both in the strategic interaction between voters' attitudes and the nature of their reasoning, as shown
in Gilli and Manzoni (2019), and in the nature of the political parties and policy platforms among which they
are called upon to choose. A further crucial aspect is a strategic use of voting for the search for alternatives to
parties that have adopted anti-people policies, analyzed in a general way and with a specific reference to the
Italian case in Di Maio et al. (2022).
In most of the literature on populism, the success of the new populist parties is interpreted as stemming
from the process of globalization, which has produced the "economic losers", those for whom the globalization
process has meant economic hardship, income and occupational uncertainty, and "cultural losers", i.e., people
who are disoriented by changes in values, by new waves of migration, and by the loss of national sovereignty.
However, the empirical disentanglement of the relative importance of the factors behind these two hypotheses
is not easy, as witnessed by the sharp confrontation between Mug and Morgan in 2018 about the possible
explanation of Trump's victory (Morgan, 2018b, 2018a; Mutz, 2018a, 2018b), and more generally by the series
of works by Colantone and Stanig (2018c, 2018a, 2019).
A general analysis of populist electoral success should consider both the demand side, i.e. the drivers of
voting for populist parties, and the supply side, i.e. the presence or the entrance of populist parties, as in Guiso
et al. (2017). Nevertheless, for the aim and scope of this paper, the study of the 2018 electoral outcome in Italy,
a partial analysis of the demand side is enough, because in Italy the populist parties were established years
before 2018, the Lega in 1991 and the M5S in 2009, even if M5S presented itself to the election for the first
time in 2013, and Lega Nord (then evolved into Lega) presented itself throughout the national territory only in
2018.
This study aims to investigate the reasons for the rise of abstention and the success of left-wing and right-
wing populist parties in Italy, relating electoral results to demographic and socio-economic factors. In
particular, we focus on abstention and on the four parties that received the largest shares of the vote in the last
three national elections: on one hand, the two populist outsider parties (M5S and Lega) that have progressively
gained ground and, on the other hand, the two main established parties (PD and FI) that alternatively have led
3 In the years following the financial crisis that began in 2007, the Italian real GDP per capita fell below the value it had
in 1998, and it was only in 2017 that it returned above this level. Appendix 8 in the Supplemental material reports a
selection of the main social and economic data for Italy.
4 In particular, in 2018 the income of Italians was still below the pre-2008 crisis levels. For instance, the amounts declared
for tax purposes in 2017 (relating to the 2016 fiscal year) were almost 2% lower, in real terms, than those declared in
2009 (Maraffi, 2018).
5 See Dassonneville and Hooghe (2017) for a review, and Gilli and Manzoni (2019) for a theoretical model.
6
the government in the last 30 years and have gradually lost votes. To this aim, after a survey of the Italian
institutional situation and of the use of the concept of populism, two facets behind the rise of Italian populism
starting from the first decade of the 2000s are examined using a geographic perspective. First, the geography
of voting is considered using cluster analysis. Second, the role of elements that might be central to populism
electoral success is studied through factor analysis and other multivariate techniques.
This paper aims to answer the following questions:
1. Is it possible to cluster electoral results to emphasize the nature of the connections of electoral results
in Italy?
2. Among the multitude of correlated variables alleged to determine populist success, is it possible to
group those concurring similarly to achieve a clearer understanding of the drivers of the electoral
outcome of the most recent Italian elections?
3. Is it possible to use such factors to further understand the populist success in Italy, and possibly to
extend the revealed pattern to other Western countries?
We will answer these questions by looking at several different sources. First, we review the 2008, 2013,
and 2018 elections, then we concentrate on the period from 2013 to 20186 at the level of the 110 Italian
provinces (NUTS 3) by using a large dataset containing different types of demographic and socio-economic
indicators, combining these data with the results of the Italian political election of 2013 and 2018. The
provincial level of analysis is a good compromise between the regional level, on the one hand, and the
municipal level, on the other hand.
Our paper takes the Italian case as a benchmark study since Italy is defined as a laboratory (Tarchi, 2015),
an enduring market (Bobba & McDonnell, 2015), and a breeding ground for populism (Bobba & Legnante,
2017) because of the strong and recurrent success of its populist formations, as argued in Section 1. To the
best of our knowledge, so far little attention has been paid to the empirical analysis of the demographic and
socio-economic factors explaining voting behavior in favor of populist parties in Italy. Some interesting
empirical analyses are provided by Caiani (2019), Corbetta et al. (2018), and Maraffi (2018); however, all their
studies are based on ITANES 2018 post-election surveys, consisting of a sample of 2,573 observations, not on
real electoral outcomes.
We make three methodological choices that differentiate our analysis of the drivers of the vote in favor of
populist parties from those done by other authors, which we consider as methodological contributions to the
research field. First, we investigate the drivers of both abstention and voting behavior, to highlight the potential
similarities between abstention and voting for populist parties. Second, we consider the total number of electors
who abstained, and the total number of votes obtained by parties as a share of the total number of citizens
entitled to vote expressed in percentage points. This methodological choice differentiates our research from
many others, which instead consider the total number of votes obtained by parties as a share of valid votes
(e.g., Dijkstra, Poelman, & Rodríguez-Pose, 2020). Even if it is interesting because it determines the allocation
6 The choice of concentrating on the 2013 and 2018 elections is due to the fact that M5S was established in 2009 and thus
did not concur in the 2008 elections.
7
of parliamentary seats, i.e., the distribution of power among parties, the share of valid votes can lead to
misleading conclusions when it is regressed on demographic and socio-economic indicators since it does not
take into account abstention. The variable of interest is misspecified because participation in the vote varies
over time and space, between successive elections, and between different territories in the same election so
that the same share of valid votes usually corresponds to different shares of citizens entitled to vote. Instead,
the total number of votes obtained by a party expressed as a share of citizens entitled to vote correctly measures
the consent of that party. Third, we perform an exploratory factor analysis to find the latent factors behind a
wide set of demographic and socio-economic variables that could affect abstention and voting behavior, instead
of pre-selecting a small number of variables, i.e., typically less than 10, as several authors do (e.g., Dijkstra et
al., 2020). This approach allows us to prevent two problems. First, we avoid a subjective selection of variables,
which, even if based on the literature, could reflect some kind of priors. Second, by considering a large set of
variables, we minimize the possibility of problems with omitted variables. The factor analysis also allows us
to highlight possible hidden relationships between the variables, which could not be identified in any other
way.
The remainder of the paper is organized as follows. Section 1 briefly reviews the use of the notion of
populism. Sections 2 and 3 present the institutional situation in Italy and the results in the 2008, 2013, and
2018 general elections, respectively. Section 4 illustrates the data set that we built by combining election results
with a broad set of demographic and socio-economic variables. Sections 5 and 6 present the methodology
applied for the analysis and the results obtained, respectively. Section 7 discusses the results obtained and
concludes.
1. WHAT IS POPULISM?
The word populism is increasingly capacious, and its definitional precariousness is proverbial. At one time
it referred specifically to political movements geared toward diminishing the political influence of economic
elites and pushing for a redistribution of incomes to the people at large, a pure people contrasted with a corrupt
elite, whose allegiance is to party and self-care rather than to the people (Canovan, 2002; Mudde, 2004). This
was the meaning that populism had in U.S. politics at the turn of the twentieth century, a left-wing version that
overlapped with socialism. This type of populism is still visible, for instance, in the Podemos, Syriza, La France
Insoumise, and the M5S movements in Spain, Greece, France, and Italy, respectively. Increasingly, however,
particularly in Europe and North America, populism has become overtly nationalist. The French National Front
and La France Insoumise are both labeled populist although the former is avowedly nationalist, and the latter
is radical socialist. This nationalist populism is regarded as fervently disruptive, looking for a totalistic change
in the status quo by challenging the mediating role of political parties and undermining established standards
of political etiquette (Norris & Inglehart, 2019). This broadening in the use of the concept leads some
commentators to wonder whether the term has lost any continuing analytical meaning (e.g., Inglehart & Norris,
2016). Attacks on vaguely specified special interests and claims to represent the people against the politicians
may seem all that most movements designated as populist have in common. The term populism now seems
related to political movements, whether right or left-wing, critical of contemporary economic and social trends,
8
particularly of economic, social, and political globalization. The fact that scholars in different regions use the
same term to analyze strongly divergent political actors raises the question of whether it is merely an
unfortunate coincidence that political actors from different times, from various places, and with different
ideologies have all been labeled populist, or whether they have something in common.
Whatever the truth to the claim that populism is an increasingly meaningless term, two constants serve to
make the term of continuing analytic use whatever other elements might be added, as argued for instance in
Pappas (2019). The first constant is taking politics to the people by questioning and challenging the dominant
political establishment of existing political parties and the experts or technocrats that are seen as unresponsive
to popular demands and the public interest, captured and corrupted by private interests. The second constant
frames the people in an entirely territorial sense of a founding or native group, particularly the ordinary people
in it who are increasingly threatened by foreign or domestic powers. In this regard, populism defines and favors
the identity of a given group against others: populism is structurally marked by a radical partiality in
interpreting the people and the majority.
According to several authors (De Benoit, 2017; Mudde, 2016), these two characteristics imply that
populism denies the necessity for institutional mediation provided by conventional political parties and
intermediate bodies and that if a populist movement comes to power, it can have a disfiguring impact on the
institutions, rule of law, and division of powers that comprise constitutional liberal democracy. However, it
should be noted that in the course of history these features have characterized many political movements aimed
at establishing a democratic regime and improving the living conditions of the people and that the criticisms
addressed to populist parties often recall the criticisms addressed over the centuries to these democratic
movements by the ruling elites of the time, e.g., the nobility and the clergy.
In general, we might say that even if populism differs in different times and places, however, it shares four
characteristics (e.g., Mudde & Kaltwasser, 2017; Urbinati, 2019): 1) the central position of the people, 2) the
critic of the elite, 3) the perception of the people as a homogeneous entity, and 4) the proclamation of a serious
crisis.
The most accredited interpretations of the success of populist parties in Western democracies (e.g.,
Inglehart & Norris, 2016; Mudde & Kaltwasser, 2017) hinge on the process of modernization and
globalization, and their effects on social and economic insecurity. On the one hand, this process has brought
about profound changes in the dominant culture, replacing old values with new ones; on the other hand, it has
altered the economic balance of the economy and society, lifting a few segments of the population upward,
while leaving many others behind and with more social and economic insecurity. Social changes have been
flanked by new ethical openings in the sphere of gay rights, homosexual civil unions, etc. The disorientation
of the more traditionalist citizens has been further raised by non-Western immigration, causing some sectors
of the population to feel like strangers in their own country. Besides this social insecurity, there is economic
insecurity. The severe crisis that afflicted many economies in the last years led to radical transformations both
in the structure of production and in the conditions of the working class, even if economic changes have started
long before. In the last thirty years, manufacturing industries have declined, industrial production has been
9
transferred abroad, automation has eliminated jobs, immigration has brought in competition for labor, trade
unions have been weakened, and the sustainability of the welfare state has been undermined. The 2008
financial crisis, followed by the Euro area “sudden stop” crisis (Baldwin & Giavazzi, 2015), has added further
economic uncertainty. All these processes have created new conditions of economic insecurity and social
deprivation: the present is uncertain and prospects non-existent. The age of the affluent society (Galbraith,
1958) has given way to the age of sad passions: a pervasive sense of helplessness and uncertainty (Benasayag
& Schmit, 2003). In such a setting, widespread resentment against the dominant elites and the mainstream
parties readily emerges, providing fertile ground for populist appeals. The socioeconomic decline, perceived
also towards one's parents, feeds support for right- and left-wing radicalism (Bolet, 2022).
Many works have established connections between economic insecurity and populism electoral success,
and more generally the significant role of the so-called ‘China effect’ (e.g., Colantone & Stanig, 2018a, 2018b,
2018c; Guiso et al., 2017 and references therein). Moreover, Dustmann et al. (2017) highlight how the populist
vote is related to distrust in institutions, which in turn is correlated to unemployment and economic difficulties.
The explanation of these findings given by Guiso et al. (2017) is that populism is a three-part phenomenon:
(1) anti-elite rhetoric; (2) immediate protection offers, and (3) hiding the future costs of the protection policies
proposed. A reduction in wages and employment creates a direct effect in terms of economic insecurity. Such
economic insecurity, if protracted and pervasive, reduces trust in current government policies and institutions
and reduces voter turnout. Then, if government policies result ineffective to counter the crisis, populist supply
arises, tempting voters with apparent protection strategies (such as trade barriers, limiting immigration from
poor countries, or reintroducing a national currency in place of the euro).
The limit of this explanation is that it is strongly based on the subjective judgment of the authors and a
paternalistic view of the voters. It is argued that those who vote for populist parties do not understand that
there is no alternative to the policies proposed by mainstream parties, even if their effects are disastrous for
most people. In practice, the preference for certain policies defines the categories of analysis: whoever
disagrees is a populist.7 On the other hand, the vote for populist parties may be considered as a lack of
confidence in the mainstream parties that dominate the institutions rather than a lack of confidence in the
institutions themselves. On the contrary, it is abstention that can signal both a lack of confidence in the
institutions and protest against mainstream parties.
The case of Italy illustrates several of the facets of populism that have recently become visible elsewhere
in Europe and North America. The use of the term populism to describe aspects of Italian politics dates to the
early 1990s, and even earlier. This followed the disintegration of the post-World War II party system that
coincided with the end of the Cold War. As the Italian economy stagnated in the 1990s and early 2000s and as
the global economic crisis in 2008 began to take its toll on Italian households, the trend toward populist politics
intensified. The Italian case is specific, though the increasing allure of populism is reflected across several
countries facing similar crises of popular economic and cultural confidence in existing political regimes and
7 Actually, Gilli and Manzoni (2019) propose a model where there is room for a reverse causality effect from lack of trust
to ineffective economic policies and consequent high electoral volatilities.
10
associated political parties. It is important to note, however, that Italian populism began well before the recent
economic crisis and associated austerity policies that are often invoked as its immediate causes. In the
immediate aftermath of World War II, there were strong signs of popular aversion to mainstream parties and
politicians.8 It is no surprise that when the post-World War II party system disintegrated between 1989 and
1992, the populist sentiment was not just in the air but already under mobilization. The initial beneficiary of
the collapse of the two main parties that had ruled Italy since 1963, Democrazia Cristiana (Christian
Democracy) and Partito Socialista Italiano (Italian Socialist Party), were Forza Italia, a personalist party,
founded by the TV tycoon Silvio Berlusconi, using his media firm as the corner stone of the future party, and
the Lega Nord, a party created in 1989 from several regionalist movements across Northern Italy. In almost all
respects this was an incarnation of an ideal-typical populism. Lega Nord was born from local activists, its
leader until 2012 was Umberto Bossi, a man of the people, who steered the party between attacks on the
national government (‘Roma ladrona’, Rome the thief), proposals for the secession of Northern regions, and,
in 1994, the joining into a governing coalition led by Silvio Berlusconi, who presented himself as a self-made
man who opposed political games and tricks. The current leader of the Lega, Matteo Salvini, has turned Lega
Nord into a nationalistic anti-immigrant party, flirting with neo-fascist groups. This allowed Lega support to
spread beyond the North but in so doing it dropped its regionalist for a nationalist populism, while the declining
leadership of Berlusconi has reduced both the electoral appeal and the populist characteristics of Forza Italia.
2. THE INSTITUTIONAL SITUATION IN ITALY
2.1 THE ITALIAN SECOND REPUBLIC
The years between 1992 and 1994 have been regarded by most observers as a turning point in Italian
politics, to the extent that this period has been referred to as the passage from the so-called ‘First Republic’ to
the ‘Second Republic’. The year 1993 saw the collapse of the five governing parties under the ‘Tangentopoli’
(‘Bribesville’) scandals, the creation and regeneration of other parties, and the abandonment of the proportional
electoral system, which had been a foundational and stable feature of the post-1945 party system, in favor of
a mixed (mainly majoritarian) system, changing the incentive structure for electoral and political strategies.
The reformed mixed system displayed a predominantly majoritarian logic as far as the mechanical effects of
translating votes into seats are concerned. This majoritarian logic provided a very strong incentive for the
formation of electoral-political alliances to win in first-past-the-post voting contexts, which was also facilitated
by a parallel process of ideological softening across the political spectrum. As a consequence, a bipolar pattern
of competition was established. The new electoral system was used in the 1994, 1996, and 2001 general
elections. Between 1994 and 2013, two coalitions organized along a basic left-right continuum increasingly
accounted for most votes across Italy. The polarizing capacity of the center-right leader, Silvio Berlusconi,
was also important because he recruited other right-wing factions into his camp and institutionalized his
alliance with the Lega Nord. Yet, there was definite geography shaping the overall national bipolarity. FI
8 This was manifested most clearly in the Fronte dell'Uomo Qualunque (Front of Ordinary Man) founded by Guglielmo
Giannini, a satirical journalist and comedian.
11
indeed had to share votes and seats with the Lega Nord in the Northern regions but was faced with serious
competition by the center-left in the South, and a dearth of opportunities in the Center, where the center-left
parties exercised a considerable draw.
2.2 THE CHANGING ELECTORAL SYSTEMS
Italy stands out among advanced industrialized democracies because of its frequency of major electoral
reforms. In the postwar period, Italy experienced four major electoral systems: the proportional representation
(PR) system (1948-1992), the mixed-member majoritarian (MMM, 1993-2005), and two varieties of PR with
majority bonus (2005-2015, 2015-now).
The proportional system was introduced with the electoral law of 1946, and, with minor variations, remained
in force for nearly fifty years, folding under heavy criticism in the early Nineties, as it was considered the main
cause of party fragmentation and government instability, and abolished by referendum in 1993, leaving the
field to a new electoral law mainly based on single-candidate constituencies, the Mattarella Law. The new law
replaced the previous system of proportional representation and remained in force until 2005 when it was
replaced by the Calderoli Law. The Mattarella Law introduced a mixed electoral system: for the Senate, it was
majoritarian with a single ballot for the allocation of 75% of parliamentary seats, the remaining 25% seats
allocated to the proportional recovery of the most-voted non-elected; for the Chamber of Deputies, it was a
proportional system with blocked lists and a 4% threshold. Hence, the Mattarella Law entailed three different
modes of seat distribution: majority in the Senate, proportional in the Camera, and proportional recovery in
the Senate. The Calderoli law of 2005 amended the Italian electoral system, introducing a radically different
scheme. The main change was the elimination of single-member constituencies, along with the re-introduction
of multi-member constituencies under proportional rules for both branches of Parliament. The law introduced
a modified proportional representation based on coalitions, a majority premium which is managed differently
in the two branches of Parliament, and blocked lists with candidates appointed by the parties with no possibility
for voters to express their preferences for individual candidates, who are elected according to their position in
the list. At the Senate, the majority premium was assigned on a regional basis, allocating at least 55% of the
seats reserved in a region to the majority coalition that won the election in that region. At the Chamber of
Deputies, a majority premium of 340 seats was given to the relative majority party or coalition with no
minimum threshold to obtain the premium. This law ruled the Italian general elections in 2006, 2008, and
2013.
In 2017, a new electoral law was approved, the Rosato Law. It calls for a mixed electoral system: 61% of
seats (386 in the Chamber of Deputies and 193 in the Senate) are allocated on a proportional basis among
parties that take more than 3% of valid votes, whereas 37% of seats (231 in the Chamber of Deputies and 115
in the Senate) is attributed following a plurality rule in single-member districts (SMDs). Before the election,
politicians and analysts were particularly curious to compare proportional results with majoritarian ones:
indeed, it was broadly expected that majoritarian competition - where ‘winner takes all’ and individual qualities
of the candidates are crucial - should have favored pre-electoral coalitions at the expenses of M5S. It did not
12
happen. On the contrary, majoritarian results were quite similar to proportional results for M5S and the other
coalitions.
3. A SURVEY OF THE 2008, 2013, AND 2018 ELECTORAL RESULTS
This section presents a survey of the results of the Italian general elections held in 2008, 2013, and 2018,
which are summarized in Table 1. Figure 1-4 provide the geographical representation of these results for
abstention and the three parties that will be considered in the empirical analysis, i.e., Lega, M5S, and PD.
3.1 2008 GENERAL ELECTIONS
The general election held on 13-14 April 2008 was conducted under the electoral rules introduced in
December 2005 by the center-right. It marked a further milestone in the reconfiguration of the Italian polity,
ongoing for over 15 years. The election took place after the collapse of the nine-party center-left coalition,
elected with a narrow majority in April 2006. The center-right won the 2008 parliamentary elections with a
significant majority. The electoral results caused a near shockwave in Italy: for the first time, only five parties
(two parties for each coalition and the small Unione di Centro, UDC) went to Parliament, and only 80.5% of
the electorate went to the ballot box, the lowest figure in a parliamentary election in the Italian history.
Compared to the 2006 elections, Italy experienced a 3.1% increase in abstainers. The party getting the best
result in terms of votes gained was the Lega (best result in its history till 2018).
3.2 2013 GENERAL ELECTIONS
Berlusconi returned to power in 2008, when Italy was struck by the Great Recession. In 2009, GDP fell
sharply and, consequently, the debt-to-GDP ratio jumped again. In April 2011, the spread between yields on
Italian and German bonds began to grow. On August 5, 2011, the European Central Bank (ECB) sent an
unheard-of letter to the Italian Government, signed by the president in office and by the one appointed, i.e.,
Trichet and Draghi (2011), calling for severe fiscal consolidation and a wide range of radical structural reforms,
starting with the pension system and the labor market.9 Six months later, on November 9, the spread reached
a peak of 575 basis points. In a climate of national emergency, on November 12, as soon as the budget law
was approved, Berlusconi resigned.10 Four days later, the technocratic government led by former EU
commissioner and economist Mario Monti took office. On November 18, the Monti government won the
confidence of the House of Deputies: 556 members voted in favor and only 61 against it. Lega (Lega Nord at
that time) and a small party called Italia dei Valori (Italy of values) were the only parties to vote against it. In
a few days, the Monti government launched a program of fiscal consolidation and structural reforms, along the
lines drawn by the ECB.11 Italy went into recession and the debt-to-GDP ratio increased, even if according to
the Government structural reforms such as the one regarding the pension system would have guaranteed easier
long-term sustainability of the debt-to-GDP ratio. However, the spread between Italian and German
9 The letter was strictly confidential but it was published by an Italian newspaper within a few days.
10 The Government was also in trouble because of Berlusconi’s sex scandals, charges for fiscal evasion, and the weakening
of its parliamentary majority.
11 In an interview with the CNN reporter Fareed Zakaria (2012) in May 2012, which went viral on Italian social media,
Monti declared: “We’re actually destroying domestic demand through fiscal consolidation.”.
13
government bond yields decreased only after the ECB announced a radical change in monetary policy, with
the famous "whatever it takes" speech delivered by Draghi on July 26, 2012 (Draghi, 2012).
In the 2013 general elections, the M5S turned out to be the party with the highest number of votes in the
Chamber of Deputies, despite being outnumbered overall by the center-left and center-right electoral alliances,
however, because of the majority prize awarded by the electoral law, the PD was able to establish a center-left
government. Although Berlusconi made a remarkable comeback in the 2013 election, with an electoral
campaign against European Union policies, he was unable to reconstitute the coalition of regional political
forces that had been the secret of his previous success, because, by early 2012, the diminished health of the
Lega Nord’s leader Umberto Bossi and charges of corruption against the party had taken a toll on Lega Nord,
which faded across the Northern regions.
3.3 2018 GENERAL ELECTIONS
On the one hand, the rise of the M5S in the 2013 election was a significant problem for the PD, even if it
was able to arrange a coalition with part of the center-right to govern till the natural end of the legislature in
2018. On the other hand, when Matteo Salvini was elected as Secretary General, in December 2013, Lega was
not in good shape. It was a regional party, unable to get votes in central and southern Italy, Berlusconi's close
ally since the late Nineties, and thus held co-responsible for his failure. In 2012, investigations on the illegal
use of Lega funds led its charismatic leader, Umberto Bossi, to resign as secretary general. Thus, it was no
surprise that in the 2013 elections, Lega got a paltry 4% of the votes.
But the political arena was ripe with opportunities. First, there was a very broad potential electoral space in
the center-right because Berlusconi was politically worn out and the post-fascist party Alleanza Nazionale
(National Alliance) was in disarray. Second, there was a widespread emotional condition catalyzed by a
specific policy that commanded the attention of the media: immigration. In particular, in 2015 there was a huge
increase of refugees and migrants in Europe, part because of the Syrian war when 1.3 million people came to
the continent to request asylum, the highest number in a single year since World War II.12 In short, Lega is one
interesting example of how a political force moved from political obscurity to political significance in the wake
of an economic, financial, social, and political crisis.
On March 4, 2018, a wind of change swept across Italy's political landscape, and indeed Europe's. The 2018
general election represented a turning point in Italian politics because of the huge success of two populist
parties, M5S and Salvini's Lega,13 and the corresponding decrease in the traditional established parties, FI and
PD. That national election made Italy the first country in Western Europe with a populist majority. Indeed,
Italy was only the last, even if one of the most relevant manifestations of a general trend: a significant number
of populist formations achieved electoral success in many countries with different economic and political
characteristics.
12 This situation is known as the “European migrant crisis”.
13 Both M5S and Lega are by common consent classified among the so-called populist parties (e.g., Corbetta et al., 2018)
and considered as opposed to European integration (e.g., Dijkstra et al., 2020).
14
Both main anti-establishment populist parties have achieved historical success, with a combined vote
representing the absolute majority of votes cast (Table 1). As regards the M5S, never before in the history of
Western Europe had a new party obtained such a high degree of support in only its second appearance at a
national election. M5S obtained 10.25 million votes, improving its already historical 2013 success with an
increase of 1.55 million votes, reaching a popular consensus equal to 22% of those entitled to vote and to
32.4% of the voters. On the other hand, in relative terms, the most significant change was in the support won
by the Lega, which saw its votes quadruple. Lega obtained its best result in a general election, both in absolute
and percentage terms, obtaining 5.59 million votes, with an increase of 4.19 million votes compared to the
previous elections, and reaching a popular consensus equal to 12% of those entitled to vote and to 17.7% of
the voters. For the first time, Lega overtook FI within the center-right coalition.
These historical electoral outcomes were accompanied by a paradigm shift that threw consolidated
territorial alignments into disarray. Salvini’s strategy of transforming the formerly Lega Nord into a national
party (Lega) proved very successful, considering that Lega even achieved an average of 8% of votes in the
South of Italy. The main parties of the Center-left and Center-right - respectively PD and FI – saw more than
five-million voters abandon them. These parties remained significant forces, but they were defeated. However,
while the electoral numbers were unequivocal, what is far less evident is why 50.1% of voters (32.4% for the
M5S and 17.7% for the Lega) were prompted to cast a populist vote. Moreover, we cannot assume that the
reasons that led to this and previous electoral outcomes were underpinned by the same motivations. No other
Western country has an internet-driven movement such as M5S gained power, and no other Western country
had a regionalist (and indeed secessionist) party such as Lega transformed itself into a champion of national
sovereignty in just a couple of years. Compared to trends seen across Europe, which had witnessed the success
of radical right anti-immigrant parties in some Northern European democracies, as well as the success of left-
wing anti-austerity parties in Southern European democracies with weaker economies, Italy proved to be a
peculiar case in which two different left-wing and right-wing 'populisms' were established in the same country.
Indeed, scholars who have dealt with populism have often made distinctions between right-wing and left-wing
populism (e.g., Mudde & Kaltwasser, 2017). The present work focuses exactly on the reasons behind
abstention and the vote for both Lega and M5S, investigating the possible drivers of such an electoral
outcome.14
14 A narrative explanation of the reasons of M5S and Lega success is Orsina (2019).
15
Table 1 – Political election results, Chamber of Deputies, 2008, 2013 and 2018
Political election 2008 2013 2018
Number % of
citizens % of
valid votes Number % of
citizens % of
valid votes Number % of
citizens % of
valid votes
Abstention and turnout
Citizens entitled to vote 47,142,436 100% 47,005,432 100% 46,604,896 100%
Abstention 10,617,017 22.5% 12,932,157 27.5% 14,955,989 32.1%
Turnout
36,525,420
77.5%
100%
34,073,272
72.5%
100%
31,648,908
67.9%
100%
Parties
Movimento 5 Stelle 8,702,987 18.5% 25.5% 10,252,280 22.0% 32.4%
Lega
3,026,844
6.4%
8.3%
1,392,537
3.0%
4.1%
5,587,146
12.0%
17.7%
Partito Democratico 12,092,998 25.7% 33.1% 8,644,542 18.4% 25.4% 5,887,357 12.6% 18.6%
Forza Italia 13,642,745 28.9% 37.4% 7,332,829 15.6% 21.5% 4,471,741 9.6% 14.1%
Fratelli d’Italia 668,886 1.4% 2.0% 1,398,109 3.0% 4.4%
Political areas and alignments
Extreme left 378,116 0.8% 1.0% 95,150 0.2% 0.3% 480,285 1.0% 1.5%
Center-left 15,343,652 32.5% 42.0% 10,852,847 23.1% 31.9% 7,085,809 15.2% 22.4%
Center-liberals
103,760
0.2%
0.3%
3,364,715
7.2%
9.9%
971,815
2.1%
3.1%
Center-right
19,130,396
40.6%
52.4%
10,180,386
21.7%
29.9%
11,905,528
25.5%
37.6%
Extreme right 1,026,485 2.2% 2.8% 421,367 0.9% 1.2% 502,238 1.1% 1.6%
Movimento 5 Stelle 8,702,987 18.5% 25.5% 10,252,280 22.0% 32.4%
Notes. In 2008, Forza Italia and Alleanza Nazionale (which later became Fratelli d’Italia) stood in the elections together with a list called Il Popolo delle Libertà (The People of
Freedom). The source of the data is the Italian Ministry of the Interior.
16
Figure 1 – Abstention in 2008, 2013, and 2018 general elections
Notes. This figure represents the abstention in each election, expressed in percentage points as a share of citizens entitled to vote, and its variations with
respect to the previous election.
17
Figure 2 – Lega in 2008, 2013, and 2018 general elections
Notes. This figure represents the consensus for the Lega in each election, expressed in percentage points as a share of citizens entitled to vote, and its variations
with respect to the previous election.
18
Figure 3 – Movimento 5 Stelle in 2008, 2013, and 2018 general elections
Notes. This figure represents the consensus for the Movimento 5 Stelle in each election, expressed in percentage points as a share of citizens entitled to vote,
and its variations with respect to the previous election.
19
Figure 4 –Partito Democratico in 2008, 2013, and 2018 general elections
Notes. This figure represents the consensus for the Partito Democratico in each election, expressed in percentage points as a share of citizens entitled to vote,
and its variations with respect to the previous election.
20
Figure 5 – Forza Italia in 2008, 2013, and 2018 general elections
Notes. This figure represents the consensus for Forza Italia in each election, expressed in percentage points as a share of citizens entitled to vote, and its
variations with respect to the previous election. In 2008, Forza Italia and Alleanza Nazionale (which later became Fratelli d’Italia) stood in the elections
together with a list called Il Popolo delle Libertà (The People of Freedoms).
21
4. DATA
To analyze the drivers of abstention and voting for populist parties we have built a dataset that contains
both the data of the political elections and a large set of socio-economic and demographic variables.
The observation unit is the province (NUTS 3) and we consider the 110 Italian provinces existing in the
period 2012-2018.
4.1 GENERAL ELECTIONS
The data relating to the general or political elections refer to the elections of the members of the Chamber
of Deputies in 2008, 2013, and 2018. For these elections, all Italian citizens who are at least 18 years old on
the date of the elections have the right to vote. The electoral data provided by the Italian Ministry of Interior
at the municipal level have been aggregated at the provincial level.15 It is important to note that all those
entitled to vote are automatically registered in the lists of the electoral section to which they belong, which is
the closest to their residential address and is usually located in a public building, e.g., schools. In practice,
citizens can vote by just going to their electoral section carrying a valid identity document.
Abstention includes both the electors who did not vote and the invalid votes, i.e., the blank and null ballots.
Because we search for the drivers of abstention and voting for populist parties, we consider the total number
of people who abstained and the total number of votes obtained by parties as a share of the total number of
citizens entitled to vote, expressed in percentage points. This methodological choice differentiates our research
from many others, which instead consider the total number of votes obtained by parties as a share of valid
votes (e.g., Dijkstra et al., 2020). Even if it is interesting because it determines the allocation of parliamentary
seats, i.e., the distribution of power among parties, we believe that the share of valid votes can lead to
misleading conclusions, since the variable of interest might be misspecified. When participation in the vote
varies over time or space, between successive elections, or between different regions in the same election, the
same share of valid votes corresponds to different shares of citizens entitled to vote. Hence, consensus for a
party can decrease without the share of valid votes registering this fact. Instead, the total number of votes
obtained by a party expressed as a share of citizens entitled to vote correctly measures the consent of that party.
This consideration is especially important when the Italian case is considered: abstention is growing nationally,
and it is very different between the various provinces.
4.2 DEMOGRAPHIC AND SOCIO-ECONOMIC VARIABLES
Demographic and socio-economic variables are relative to 41 indicators in 2012 and 2017: 14 on crimes, 6
on demography, 14 on the economy, 1 on education, 1 on innovation, 3 on migration, 1 on the emergency
related to uncontrolled immigration from poor countries, and 1 related to newspapers circulation (Table 2).
The data source is ISTAT, i.e., the Italian national institute of statistics, for almost all indicators.
This wide set of indicators aims to include those factors that may explain abstention and voting for populist
parties, based on the literature and the specific characteristics of the Italian case. In particular, voting for
15 Appendix 1 in the Supplemental material reports some descriptive statistics of the results of the general elections held
in 2008, 2013, and 2018.
22
populist and anti-European integration parties has been explained by considering individual and territorial
factors such as age, education, income, unemployment, inequality, geographic mobility, migration, population
density, geographical isolation, brain drain, and industrial decline (e.g., Dijkstra et al., 2020 and the references
therein).
We have added to the indicators related to these factors several indicators related to crime. The first reason
for this choice is that crime occupies a central place in public discourse in Italy and the lack of security, real
or alleged, is one of the most used arguments by populist parties. Secondly, we have added some indicators
that can be considered as proxies of the presence and activity of organized crime, which is very relevant not
only in the Southern regions of origin of the main criminal organizations but throughout the entire national
territory.
Finally, we considered one indicator relating to the management of uncontrolled immigration by the
government and the circulation of newspapers. From 2011 onwards, many thousands of immigrants from
poorer countries arrived in Italy in an uncontrolled way, i.e., not according to the provisions of the laws in
force. Similar to what happened in other countries, these people fleeing war or poverty or simply searching for
a better future arrived in Italy clandestinely or aboard boats. Many of these people then applied for asylum,
and many asylum applications were rejected. A peculiar choice of the Italian government was to manage this
immigration by distributing people throughout the national territory, based on agreements with local
authorities, in emergency residences. The whole issue was the subject of a heated political confrontation with
the opposition parties accusing the PD’s government of not being able to defend national borders, wasting
money on welcoming migrants, and not being able to repatriate people for whom the asylum application was
not accepted.
We included the number of emergency residence beds in each province as an indicator of the spillovers of
the government management of this uncontrolled immigration. Beyond the controversy at the national level,
this indicator can be considered as a proxy of the repercussions of this phenomenon at the provincial level and
therefore a specific factor that may have been considered by voters.
23
Table 2 – Demographic and socio-economic explanatory variables
Variable Description
Variable Description
Crime Economics
1. Arsons Reported crimes per 10,000 inhabitants 21. Isolation (highways, airports, and ports) Travel times to urban and logistic nodes
2. Attempted homicides Reported crimes per 10,000 inhabitants 22. Participation in the labor market
Labor force aged 15-64 years out of the total population
aged 15-64 (percentage)
3. Bag theft Reported crimes per 10,000 inhabitants 23.
Participation in the labor market: the difference
between men and women
Percentage
4. Home burglaries Reported crimes per 10,000 inhabitants 24. Exports per capita Euro per inhabitant
5. Drug-related crimes Reported crimes per 10,000 inhabitants 25. Income inequality
Gini concentration index on equivalent net household
income
6. Extortions Reported crimes per 10,000 inhabitants 26. Non-performing entry rate of loans to households Percentage of loans to households
7. House robberies Reported crimes per 10,000 inhabitants 27. Unemployment: job seekers aged 15 and over Percentage of the population between 15 and 64 years
8. Intentional homicides Reported crimes per 10,000 inhabitants 28. Value added: manufacturing Percentage of the total value added
9. Mafia homicides Reported crimes per 10,000 inhabitants 29. Value added: public sector Percentage of the total value added
10. Micro criminality Reported crimes per 10,000 inhabitants 30. Value added: per capita Euro per inhabitant
11. Prostitution-related crimes Reported crimes per 10,000 inhabitants 31.
Median gross hourly wage of employees born
abroad
Euro
12. Sexual violence Reported crimes per 10,000 inhabitants 32.
Median gross hourly wage of employees born in
Italy
Euro
13. Robbery Reported crimes per 10,000 inhabitants 33. Mean wage of employees Euro
14. Robbery homicides Reported crimes per 10,000 inhabitants 34. Mean wealth per capita Euro
Demography Education, innovation, migration
15. Fertility rate Number of children per woman 35. Population having at least a secondary degree Percentage of the population between 25 and 64 years
16. Total growth rate of the population Rate per thousand inhabitants 36. Immigration of graduates between 25 and 39 years Rate per 1,000 resident graduates
17. Population between 15 and 64 years Percentage on January 1 37. Foreign residents Rate per 10,000 inhabitants between 15 and 64 years
18. Population over 64 years Percentage on January 1 38. Emigration to other Italian regions
Number of residents who emigrated to other Italian
regions per 10,000 inhabitants
19. Population density Number of inhabitants per square kilometer 39. Emigration abroad
Number of residents who emigrated abroad per 10,000
inhabitants
20. Total immigration Rate per thousand inhabitants 40. Beds in emergency residences for migrants Rate per 10,000 inhabitants between 15 and 64 years
41. Newspaper circulation
Average number of newspapers distributed per day per
10,000 inhabitants above 14 years
Notes. The Ministry of Interior is the source for variable 40, and ADS is the source for variable 41. For all other variables, the source is Istat. All the variables are at the provincial level except
variable 25, which is at the regional level.
24
5. METHODOLOGY
In the debate about cultural vs. economic drivers of populist voting, several studies adopt a research design
that involves regressing vote choices against broad sets of explanatory variables. These variables jointly
include both cultural attitudes and measures of economic distress, and the lack of significance of the economic
indicators in these regressions is then interpreted as evidence that economic factors do not matter for vote
choice. The article by Mutz (2018b) on the Trump election is probably the most prominent example of this
approach. Many subsequent works (e.g., Colantone & Stanig, 2018b, 2019; Morgan, 2018b) challenged these
results because cultural attitudes can be considered “bad controls” (Angrist & Pischke, 2009) since changes in
attitudes are themselves an important channel through which economic variables might affect voting.
To overcome the problem of "bad controls", the approach we use in this paper is to start with a very large
set of demographic and socio-economic variables that may affect voting through many channels, however, we
perform an exploratory factor analysis to find the latent factors behind these explanatory variables. Indeed,
many of the demographic and socio-economic variables that could affect the vote are highly correlated with
each other, both positively and negatively.16 It should also be noted that we have only 110 observations
available for each general election. Pre-selecting a small number of demographic and socio-economic
variables, i.e., less than 10, as several authors do (e.g., Dijkstra et al., 2020), does not seem an appropriate
solution, even if this selection is based on the literature. On the one hand, this approach would introduce an
element of discretion in the analysis, which could reflect the authors’ preferences. On the other hand, it could
create an omitted variable problem. The factor analysis allows us to solve these problems, and it also allows
us to highlight hidden relationships between the variables, which could not be identified in another way.
We then regress the electoral results on the factor scores obtained for each province to find the demographic
and socio-economic determinants of abstention and voting behavior.
In the rest of this section, we illustrate our methodological approach in more detail.
5.1 ANALYTICAL FRAMEWORK
We base our analysis on the assumption that voting decisions in a province, in particular abstention and
voting for populist parties, are correlated with at least some of the demographic and socio-economic factors
that characterize that province. Figure 6 represents our analytical framework. Potential voters know the socio-
economic and demographic characteristics of the province they live in directly, based on their personal
experience or through the social networks they belong to, or through both mass media and parties.17 Whether
by personal inclination or because of the activity of mass media and parties, voters may attach more importance
to one factor rather than to another. Voting decisions are based on socio-economic and demographic factors,
as perceived and interpreted by voters, also depending on their socio-economic conditions, values, preferences,
and beliefs, but also on the electoral law and the political offer available when the elections are held.
16 See Figure A3.1 and Table A3.2 in Appendix 3 in the Supplemental material.
17 However, we do not assume that voters have a perfect knowledge of the society in which they live.
25
The political offer plays a role in determining both abstentions and voting for parties. When considering
the Italian case, it is important to keep in mind two facts that changed substantially the political offer. The first
is that the M5S only presented itself in the two most recent political elections, in 2013 and 2018. The second
is that the Lega presented itself throughout the national territory only in 2018. These two changes in the
political offer make it difficult to conduct an empirical analysis that considers all the elections in 2008, 2013,
and 2018 at the same time. On the contrary, they suggest focusing on the political elections of 2018.
Figure 6 – Analytical Framework
SOCIETY
(Social,
Economic, and
Demoraphic
Factors)
VOTERS
(Personal
Condition,
Values,
Beliefs,
Preferences,
Social
Networks)
MASS
MEDIA
(Selected Facts,
Fake News,
Narrative)
POLITICAL
PARTIES
(Propaganda,
Programs,
Past Behavior)
VOTE?
POLITICAL
ELECTIONS
(Electoral law,
Political
offer)
NO
YES
ABSTENTION
(Null and
Blank Ballots
Included)
VALID
VOTES
Notes. This figure illustrates the analytical framework underlying our study. We assume that voting decisions in a
territory, e.g., abstention and voting for populist parties, are driven by at least some of the demographic and socio-
economic factors that characterize that territory. Many other factors may be involved in determining the voting behavior
without, however, generally altering this fundamental relationship.
26
5.2 CORRELATION ANALYSIS
Having assumed that the consensus for one party in a province is associated with the socio-economic and
demographic factors of that province, the first step of our analysis is to study the correlations between the votes
for the different parties across provinces. We can put forward the following three working hypotheses. First,
if the consensus for two parties is associated with the same factors, which the voters consider important or
characterize the voters, but with correlations of the opposite sign with these factors, e.g., the consensus for one
party grows and that for another party decreases as the average income increases, then the votes for these two
parties in the different provinces will be negatively correlated because the voters consider these two parties as
substitutes. Second, by the same logic, the votes for two parties will be positively correlated with each other
when the voters consider them as complements, i.e., the voters regard them as similar, and the choice between
one and another may be determined by factors other than demographic and socio-economic ones. The
participation in the government of only one of the two parties or a scandal that hit one of them may be factors
not correlated with socio-economic and demographic factors that explain the transfer of votes between two
parties that are complements.18 Third, we interpret the absence of a significant correlation between consensus
for two parties as the indication that the voting behavior is driven by factors that are national or specific to
some provinces, such as cultural factors or the presence of linguistic minorities.
5.3 CLUSTER ANALYSIS
The second step of our empirical strategy is to apply cluster analysis for grouping the Italian provinces
according to their electoral results in the political elections of 2008, 2013, and 2018. We consider abstention
and both parties' and coalitions’ votes. We aim to obtain an “objective” representation of the electoral results
that can provide some early insights into the 2018 results as compared to previous elections’ results.
We use cluster analysis to partition the 330 observations in the data set, i.e., the 110 Italian provinces
observed in 2008, 2013, and 2018, into distinct groups so that provinces within each cluster are quite similar
to each other, according to their electoral results, while provinces in different groups are quite different from
each other. In detail, the similarity is measured by Euclidean (straight-line) distance and Manhattan (city-
block) distance, computed on non-scaled electoral results. We use four linkage methods: the single linkage
(i.e., the minimum distance), complete linkage (i.e., the maximum distance), average linkage (i.e., the average
distance), and Ward's linkage (i.e., the smallest increase in error sum of squares). We apply a hierarchical
clustering by building a dendrogram, i.e., a tree-like visual representation of the clustering, and decide where
to cut the dendrogram, i.e., the number of the clusters, by looking at the heights of the branches of the tree,
which indicate how different are the clusters that are joined from time to time, starting from the bottom of the
tree where observations are represented as leaves of the tree. Finally, the selected number of clusters is
validated by statistical analysis, and the clusters obtained are characterized by considering the average values
18 An example of this case are Forza Italia and Lega. While Forza Italia participated in the government shortly before the
2018 elections, the Lega remained in the opposition.
27
of abstention and votes obtained by parties and coalitions in each group. An F test is used to verify the
significance of the variables used for clustering.
5.4 FACTOR ANALYSIS
In the third step of our analysis, we apply exploratory factor analysis to the data set of demographic and
socio-economic variables to identify the latent factors that may affect abstention and voting behavior. The
factor analysis, developed using the principal component factor as a method of estimation, allows us to uncover
the underlying structure of the set of variables, i.e., the underlying relationships between the variables, and to
reduce the dimensionality of the data set. Applying the Kaiser-Guttman criterion, we retain all factors with
eigenvalues greater than one. We apply the factor analysis to the data set of the demographic and socio-
economic variables measured in 2017, to obtain factors that are orthogonal to each other in this year. Finally,
we calculate the factor scores relating to 2012, to be able to obtain the changes between 2012 and 2017.
5.5 REGRESSIONS ANALYSIS ON FACTOR SCORES
In the fourth and last step of our analysis, we estimate a province OLS regression model to investigate the
determinants of abstention and voting for populist parties in the 2018 Italian general elections.
We specify the baseline linear model as follows:
,2018 =1+1 ,2017 + (1)
where yi,2018 is the share, expressed in percentage points, of citizens entitled to vote who abstained or voted
for a given party, i.e., M5S, Lega, or PD, in province i in 2018, and Fi,2017 are the factor scores for province i
in 2017. Factor scores are lagged at time t-1, i.e., 2017, to limit problems of reverse causality.
Through specification (1), we estimate the drivers of abstention and voting behavior in the 2018 general
elections. However, because both M5S and Lega have been opposed to the government in office in the period
2013-2017, which has been led by PD, a possible problem of reverse causality arises more for the PD than for
the two populist parties on which our analysis focuses. To limit reverse causality stemming from this aspect,
we also estimate the following additional model:
,2018 =2+2 ,2017 +2,2013 + (2)
in which we add the lagged dependent variable in 2013 (i.e., the results of the previous general elections)
as a control.
With similar purposes, we further enrich the model including the change of the factor scores between 2012
and 2017, in addition to their level in 2017.
,2018 =3+3 ,2017 +3,2013 +3∆,2017−2012 + (3)
28
In summarizing, baseline specification (1) aims at estimating the relevance of social and economic factors
on both abstention and voting behavior in the 2018 general elections, and specification (2) allows us to check
whether these factors also drive the variations of abstention and voting behavior between 2013 and 2018, and
specification (3) allows us to check the relevance of the variation of the factor scores between 2012 and 2017
in addition to their level in 2017.
5.6 SUPPLEMENTAL ANALYSIS
Whereas the main analysis is performed with OLS regressions on factor scores, we also performed two
supplemental analyses. The first, more aligned with the mainstream empirical literature, employs OLS
regressions on a smaller set of demographic and socio-economic variables, which are considered important in
the literature, taking as reference the study done by Dijkstra et al. (2020). The second supplemental analysis
employs fixed effects panel regressions on the same set of selected variables. These supplemental analyses
aim to validate the methodological choices illustrated in the previous sections.
Going to the results, which are presented in the Supplemental material only, the first supplemental analysis
(Appendix 6) shows that the choice of considering the two populist parties separately is crucial to
understanding the Italian case and that the use of the factor analysis allows a deeper understanding of abstention
and voting decisions. The second supplemental analysis (Appendix 7) shows that it is also crucial to focus on
the results of the 2018 general elections. The reason is simple: between elections, voters have the opportunity
to observe the action of parties and to change their views on them. This alters the relationship between
demographic and socio-economic variables and voting decisions. Leaving aside this fact leads to unreliable
results on the relationship between the voting decisions and the variables considered, because of the implicit
assumption of an invariant relationship.
6. RESULTS
6.1 CORRELATION ANALYSIS
For each election, Table 3 reports the correlation matrixes between the electoral outcomes of the four main
parties, i.e., FI, Lega, PD, and M5S, and abstention.19 Correlations reveal particularly informative and
sometimes uncover unexpected aspects.
In 2008, abstention is significantly negatively correlated with voting for PD and Lega, while it is positively
correlated with voting for FI, which means that abstention is a substitute for voting for PD and Lega, while it
is a complement with voting for FI. As expected, there is a negative correlation between PD and FI, whereas
the negative correlation between Lega and FI, is quite unexpected, as well as the absence of a significant
correlation between voting PD and Lega. This suggests that in 2008 both PD and FI, on one side, and Lega
and FI, on the other side, were both considered substitutes by voters, while there was no such competition
between PD and Lega.
19 Corresponding correlation plots are reported in Appendix 1.
29
In 2013, when the M5S adds to the political offer, there are some significant changes. In this election, many
correlation coefficients are not significant anymore. Abstention is still significantly and negatively correlated
with voting for PD and Lega, while now there is no significant correlation between voting for FI and M5S,
which means that abstention is a substitute for voting for PD and Lega, while there is no correlation between
voting for FI or M5S. The negative correlation between PD and FI is still present, while now there is no
significant correlation between Lega and FI, and Lega and PD. Quite interesting is also the emergence of a
significant negative correlation between voting for M5S and Lega. This means that PD and FI are again
competing within a shared bunch of voters, while there is no such competition between Lega and FI, and
between Lega and PD: the possibility of transferring voting from PD to Lega, or from Lega to FI, and vice
versa is negligible in 2013. Finally, there is an interesting competition for voting between Lega and M5S.
In 2018, when the Lega becomes present throughout the national territory, the situation changes again in
an interesting way. All correlation coefficients are now significant. Abstention is still significantly negatively
correlated with voting for PD and Lega, while now there is a positive significant correlation between voting
for FI and for M5S, which means that abstention is a substitute for voting for PD and Lega, while it is
complementary with voting for FI or M5S. The negative correlation between PD and FI is still present, while
now there is a significant negative correlation between PD and M5S. Moreover, there is a significant negative
correlation between FI and Lega, and a significant positive correlation between FI and M5S. Finally, voting
for Lega and M5S is significantly and highly negatively correlated. This means that PD and FI are again
competing within a shared bunch of voters, but now there is also competition between M5S and PD and
between Lega and FI while voting for FI complements voting for M5S. Finally, the competition for voting
between Lega and M5S is confirmed. This means that the possibility of transferring voting from PD to Lega
or FI or M5S and vice versa in 2018 is possible, while there is no such possibility between FI and M5S.
These results show how voters’ political attitudes changed trough time from 2008 to 2018: while
maintaining some invariant aspects, such as the substitutability between abstention and voting for PD and
Lega, the emergence of M5S as a key player changes many aspects of the voters’ choices. A significant
negative correlation means that voters’ intentions may switch from one party to the other depending on the
underlying determinants, hence we expect these factors to affect the voting choices oppositely. Finally, notice
that substitutability makes it difficult to cooperate between parties because they compete on the same bunch
of voters, which might partially explain the fragility of the 2018 coalition between Lega and M5S.
30
Table 3 - Political elections 2008, 2013, and 2018, correlation matrixes
Political elections 2008
Abstention
PD
Forza Italia
Lega
Abstention
1.000***
-0.457***
0.273***
-0.559***
(0.000)
(0.000)
(0.004)
(0.000)
Partito Democratico
-0.457***
1.000***
-0.289***
-0.200
(0.000)
(0.000)
(0.002)
(0.037)
Forza Italia
0.273***
-0.289***
1.000***
-0.411***
(0.004)
(0.002)
(0.000)
(0.000)
Lega
-0.559***
-0.200
-0.411***
1.000***
(0.000)
(0.037)
(0.000)
(0.000)
Political elections 2013
Abstention
PD
Forza Italia
Lega
M5S
Abstention
1.000***
-0.675***
0.174
-0.436***
-0.207
(0.000)
(0.000)
(0.071)
(0.000)
(0.031)
Partito Democratico
-0.675***
1.000***
-0.380***
0.022
0.165
(0.000)
(0.000)
(0.000)
(0.817)
(0.086)
Forza Italia
0.174
-0.380***
1.000***
-0.002
-0.019
(0.071)
(0.000)
(0.000)
(0.982)
(0.841)
Lega
-0.436***
0.022
-0.002
1.000***
-0.277***
(0.000)
(0.817)
(0.982)
(0.000)
(0.004)
Movimento 5 Stelle
-0.207
0.165
-0.019
-0.277***
1.000***
(0.031)
(0.086)
(0.841)
(0.004)
(0.000)
Political elections 2018
Abstention
PD
Forza Italia
Lega
M5S
Abstention
1.000***
-0.768***
0.434***
-0.766***
0.558***
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
Partito Democratico
-0.768***
1.000***
-0.409***
0.468***
-0.481***
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
Forza Italia
0.434***
-0.409***
1.000***
-0.345***
0.402***
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
Lega
-0.766***
0.468***
-0.345***
1.000***
-0.817***
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
Movimento 5 Stelle
0.558***
-0.481***
0.402***
-0.817***
1.000***
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
6.2 CLUSTER ANALYSIS
Cluster analysis of political election patterns in 2008, 2013, and 2018 leads to the identification of four
clusters of the Italian provinces based on the similarities of their political election patterns.20
These four clusters (Table 4) are Cluster 1, including the provinces with a prevalence of FI and Center-
right prevalence; Cluster 2 including the provinces with a prevalence of the M5S and abstention; Cluster 3
including the provinces with a prevalence of Lega and Center-right, and Cluster 4 including the provinces with
a prevalence of PD and Center-left.
20 See Appendix 2 in the Supplemental material for the details. All linkage criteria suggest the choice of four clusters. The
statistical analysis, in particular the sharp decrease of pseudo-F in the shift from one class to the next, and a relatively
high value of pseudo-t-squared going from class 3 to 4 confirms this choice (Table A2.1). Furthermore, according to
ANOVA tests all variables involved in the cluster analysis are significant (Table A2.3).
31
Table 4 – Characterization of clusters
Cluster 1
Cluster 2
Cluster 3
Cluster 4
FI and Centre-right
M5S and Abstention
Lega and Centre-right
PD and Centre-left
“Blue”
“Yellow”
“Green”
“Red”
No.
Mean
No.
Mean
No.
Mean
No.
Mean
Abstention
108
22.96
45
36.53
59
26.58
118
26.91
Extreme left
108
0.86
45
0.73
59
0.99
118
0.41
Center-left
108
32.89
45
11.07
59
15.55
118
24.48
Liberals
108
0.25
45
1.58
59
2.22
118
6.42
Center-right
108
39.80
45
19.06
59
30.40
118
21.01
Extreme right
108
2.19
45
0.76
59
1.25
118
0.89
Partito Democratico
108
26.21
45
8.61
59
13.38
118
18.94
Forza Italia
108
28.58
45
11.56
59
9.07
118
14.09
Lega
108
5.95
45
3.74
59
17.32
118
3.84
Movimento 5 Stelle
108
0
45
26.64
59
19.01
118
18.19
The maps of these four clusters (Figure 7) provide an immediate representation of the geographical
distribution of the political earthquake that hit Italy in 2018. The elections in 2008 and 2013 rewarded Forza
Italia before, and then the Democratic Party, continuing the alternation between the two parties that has
characterized Italian politics since 1994, based on the rule that the ruling party loses the next elections. In
2013, there is also a first affirmation of the M5S in Sicily. The picture changes completely in 2018. The Lega
in the center-north and the M5S in the center-south won the elections, while the PD prevailed in some of the
provinces of central Italy with an old communist tradition. The country's political representation is shattered
along historical borders.21 The geographical fragmentation corresponds to political fragmentation, with the
transition from a bipolar Forza Italia-PD system to a tripolar Lega-PD-M5S system (Figure 8).
The marked regional differences in voting call into question the traditional geopolitical divisions of Italy.
Some have noted they are fading away (Agnew & Shin, 2017). For instance, in 2008 PD was globally residual,
in 2013 it was globally dominant, whereas in 2018 it was competitive only in some parts of the territory where
it has been the most voted party for many years, possibly due to the equilibrium between the M5S (which was
dominant in the South) and the Lega (which led the way in the North) in this “intermediate area”. The markedly
territorial nature of the vote reflects the importance given by voters to the demographic and socio-economic
trends that we will analyze in the next section.
21 For example, the provinces where the M5S is successful coincide quite precisely, excluding Sardinia, with the Kingdom
of the Two Sicilies.
32
Figure 7 – Maps of clusters, 2008-2018
Figure 8 – Plot of clusters in the Center-right vs. Center-right plane, 2008-2018
33
6.3 FACTOR ANALYSIS
Based on the varimax rotated factor loadings and characterization of factors (Table 5), and the scree plot of
eigenvalues after factors22, the factor analysis developed using the principal component factor as a method of
estimation allows us to characterize nine factors, which are by construction orthogonal in 201723. We interpret
these factors, sorted in decreasing order of explained variance, as follows: Economic well-being (F1), Crime
in densely populated areas (F2), Demographic growth (F3), Crime in less industrialized areas (F4), Organized
crime violence (F5), Arsons and extortions in areas with high emigration (F6), Government management of
uncontrolled immigration (F7), Crimes against women (F8), and House robberies (F9). The cumulative
proportion of the total variance of the data set explained by these nine factors is 77.8% (Table 5). Figure 9
shows the geographical distribution of the scores for each factor in 2017.24
Economic well-being (F1) is the most important factor, explaining 38% of the variance of the data set
(Table 5). This factor captures the multi-dimensionality of economic well-being and shows that many of the
variables usually considered in the explanations of the vote for populist parties represent different aspects of
the same phenomenon. Higher economic well-being corresponds with a higher per capita added value, lower
unemployment and higher average wages of employees, lower income inequality, a higher average per-capita
wealth, a greater share of the added value of manufacturing and higher exports, and a greater household
financial strength. Furthermore, higher economic well-being corresponds with a higher growth rate of
population, thanks to greater immigration also from abroad, higher participation in the labor market, a smaller
difference between male and female participation in the labor market, a population having a higher level of
education, also thanks to the ability to attract young graduates, and a greater circulation of newspapers. The
multi-dimensionality of economic well-being resulting from the factor analysis is the result of well-known
virtuous and vicious processes of cumulative circular causation (Myrdal, 1957), which give the geographical
representation of the scores of this factor the trend described in Figure 9, with increasingly lower values as you
move from North to South Italy. Economic well-being is also associated with some forms of micro-crime and
with greater emigration abroad, in particular of skilled workers.
The factor Demographic growth (F3) shows that the population trend cannot be reduced to economic well-
being. In particular, it is possible to identify two distinct areas with higher population growth in relative terms:
one in Southern Italy, where the population grows more thanks to the higher birth rate, and one in Northern
Italy where the population grows more, mainly thanks to immigration (Figure 9).
22 See Figure A4.1 in Appendix 4 in the Supplemental material.
23 See Table A4.3 and A4.4 in Appendix 4 in the Supplemental material.
24 Appendix 4 provides more details on the factor analysis and its results. In particular, it contains graphic and
geographical representations of the factor scores for both 2012 and 2017.
34
Table 5 - Varimax rotated factor loadings and characterization of factors
Variable
Factor 1
Factor 2
Factor 3
Factor 4
Factor 5
Factor 6
Factor 7
Factor 8
Factor 9
Communality
Uniqueness
1. Arsons
-0.384
0.682
0.709
0.291
2. Attempted homicides
-0.380
0.328
0.593
0.680
0.320
3. Bag theft
0.917
0.868
0.132
4. Home burglaries
0.517
0.339
-0.314
0.428
0.767
0.233
5. Drug-related crimes
0.344
0.682
0.644
0.356
6. Extortions
0.311
0.319
0.379
0.329
0.564
0.436
7. House robberies
0.482
0.671
0.750
0.250
8. Intentional homicides
0.804
0.778
0.222
9. Mafia homicides
0.345
0.606
0.734
0.266
10. Micro criminality
0.403
0.796
0.904
0.096
11. Prostitution-related crimes
0.757
0.697
0.303
12. Sexual violence
0.369
0.474
0.340
0.471
0.773
0.227
13. Robbery
0.901
0.890
0.110
14. Robbery homicides
0.384
0.474
-0.323
0.398
0.692
0.308
15. Fertility rate
0.481
0.418
0.529
0.809
0.191
16. Total growth rate of population
0.781
0.456
0.893
0.107
17. Population between 15 and 64 years
-0.505
0.790
0.908
0.092
18. Population over 64 years
0.319
-0.875
0.933
0.067
19. Population density
0.655
-0.335
0.738
0.262
20. Total immigration
0.869
0.887
0.113
21. Isolation (highways, airports, and ports)
-0.491
0.421
-0.329
0.645
0.355
22. Participation in the labor market
0.882
0.911
0.089
23. Participation in the labor market: difference between men
and women
-0.804
0.325
0.812
0.188
24. Exports per capita
0.668
-0.482
0.751
0.249
25. Income inequality
-0.656
0.391
0.697
0.303
26. Non-performing entry rate of loans to households
-0.648
0.646
0.354
27. Unemployment: job seekers aged 15 and over
-0.805
0.717
0.283
28. Value added: manufacturing
0.601
-0.668
0.881
0.119
29. Value added: public sector
-0.761
0.426
0.878
0.122
30. Value added: per capita
0.907
0.895
0.105
35
Variable
Factor 1
Factor 2
Factor 3
Factor 4
Factor 5
Factor 6
Factor 7
Factor 8
Factor 9
Communality
Uniqueness
31. Median gross hourly wage of employees born abroad
0.639
-0.364
0.352
0.747
0.253
32. Median gross hourly wage of employees born in Italy
0.869
0.893
0.107
33. Mean wage of employees
0.833
0.874
0.126
34. Mean wealth per capita
0.870
0.868
0.132
35. Population having at least a secondary degree
0.699
-0.323
0.762
0.238
36. Immigration of graduates between 25 and 39 years
0.848
0.844
0.156
37. Foreign residents
0.763
0.768
0.232
38. Emigration to other Italian regions
-0.335
0.765
0.745
0.255
39. Emigration abroad
0.413
0.579
0.656
0.344
40. Beds in emergency residences for migrants
0.671
0.557
0.443
41. Newspaper circulation
0.708
0.747
0.253
Eigenvalues
15.569
4.517
2.981
2.473
1.607
1.352
1.330
1.056
1.027
Difference
11.051
1.537
0.508
0.866
0.255
0.022
0.273
0.030
Proportion
0.380
0.110
0.073
0.060
0.039
0.033
0.032
0.026
0.025
Cumulative proportion
0.380
0.490
0.563
0.623
0.662
0.695
0.728
0.753
0.778
Explained variance
13.468
4.209
2.890
2.691
2.354
1.903
1.640
1.453
1.302
Number of variables
41.000
Number of retained factors
9.000
Notes. Factor loadings below 0.3 are omitted. The full table and all the details of the factor analysis performed are contained in Appendix 4.
36
Figure 9 - Maps of factor scores in 2017
37
Notes. This figure shows the geographical distribution of the scores in 2017 for each of the nine factors identified with the exploratory factor analysis. Higher scores are indicated with a darker
green (red) when they are related to a factor that can be interpreted as having a positive (negative) meaning.
38
The presence of the factor Government management of uncontrolled immigration (F7) reflects the approach
given by the government to the hospitality program for irregular immigrants, showing that it has been
successful in obtaining a distribution in the national territory as homogeneous as possible with the resident
population. However, it should be noted that some local administrations have refused to participate in the
government program or have given very limited availability.
Finally, the factor analysis has highlighted some types of crime (F2, F4, F5, F6) that have a particular
geographic distribution, distinct from those of the other factors. It is interesting to note that some of these are
the types of crime (i.e., organized crime violence, crimes against women, crime in densely populated urban
areas, arsons, and extortions in Southern provinces) that are most often discussed in the mass media and often
feed the political controversy.
6.4 REGRESSION ANALYSIS ON FACTOR SCORES
Results of the regression analysis of voting behavior on factor scores are reported in Table 6.
Abstention (Table 6 Column (1)-(3)), is associated with three factors, as shown by the positive and
significant coefficients in all the specifications: Crime in less industrialized areas (F4), Organized crime
violence (F5), and Government management of uncontrolled immigration (F7). However, the factor with the
highest coefficient in absolute value (with a negative sign, and significant only in the first specification) is
Economic well-being (F1), which looks as an important determinant of abstention but does not explain its
increase between 2013 and 2018. Based on these results, the drivers of abstention can be considered as three
long-term failures of the Italian State: the lack of socio-economic development, the persistence of organized
crime, and the inability to limit irregular immigration from poor countries. As such, abstention can be
considered as a sign of distrust in the Italian political system as a whole, as a “none of the above” vote, and it
signals a demand for protection that is not satisfied by the available political offer.
It is interesting to note that the factor Government management of uncontrolled immigration (F7), reports
a significant parameter with a positive sign only for abstention, while it always has a negative sign, when
significant, for all parties. This indicates that the government policy for the management of irregular
immigration has penalized all parties, through its repercussions on the territories where immigrants and asylum
seekers were hosted. A possible explanation is that the activation of the program required the consent of the
local administrations, in which both the M5S and the Lega may have been involved.
Even when voting for parties is considered (Table 6 Column (4)-(15)), Economic well-being (F1) is the
most important factor, having the greatest absolute value and being significant for all specifications and all
parties. The signs and the values of the coefficients show that the M5S collects more votes in the most
backward provinces, the Lega in the most advanced ones, and the PD and FI in those set in an intermediate
position. The provinces in which the M5S collects the most votes are also characterized by higher demographic
growth (F3) and higher crime rates (F2, F4, and F6). These results correspond with the importance given by
the movement both to measures in favor of the poorest people and for the restoration of legality, recognized as
contrasting crime at all levels.
39
Table 6 –Regression Analysis on Factor Scores
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
(13)
(14)
(15)
Abstention
Abstention
Abstention
M5S
M5S
M5S
Lega
Lega
Lega
PD
PD
PD
FI
FI
FI
F1 Economic well-being
-3.973***
-0.320
-0.473
-4.531***
-4.327***
-4.551***
5.106***
2.722***
2.751***
2.962***
0.825***
0.796***
-1.311***
-0.784***
-0.621***
(0.266)
(0.265)
(0.357)
(0.266)
(0.251)
(0.294)
(0.437)
(0.259)
(0.314)
(0.258)
(0.176)
(0.243)
(0.156)
(0.126)
(0.133)
F2 Crime in densely
populated areas
-0.438*
-0.025
-0.019
0.697***
0.423**
0.760***
-0.862**
-0.222
-0.435
0.796***
0.254**
0.252
0.252
-0.109
-0.142
(0.237)
(0.146)
(0.142)
(0.199)
(0.212)
(0.241)
(0.341)
(0.230)
(0.282)
(0.236)
(0.113)
(0.159)
(0.193)
(0.136)
(0.151)
F3 Demographic growth
1.146***
0.297
0.385
0.900***
1.868***
1.897***
-1.170***
-1.966***
-1.870***
-1.286***
-0.049
-0.064
0.472***
0.315***
0.224**
(0.239)
(0.223)
(0.252)
(0.222)
(0.230)
(0.259)
(0.427)
(0.214)
(0.206)
(0.280)
(0.129)
(0.153)
(0.163)
(0.107)
(0.108)
F4 Crime in less
industrialized areas
2.028***
1.067***
1.130***
0.600**
0.572**
0.789**
-2.496***
-0.865***
-0.922***
-0.447*
-0.490***
-0.573***
-0.400**
0.090
0.029
(0.233)
(0.230)
(0.268)
(0.262)
(0.247)
(0.307)
(0.450)
(0.233)
(0.272)
(0.227)
(0.102)
(0.129)
(0.162)
(0.127)
(0.137)
F5 Organized crime violence
1.667***
0.494***
0.553***
-0.175
0.422
0.392
-0.910***
-0.808***
-0.653***
-0.734***
-0.393***
-0.460***
0.217
0.435**
0.373***
(0.282)
(0.174)
(0.152)
(0.385)
(0.317)
(0.302)
(0.185)
(0.139)
(0.166)
(0.226)
(0.131)
(0.144)
(0.172)
(0.184)
(0.112)
F6 Arsons and extortions in
areas with high emigration
0.446*
-0.186
-0.127
0.313
0.678***
0.580**
-1.327***
-0.315*
-0.299
-0.091
-0.128
-0.143
0.419**
0.350**
0.374***
(0.238)
(0.145)
(0.165)
(0.253)
(0.225)
(0.244)
(0.270)
(0.177)
(0.190)
(0.204)
(0.112)
(0.142)
(0.175)
(0.135)
(0.121)
F7 Government management
of uncontrolled immigration
1.185***
0.908***
0.972**
-1.116***
-1.016***
-1.308***
0.356
-0.436**
-0.269
-0.837***
0.042
0.195
-0.174
0.161
0.071
(0.253)
(0.287)
(0.426)
(0.222)
(0.236)
(0.347)
(0.584)
(0.180)
(0.269)
(0.285)
(0.135)
(0.204)
(0.164)
(0.104)
(0.183)
F8 Crimes against women
-0.001
-0.113
-0.143
0.435*
0.238
-0.164
-0.311
0.024
0.499**
-0.373*
-0.080
0.029
0.061
-0.129
-0.238*
(0.255)
(0.110)
(0.158)
(0.227)
(0.194)
(0.312)
(0.282)
(0.160)
(0.208)
(0.214)
(0.108)
(0.137)
(0.115)
(0.081)
(0.124)
F9 House robberies
-0.268
-0.048
0.036
0.351
0.068
-0.082
-0.310
0.333*
0.313
0.747***
0.076
0.043
0.093
-0.010
0.030
(0.252)
(0.144)
(0.137)
(0.255)
(0.239)
(0.240)
(0.330)
(0.195)
(0.233)
(0.254)
(0.119)
(0.152)
(0.153)
(0.106)
(0.117)
Constant
29.684***
9.535***
9.661***
21.819***
11.312***
11.636***
12.278***
9.161***
8.469***
12.531***
0.334
0.306
9.507***
1.596**
1.260*
(0.247)
(1.554)
(2.032)
(0.284)
(1.746)
(1.961)
(0.364)
(0.294)
(0.765)
(0.272)
(0.873)
(0.954)
(0.152)
(0.688)
(0.678)
Lagged dependent variable,
2013
NO
YES
YES
NO
YES
YES
NO
YES
YES
NO
YES
YES
NO
YES
YES
Factor score changes, 2017-
2012
NO
NO
YES
NO
NO
YES
NO
NO
YES
NO
NO
YES
NO
NO
YES
No. of observations
110
110
110
110
110
110
110
110
110
109
109
109
109
109
109
R-squared
.808
.938
.942
.746
.817
.851
.736
.914
.925
.645
.929
.935
.514
.764
.815
F test
51.6***
255***
145***
57.3***
72.3***
44.8***
59.9***
145***
95.5***
20.1***
77.9***
58.3***
11.3***
27.7***
21.5***
Notes. Robust standard errors in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01.
40
As far as the Lega is concerned, is interesting to note that it collects more votes where population growth
is lower, i.e., in provinces where immigration is lower, given that this is the main factor of demographic growth
in the provinces of northern Italy. Furthermore, the parameter associated with the factor Crimes against women
(F8) is positive and significant in the third specification (Col. (9)), indicating that the vote for the Lega may
have been driven by a demand for greater protection, which is a major focus on the party's agenda.
The fit of the models can be considered more than satisfactory in all specifications and for all the dependent
variables.
7. DISCUSSION AND CONCLUSIONS
This paper aimed to understand the demographic and socio-economic drivers of the electoral success in
Italy in 2018 of two parties, i.e., M5S and Lega, usually considered left-wing and right-wing populists, and of
the decreasing electoral turnout. Our contribution consists of the use of an innovative empirical methodology
applied to the above problem.
For this purpose, we first provided detailed descriptive evidence useful to frame the problem, then we used
hierarchical cluster analysis to unfold the natural groupings of the provinces based on the similarities of their
political election results in 2008, 2013, and 2018. We identified four clusters, one with FI and center-right
prevalence, a second with M5S and abstention prevalence, a third with Lega and center-right prevalence, and
finally a cluster with PD and center-left prevalence. This evidence was particularly insightful in raising
concerns about abstention choices.
Then, we proceeded with factor analysis to characterize nine factors out of 41 demographic and socio-
economic variables that are likely to influence voting behavior. These factors encompass different phenomena,
such as economic well-being, demographic growth, organized crime violence, government management of
irregular immigration, and some types of crimes that are subject to political debate, such as extortion, home
robbery, and crimes against women. One specific advantage of factorizing a large number of available
variables into nine main factors is to allow bypassing the non-conclusive contraposition between cultural or
economic determinants of populist electoral success. Finally, we performed a regression analysis to understand
the extent to which each factor affected voters’ electoral preferences.
In conclusion, this paper offered a methodological contribution that simultaneously uses machine learning
techniques (cluster analysis and factor analysis) and more traditional regression techniques to explain the
drivers of voting behavior, with specific attention to the phenomenon of abstention. We applied this
methodology using an original dataset including a set of indicators much larger than those normally employed
by other scholars in the field. We leave to future research the task of improving the analysis through a perfect
identification of the relationships of main interest.
Results show that an important phenomenon emerging from the Italian 2018 general elections, and strongly
intertwined with the persistence of electoral instability, is the resurgence of territory as one of the major
elements affecting voting patterns. However, when only the national context is considered, Italy does not
exactly fit into the narrative that sees populism as revenge for places that don't matter or have been left behind,
as proposed, for example, by Rodríguez-Pose (2018). With this perspective, it would not be possible to explain
41
the success of the Lega in the most economically advanced provinces of Northern Italy. However, if we
consider Italy in the international context, the whole country has lagged behind the other main European
countries, and its political and economic weight in the international context, and Europe in particular, has
decreased. In this broader context, the success of the M5S, on the one hand, can be seen as the revenge of the
Southern provinces, which have remained behind and marginalized compared to the rest of the country. The
success of the Lega, on the other hand, can be seen as the revenge of the Northern provinces, whose reference
points are the most advanced European regions with which the gap has widened. The holding of the PD in
some central provinces can be considered as an intermediate position of temporary satisfaction with the status
quo, especially by older people.
In the Italian case, the mainstream parties are besieged by populism that comes from both the richer and
poorer parts of society. On the one hand, this particularity of the Italian case indicates that the interpretative
category of populism should be qualified, not being able to fully convey the complexity of this phenomenon.
On the other hand, it indicates that the Italian case has European relevance since the success of the Lega is at
least in part the result of a contestation of European policies, supported and endorsed by the mainstream parties,
which come from a large part of the voters of the more economical advanced Italia provinces. As regards the
ever-decreasing participation in the vote, our analysis highlighted that abstention is associated with three
factors that can be considered three long-term failures of the Italian State: the lack of socio-economic
development and security, the persistence of organized crime, and the inability to limit irregular immigration
from poor countries. As such, abstention can be considered as a sign of distrust in the Italian political system
as a whole, as a “none of the above” vote, and a demand for protection not satisfied by the political offer.
Despite some measures, it cannot be said that the expectations of the citizens who voted for populist parties
in 2018 have been met, and the latest emergency government chaired by Mario Draghi, supported by all Italian
parties except Fratelli d’Italia (FdI, Brothers of Italy), testifies to the normalization of these parties. Part of the
consensus gathered in 2018 by Lega and M5S, and that these two parties will lose, can flow into FdI, and part
into abstention.
We think the most likely outcome will be a further drastic increase in abstention, which could lead to a new
unpredictable balance between mainstream and populist parties. The main unknown factor is the possibility
that new parties can be formed and stand for election. With a higher abstention, even small shifts in consensus
can prove decisive.
8. REFERENCES
Agnew, J., & Shin, M. (2017). Spatializing Populism: Taking Politics to the People in Italy. Annals of the
American Association of Geographers, 107(4), 915–933.
https://doi.org/10.1080/24694452.2016.1270194
Angrist, J. D., & Pischke, J.-S. (2009). Mostly harmless econometrics: An empiricist’s companion. Princeton,
NJ: Princeton University Press.
Baldwin, R., & Giavazzi, F. (Eds.). (2015). The Eurozone Crisis: A Consensus View of the Causes and a
42
Few Possible Solutions. London: Centre for Economic Policy Research.
Benasayag, M., & Schmit, G. (2003). Les passions tristes: Souffrance psychique et crise sociale. Paris: La
Découverte.
Bobba, G., & Legnante, G. (2017). Italy: A Breeding Ground for Populist Political Communication. In T.
Aalberg, F. Esser, C. Reinemann, J. Stromback, & C. H. de Vreese (Eds.), Populist Political
Communication in Europe (pp. 221–234). New York, NY: Routledge.
Bobba, G., & McDonnell, D. (2015). Italy: a strong and enduring market for populism. In H. Kriesi & T. S.
Pappas (Eds.), European Populism in the Shadow of the Great Recession (pp. 163–180). Colchester:
ECPR Press.
Bolet, D. (2022). The janus-faced nature of radical voting: Subjective social decline at the roots of radical
right and radical left support. Party Politics, 0(0), 1–14. https://doi.org/10.1177/13540688221085444
Caiani, M. (2019). The populist parties and their electoral success: different causes behind different
populisms? The case of the Five-star Movement and the League. Contemporary Italian Politics, 11(3),
236–250. https://doi.org/10.1080/23248823.2019.1647681
Canovan, M. (2002). Taking Politics to the People: Populism as the Ideology of Democracy. In Y. Mény &
Y. Surel (Eds.), Democracies and the Populist Challenge (pp. 25–44). London: Palgrave Macmillan.
https://doi.org/10.1057/9781403920072_2
Colantone, I., & Stanig, P. (2018a). Global Competition and Brexit. American Political Science Review,
112(2), 201–218. https://doi.org/10.1017/S0003055417000685
Colantone, I., & Stanig, P. (2018b). The Economic Determinants of the “Cultural Backlash”: Globalization
and Attitudes in Western Europe. BAFFI CAREFIN Working Paper Series, 91.
Colantone, I., & Stanig, P. (2018c). The Trade Origins of Economic Nationalism: Import Competition and
Voting Behavior in Western Europe. American Journal of Political Science, 62(4), 936–953.
https://doi.org/10.1111/ajps.12358
Colantone, I., & Stanig, P. (2019). The Surge of Economic Nationalism in Western Europe. Journal of
Economic Perspectives, 33(4), 128–151. https://doi.org/10.1257/jep.33.4.128
Corbetta, P., Colloca, P., Cavazza, N., & Roccato, M. (2018). Lega and Five-star Movement voters:
exploring the role of cultural, economic and political bewilderment. Contemporary Italian Politics,
10(3), 279–293. https://doi.org/10.1080/23248823.2018.1524678
Dassonneville, R., & Hooghe, M. (2017). Economic indicators and electoral volatility: Economic effects on
electoral volatility in Western Europe, 1950-2013. Comparative European Politics, 15(6), 919–943.
https://doi.org/10.1057/cep.2015.3
De Benoit, A. (2017). Populismo. La fine della destra e della sinistra. Bologna: Arianna Editrice.
Di Maio, G., Gilli, M., & Dalla Pellegrina, L. (2022). Populist Voting as Search for Alternatives: The Case
of Italy. Mimeo.
Dijkstra, L., Poelman, H., & Rodríguez-Pose, A. (2020). The geography of EU discontent. Regional Studies,
54(6), 737–753. https://doi.org/10.1080/00343404.2019.1654603
43
Draghi, M. (2012, July 26). Speech at the Global Investment Conference in London. Retrieved July 29, 2022,
from https://www.ecb.europa.eu/press/key/date/2012/html/sp120726.en.html
Dustmann, C., Eichengreen, B., Otten, S., Sapir, A., Tabellini, G., & Zoega, G. (2017). Europe’s Trust
Deficit: Causes and Remedies. London: CEPR PRESS.
Galbraith, J. K. (1958). The Affluent Society. Boston, MA: Houghton Mifflin.
Gilli, M., & Manzoni, E. (2019). Populism, the Backlash against Ruling Politicians and the Possible
Malfunctioning of Representative Democracy. CefES-DEMS Working Paper Series, 417.
Guiso, L., Herrera, H., Morelli, M., & Sonno, T. (2017). Populism: Demand and Supply. Mimeo.
Inglehart, R. F., & Norris, P. (2016). Trump, Brexit, and the Rise of Populism: Economic Have-Nots and
Cultural Backlash. HKS Faculty Research Working Paper Series, 26, 1–52.
Itanes. (2018). Vox populi: Il voto ad alta voce del 2018. Bologna: Il Mulino.
Maraffi, M. (2018). The social stratification of the 2018 vote in Italy: between continuity and change.
Contemporary Italian Politics, 10(3), 267–278. https://doi.org/10.1080/23248823.2018.1531932
Morgan, S. L. (2018a). Correct Interpretations of Fixed-effects Models, Specification Decisions, and Self-
reports of Intended Votes: A Response to Mutz. Socius: Sociological Research for a Dynamic World, 4,
237802311881150. https://doi.org/10.1177/2378023118811502
Morgan, S. L. (2018b). Status Threat, Material Interests, and the 2016 Presidential Vote. Socius:
Sociological Research for a Dynamic World, 4, 1–17. https://doi.org/10.1177/2378023118788217
Mudde, C. (2004). The Populist Zeitgeist. Government and Opposition, 39(4), 541–563.
https://doi.org/10.1111/j.1477-7053.2004.00135.x
Mudde, C. (2016). On Extremism and Democracy in Europe. New York, NY: Routledge.
Mudde, C., & Kaltwasser, C. R. (2017). Populism. A very short introduction. Oxford: Oxford University
Press.
Mutz, D. C. (2018a). Response to Morgan: On the Role of Status Threat and Material Interests in the 2016
Election. Socius: Sociological Research for a Dynamic World, 4, 237802311880861.
https://doi.org/10.1177/2378023118808619
Mutz, D. C. (2018b). Status threat, not economic hardship, explains the 2016 presidential vote. Proceedings
of the National Academy of Sciences, 115(19), E4330–E4339. https://doi.org/10.1073/pnas.1718155115
Myrdal, G. (1957). Economic theory and under-developed regions. London: General Duckworth & Co.
Norris, P., & Inglehart, R. (2019). Cultural Backlash: Trump, Brexit, and Authoritarian Populism.
Cambridge: Cambridge University Press. https://doi.org/10.1017/9781108595841
Orsina, G. (2019). Genealogy of a Populist Uprising. Italy, 1979-2019. The International Spectator, 54(2),
50–66. https://doi.org/10.1080/03932729.2019.1603896
Pappas, T. S. (2019). Populism and Liberal Democracy: A Comparative and Theoretical Analysis. Oxford:
Oxford University Press. https://doi.org/10.1093/oso/9780198837886.001.0001
Rodríguez-Pose, A. (2018). The revenge of the places that don’t matter (and what to do about it). Cambridge
Journal of Regions, Economy and Society, 11(1), 189–209. https://doi.org/10.1093/cjres/rsx024
44
Tarchi, M. (2015). L’Italia populista: Dal qualunquismo a Beppe Grillo. Bologna: Il Mulino.
Trichet, J.-C., & Draghi, M. (2011, August 5). Letter to Italian Prime Minister. Retrieved July 29, 2022, from
https://www.ecb.europa.eu/ecb/access_to_documents/document/pa_document/shared/data/ecb.dr.par20
21_0001lettertoItalianPrimeMinister.en.pdf
Urbinati, N. (2019). Political Theory of Populism. Annual Review of Political Science, 22(1), 111–127.
https://doi.org/10.1146/annurev-polisci-050317-070753
Zakaria, F. (2012). Interview with Italian Prime Minister Mario Monti. Aired May 20, 2012 - 10:00 ET.
Retrieved July 29, 2022, from transcripts.cnn.com/show/fzgps/date/2012-05-20/segment/01%0AFareed
1
APPENDIX - SUPPLEMENTAL MATERIAL
Appendix 1 Political elections results, 2008-2018
Table A1.1 - Political elections 2008, 2013, and 2018: Results
Political elections 2008 2013 2018
Number
% of
citizens
% of
valid votes Number
% of
citizens
% of
valid votes Number
% of
citizens
% of
valid votes
Abstention and turnout
Citizens entitled to vote 47,142,436 100% 47,005,432 100% 46,604,896 100%
Abstention 10,617,017 22.5% 12,932,157 27.5% 14,955,989 32.1%
Turnout 36,525,420 77.5% 100% 34,073,272 72.5% 100% 31,648,908 67.9% 100%
Parties
Movimento 5 Stelle 8,702,987 18.5% 25.5% 10,252,280 22.0% 32.4%
Lega 3,026,844 6.4% 8.3% 1,392,537 3.0% 4.1% 5,587,146 12.0% 17.7%
Partito Democratico
12,092,998
25.7%
33.1%
8,644,542
18.4%
25.4%
5,887,357
12.6%
18.6%
Forza Italia 13,642,745 28.9% 37.4% 7,332,829 15.6% 21.5% 4,471,741 9.6% 14.1%
Fratelli d’Italia 668,886 1.4% 2.0% 1,398,109 3.0% 4.4%
Political areas and alignments
Extreme Left
378,116
0.8%
1.0%
95,150
0.2%
0.3%
480,285
1.0%
1.5%
Center-Left 15,343,652 32.5% 42.0% 10,852,847 23.1% 31.9% 7,085,809 15.2% 22.4%
Center-Liberals 103,760 0.2% 0.3% 3,364,715 7.2% 9.9% 971,815 2.1% 3.1%
Center-Right 19,130,396 40.6% 52.4% 10,180,386 21.7% 29.9% 11,905,528 25.5% 37.6%
Extreme Right
1,026,485
2.2%
2.8%
421,367
0.9%
1.2%
502,238
1.1%
1.6%
Movimento 5 Stelle 8,702,987 18.5% 25.5% 10,252,280 22.0% 32.4%
Notes. In 2008, Forza Italia and Alleanza Nazionale (which later became Fratelli d’Italia) stood in the elections together with a list called Il popolo delle libertà (The people of
freedom). The source of the data is the Italian Ministry of the Interior.
2
Table A1.2 - Political elections 2008, 2013, and 2018: Descriptive statistics
No.
Mean
Std. Dev.
Min.
1st Quartile
Median
3rd Quartile
Max
Abstention
330
26.9
6.53
14.7
22.1
26.2
30.9
44.1
Extreme left
330
.703
.498
0
.39
.63
1
3.28
Center-left
330
23.8
9.79
6.52
16.4
22.7
29.5
53.4
Liberals
330
2.99
3.16
0
.31
1.9
5.07
16.3
Center-right
330
28.6
10.2
6.35
20
26.5
36.2
55.6
Extreme right
330
1.36
.79
.31
.79
1.12
1.75
3.88
Partito Democratico
330
18.9
8.21
0
12.9
17.9
23.6
44.3
Forza Italia
330
17.6
8.73
0
10.6
15.1
25
43.3
Lega
330
6.93
7.67
0
.17
3.73
12.6
29
Movimento 5 Stelle
330
13.5
10.4
0
0
16.6
20.9
33.4
Notes. The total number of voters who abstained or did not cast a valid vote (abstention) and the total number of valid
votes obtained by each party or political area are expressed in percentage points as a share of citizens entitled to vote.
Table A1.3 - Political elections 2018: Descriptive statistics
No.
Mean
Std. Dev.
Min.
1st Quartile
Median
3rd Quartile
Max
Abstention
110
29.7
5.66
21.7
25.2
28.2
33.7
43.3
Extreme left
110
1.03
.528
.3
.62
.92
1.35
3.28
Center-left
110
15
5.37
6.52
11.1
14.5
17.5
36.9
Liberals
110
1.87
.76
0
1.32
1.91
2.32
4.03
Center-right
110
25.5
6.79
10.3
20.2
24.4
30.5
40.8
Extreme right
110
1.08
.355
.38
.84
1.02
1.28
2.56
Partito Democratico
110
12.4
4.68
0
8.84
12.2
14.9
27.7
Forza Italia
110
9.43
2.35
0
7.87
9.26
10.8
14.6
Lega
110
12.3
7.12
1.82
4.78
12.6
18.1
28.4
Movimento 5 Stelle
110
21.8
5.66
8.32
16.9
20.5
26.8
33.4
Notes. The total number of voters who abstained or did not cast a valid vote (abstention) and the total number of valid
votes obtained by each party or political area are expressed in percentage points as a share of citizens entitled to vote.
3
Figure A1.1 - Political elections 2008
Notes. The total number of voters who abstained or did not cast a valid vote (abstention) and the total number
of valid votes obtained by each party or political area are expressed in percentage points as a share of citizens
entitled to vote.
4
Figure A1.2 - Political elections 2013
Notes. The total number of voters who abstained or did not cast a valid vote (abstention) and the total number of valid votes obtained by each party or political area
are expressed in percentage points as a share of citizens entitled to vote.
5
Figure A1.3 - Political elections 2018
Notes. The total number of voters who abstained or did not cast a valid vote (abstention) and the total number of valid votes obtained by each party or political area
are expressed in percentage points as a share of citizens entitled to vote.
6
Figure A1.4 - Abstention, 2008-2018
Notes. The total number of voters who abstained or did not cast a valid vote is expressed in percentage points as a share of citizens entitled to vote.
7
Figure A1.5 - Movimento 5 Stelle, 2008-2018
Notes. The total number of valid votes obtained by the party is expressed in percentage points as a share of citizens entitled to vote.
8
Figure A1.6 - Lega, 2008-2018
Notes. The total number of valid votes obtained by the party is expressed in percentage points as a share of citizens entitled to vote.
9
Figure A1.7 - Partito Democratico, 2008-2018
Notes. The total number of valid votes obtained by the party is expressed in percentage points as a share of citizens entitled to vote.
10
Figure A1.8 - Forza Italia, 2008-2018
Notes. The total number of valid votes obtained by the party is expressed in percentage points as a share of citizens entitled to vote. In 2008, Forza Italia and
Alleanza Nazionale (which later became Fratelli d’Italia) stood in the elections together with a list called Il popolo delle libertà (The people of freedom).
11
Figure A1.9 - Lega and Movimento 5 Stelle, 2008-2018
Notes. The total number of valid votes obtained by the parties is expressed in percentage points as a share of citizens entitled to vote.
12
Figure A1.10 - Political elections 2008: Scatterplots
Figure A1.11 - Political elections 2013: Scatterplots
13
Figure A1.12 - Political elections 2018: Scatterplots
14
Figure A1.13 - Center-left vs. Center-right, 2008-2018: Scatterplot
Figure A1.14 - Lega vs. Forza Italia, 2008-2018: Scatterplot
15
Table A1.4 - Political elections 2008: Correlation matrix
2008
Abstention
Extreme left
Center-left
Liberals
Center-right
Extreme right
PD
Forza Italia
Lega
Abstention
1.000***
-0.452***
-0.453***
-0.013
-0.207
-0.585***
-0.457***
0.273***
-0.559***
(0.000)
(0.000)
(0.000)
(0.897)
(0.031)
(0.000)
(0.000)
(0.004)
(0.000)
Extreme left
-0.452***
1.000***
0.604***
0.436***
-0.333***
0.580***
0.718***
-0.103
-0.097
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.284)
(0.318)
Center-left
-0.453***
0.604***
1.000***
0.232
-0.770***
0.315***
0.853***
-0.447***
-0.274***
(0.000)
(0.000)
(0.000)
(0.015)
(0.000)
(0.001)
(0.000)
(0.000)
(0.004)
Liberals
-0.013
0.436***
0.232
1.000***
-0.258***
0.308***
0.248***
-0.083
-0.161
(0.897)
(0.000)
(0.015)
(0.000)
(0.007)
(0.001)
(0.009)
(0.388)
(0.095)
Center-right
-0.207
-0.333***
-0.770***
-0.258***
1.000***
0.027
-0.574***
0.319***
0.682***
(0.031)
(0.000)
(0.000)
(0.007)
(0.000)
(0.780)
(0.000)
(0.001)
(0.000)
Extreme right
-0.585***
0.580***
0.315***
0.308***
0.027
1.000***
0.412***
0.036
0.116
(0.000)
(0.000)
(0.001)
(0.001)
(0.780)
(0.000)
(0.000)
(0.707)
(0.231)
Partito Democratico
-0.457***
0.718***
0.853***
0.248***
-0.574***
0.412***
1.000***
-0.289***
-0.200
(0.000)
(0.000)
(0.000)
(0.009)
(0.000)
(0.000)
(0.000)
(0.002)
(0.037)
Forza Italia
0.273***
-0.103
-0.447***
-0.083
0.319***
0.036
-0.289***
1.000***
-0.411***
(0.004)
(0.284)
(0.000)
(0.388)
(0.001)
(0.707)
(0.002)
(0.000)
(0.000)
Lega
-0.559***
-0.097
-0.274***
-0.161
0.682***
0.116
-0.200
-0.411***
1.000***
(0.000)
(0.318)
(0.004)
(0.095)
(0.000)
(0.231)
(0.037)
(0.000)
(0.000)
16
Table A1.5 - Political elections 2013: Correlation matrix
2013
Abstention
Extreme left
Center-left
Liberals
Center-right
Extreme right
PD
Forza Italia
Lega
M5S
Abstention
1.000***
-0.160
-0.617***
-0.667***
-0.089
-0.103
-0.675***
0.174
-0.436***
-0.207
(0.000)
(0.096)
(0.000)
(0.000)
(0.356)
(0.287)
(0.000)
(0.071)
(0.000)
(0.031)
Extreme left
-0.160
1.000***
0.423***
-0.266***
-0.498***
0.179
0.452***
-0.130
-0.450***
0.348***
(0.096)
(0.000)
(0.000)
(0.005)
(0.000)
(0.063)
(0.000)
(0.178)
(0.000)
(0.000)
Center-left
-0.617***
0.423***
1.000***
0.124
-0.567***
-0.062
0.815***
-0.556***
-0.138
-0.049
(0.000)
(0.000)
(0.000)
(0.200)
(0.000)
(0.523)
(0.000)
(0.000)
(0.152)
(0.611)
Liberals
-0.667***
-0.266***
0.124
1.000***
0.403***
-0.124
0.257***
-0.129
0.754***
-0.073
(0.000)
(0.005)
(0.200)
(0.000)
(0.000)
(0.198)
(0.007)
(0.183)
(0.000)
(0.450)
Center-right
-0.089
-0.498***
-0.567***
0.403***
1.000***
0.089
-0.324***
0.659***
0.699***
-0.288***
(0.356)
(0.000)
(0.000)
(0.000)
(0.000)
(0.357)
(0.001)
(0.000)
(0.000)
(0.002)
Extreme right
-0.103
0.179
-0.062
-0.124
0.089
1.000***
-0.081
0.338***
-0.186
0.143
(0.287)
(0.063)
(0.523)
(0.198)
(0.357)
(0.000)
(0.405)
(0.000)
(0.053)
(0.139)
Partito Democratico
-0.675***
0.452***
0.815***
0.257***
-0.324***
-0.081
1.000***
-0.380***
0.022
0.165
(0.000)
(0.000)
(0.000)
(0.007)
(0.001)
(0.405)
(0.000)
(0.000)
(0.817)
(0.086)
Forza Italia
0.174
-0.130
-0.556***
-0.129
0.659***
0.338***
-0.380***
1.000***
-0.002
-0.019
(0.071)
(0.178)
(0.000)
(0.183)
(0.000)
(0.000)
(0.000)
(0.000)
(0.982)
(0.841)
Lega
-0.436***
-0.450***
-0.138
0.754***
0.699***
-0.186
0.022
-0.002
1.000***
-0.277***
(0.000)
(0.000)
(0.152)
(0.000)
(0.000)
(0.053)
(0.817)
(0.982)
(0.000)
(0.004)
Movimento 5 Stelle
-0.207
0.348***
-0.049
-0.073
-0.288***
0.143
0.165
-0.019
-0.277***
1.000***
(0.031)
(0.000)
(0.611)
(0.450)
(0.002)
(0.139)
(0.086)
(0.841)
(0.004)
(0.000)
17
Table A1.6 - Political elections 2018: Correlation matrix
2018
Abstention
Extreme left
Center-left
Liberals
Center-right
Extreme right
PD
Forza Italia
Lega
M5S
Abstention
1.000***
-0.333***
-0.643***
-0.675***
-0.664***
-0.471***
-0.768***
0.434***
-0.766***
0.558***
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
Extreme left
-0.333***
1.000***
0.531***
0.209
-0.238
0.033
0.557***
-0.433***
-0.088
-0.028
(0.000)
(0.000)
(0.000)
(0.029)
(0.012)
(0.729)
(0.000)
(0.000)
(0.358)
(0.772)
Center-left
-0.643***
0.531***
1.000***
0.515***
0.131
0.279***
0.813***
-0.647***
0.375***
-0.592***
(0.000)
(0.000)
(0.000)
(0.000)
(0.172)
(0.003)
(0.000)
(0.000)
(0.000)
(0.000)
Liberals
-0.675***
0.209
0.515***
1.000***
0.486***
0.194
0.641***
-0.249***
0.552***
-0.553***
(0.000)
(0.029)
(0.000)
(0.000)
(0.000)
(0.042)
(0.000)
(0.009)
(0.000)
(0.000)
Center-right
-0.664***
-0.238
0.131
0.486***
1.000***
0.481***
0.324***
-0.003
0.930***
-0.711***
(0.000)
(0.012)
(0.172)
(0.000)
(0.000)
(0.000)
(0.001)
(0.975)
(0.000)
(0.000)
Extreme right
-0.471***
0.033
0.279***
0.194
0.481***
1.000***
0.215
-0.312***
0.518***
-0.462***
(0.000)
(0.729)
(0.003)
(0.042)
(0.000)
(0.000)
(0.024)
(0.001)
(0.000)
(0.000)
Partito Democratico
-0.768***
0.557***
0.813***
0.641***
0.324***
0.215
1.000***
-0.409***
0.468***
-0.481***
(0.000)
(0.000)
(0.000)
(0.000)
(0.001)
(0.024)
(0.000)
(0.000)
(0.000)
(0.000)
Forza Italia
0.434***
-0.433***
-0.647***
-0.249***
-0.003
-0.312***
-0.409***
1.000***
-0.345***
0.402***
(0.000)
(0.000)
(0.000)
(0.009)
(0.975)
(0.001)
(0.000)
(0.000)
(0.000)
(0.000)
Lega
-0.766***
-0.088
0.375***
0.552***
0.930***
0.518***
0.468***
-0.345***
1.000***
-0.817***
(0.000)
(0.358)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
Movimento 5 Stelle
0.558***
-0.028
-0.592***
-0.553***
-0.711***
-0.462***
-0.481***
0.402***
-0.817***
1.000***
(0.000)
(0.772)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
18
Appendix 2 Cluster analysis
Figure A2.1 - Dendrograms with Euclidean distance
Figure A2.2 - Dendrograms with Manhattan distance
19
Figure A2.3 - Dendrograms with Ward's linkage
Table A2.1 – Choice of the number of clusters: statistics
Number of clusters
Calinski-Harabasz
pseudo-F L2
Duda-Hart pseudo-T-
squared L2
Calinski-Harabasz
pseudo-F L1
1
344.793
2
344.793
87.366
344.793
3
252.217
181.892
238.913
4
230.279
57.491
214.610
5
225.012
75.858
208.055
6
227.702
35.310
208.508
7
225.447
25.644
212.473
8
220.645
54.725
207.598
9
212.539
35.800
211.820
10
211.416
63.209
211.062
11
208.276
14.994
215.424
12
210.438
16.529
219.444
13
216.695
23.891
217.707
14
212.477
23.143
217.332
15
211.914
20.620
214.498
20
Table A2.2 – Cluster description
Cluster 1
Cluster 2
Cluster 3
Cluster 4
Total
“Blue”
“Yellow”
“Green”
“Red”
No.
Mean
No
Mean
No.
Mean
No.
Mean
No.
Mean
Abstention
108
22.96
45
36.53
59
26.58
118
26.91
330
26.87
Extreme left
108
0.86
45
0.73
59
0.99
118
0.41
330
0.70
Center-left
108
32.89
45
11.07
59
15.55
118
24.48
330
23.81
Liberals
108
0.25
45
1.58
59
2.22
118
6.42
330
2.99
Center-right
108
39.80
45
19.06
59
30.40
118
21.01
330
28.57
Extreme right
108
2.19
45
0.76
59
1.25
118
0.89
330
1.36
Partito Democratico
108
26.21
45
8.61
59
13.38
118
18.94
330
18.92
Forza Italia
108
28.58
45
11.56
59
9.07
118
14.09
330
17.59
Lega
108
5.95
45
3.74
59
17.32
118
3.84
330
6.93
Movimento 5 Stelle
108
0
45
26.64
59
19.01
118
18.19
330
13.54
Table A2.3 - ANOVA table
F_test
p-value
Abstention
77.87
0
Extreme left
30.34
0
Center-left
195.45
0
Liberals
266.69
0
Center-right
271.31
0
Extreme right
145.99
0
Partito Democratico
136.52
0
Forza Italia
473.05
0
Lega
77.72
0
Movimento 5 Stelle
918.58
0
21
Figure A2.4 - Map of clusters
22
Figure A2.5 – Plot of clusters in the Center-right vs. Center-left plane, 2008-2018
Figure A2.6 – Plot of clusters in the Movimento 5 Stelle vs. Center-left plane, 2013-2018
23
Figure A2.7 – Plot of clusters in the Movimento 5 Stelle vs. Abstention plane, 2013-2018
Figure A2.8 – Plot of clusters in the Movimento 5 Stelle vs. Center-right plane, 2013-2018
24
Appendix 3 Data description
Table 3.1 – Summary statistics of explanatory variables used for factor analysis
Variable Description No. Mean
Standard.
Deviation
Min.
1st
Quartile
Median
3rd
Quartile
Max
Crime
1. Arsons Reported crimes per 10,000 inhabitants 220 2.325 2.359 0.158 0.919 1.403 2.738 11.891
2. Attempted homicides Reported crimes per 10,000 inhabitants 220 0.214 0.157 0.000 0.109 0.181 0.277 1.114
3. Bag theft Reported crimes per 10,000 inhabitants 220 1.877 1.844 0.048 0.747 1.246 2.267 10.686
4. Home burglaries Reported crimes per 10,000 inhabitants 220 34.877 14.258 10.837 23.911 32.863 45.341 74.711
5. Drug-related crimes Reported crimes per 10,000 inhabitants 220 5.709 2.088 1.449 4.047 5.521 7.185 11.399
6. Extortions Reported crimes per 10,000 inhabitants 220 1.184 0.493 0.411 0.833 1.114 1.452 2.861
7. House robberies Reported crimes per 10,000 inhabitants 220 0.454 0.244 0.079 0.295 0.398 0.576 1.769
8. Intentional homicides Reported crimes per 10,000 inhabitants 220 0.079 0.086 0.000 0.031 0.059 0.102 0.735
9. Mafia homicides Reported crimes per 10,000 inhabitants 220 0.006 0.027 0.000 0.000 0.000 0.000 0.245
10. Micro criminality Reported crimes per 10,000 inhabitants 220 187 84.145 23.000 131.250 174.400 229.250 531.500
11. Prostitution-related crimes Reported crimes per 10,000 inhabitants 220 0.172 0.138 0.000 0.078 0.136 0.220 0.739
12. Sexual violence Reported crimes per 10,000 inhabitants 220 0.738 0.262 0.234 0.561 0.701 0.864 2.344
13. Robbery Reported crimes per 10,000 inhabitants 220 3.863 3.216 0.551 1.958 3.182 4.309 26.107
14. Robbery homicides Reported crimes per 10,000 inhabitants 220 0.005 0.012 0.000 0.000 0.000 0.000 0.064
Demography
15. Fertility rate Number of children per woman 220 1.330 0.127 0.930 1.240 1.325 1.420 1.740
16. Total growth rate of population Rate per thousand inhabitants 220 -1.146 4.582 -12.900 -4.500 -1.500 1.750 15.700
17. Population between 15 and 64 years Percentage at January 1 220 64.344 1.747 59.800 63.200 64.500 65.550 68.700
18. Population over 64 years Percentage at January 1 220 22.425 2.655 15.300 20.650 22.350 24.250 28.900
19. Population density Number of inhabitants per square kilometer 220 260 371 30.8 104 173 274 2,635
20. Total immigration Rate per thousand inhabitants 220 2.013 3.872 -8.300 -0.700 2.200 4.350 15.800
Economics
21. Isolation (highways, airports, and ports) Travel times to urban and logistic nodes 220 52.245 15.735 25.481 41.419 50.220 58.552 120.355
22. Participation to labor market
Labor force aged 15-64 years out of the total
population aged 15-64 (percentage)
220 64.538 8.067 45.409 57.134 67.897 70.769 75.747
25
Variable Description No. Mean
Standard.
Deviation
Min.
1st
Quartile
Median
3rd
Quartile
Max
23.
Participation to labor market: difference between
men and women Percentage 220 19.331 6.590 8.288 14.449 17.749 23.981 37.005
24. Exports per capita Euro per inhabitant 220 6,569 5,435 2,139 1,441 5,9918 10,629 25,905
25. Income inequality
Gini concentration index on equivalent net
household income
220 0.312 0.030 0.251 0.290 0.300 0.336 0.389
26. Non-performing entry rate of loans to households Percentage of loans to households 220 1.320 0.391 0.400 1.000 1.300 1.600 2.300
27. Unemployment: job seekers aged 15 and over Percentage of population between 15 and 64 years 220 7.292 2.726 2.353 5.171 6.756 8.984 16.953
28. Value added: manufacturing Percentage of the total value added 220 15.399 8.546 2.925 7.828 14.067 21.406 37.663
29. Value added: public sector Percentage of the total value added 220 19.563 6.363 9.453 14.147 18.219 24.930 32.968
30. Value added: per capita Euro per inhabitant 220 23,079 6,206 13,251 18,137 22,635 26,967 48,751
31.
Median gross hourly wage of employees born
abroad
Euro 220 9.859 0.536 8.330 9.550 9.915 10.185 11.810
32.
Median gross hourly wage of employees born in
Italy Euro 220 11.180 0.843 9.520 10.480 11.165 11.790 13.550
33. Mean wage of employees Euro 220 18,652 3,543 11,720 15,312 18,685 21,649 29,714
34. Mean wealth per capita Euro 220 148,340 44,806 69,310 108,594 154,120 181,852 295,154
Education, innovation, migration
35. Population having at least a secondary degree
Percentage of the population between 25 and 64
years 220 57.772 7.849 39.300 51.900 59.050 63.750 75.700
36. Immigration of graduates between 25 and 39 years Rate per 1,000 resident graduates 220 -10.160 16.290 -58.700 -20.400 -7.550 0.300 36.600
37. Foreign residents
Rate per 10,000 inhabitants between 15 and 64
years
220 10.996 5.514 1.274 5.761 11.692 15.524 26.135
38. Emigration to other Italian regions
Number of residents emigrated to other Italian
regions per 10,000 inhabitants
220 62.634 21.213 25.675 47.902 58.412 75.677 148.959
39. Emigration abroad
Number of residents emigrated abroad per 10,000
inhabitants 220 22.397 7.991 6.242 16.888 22.036 27.776 50.510
40. Beds in emergency residences for migrants
Rate per 10,000 inhabitants between 15 and 64
years
220 8.829 13.907 0.000 2.086 5.080 10.151 134.386
41. Newspaper circulation
Average number of newspapers distributed per day
per 10,000 inhabitants above 14 years
220 560 326 62 313 510 744 1.932
Notes. The Ministry of Interior is the source for the variable 40, and ADS is the source for the variable 41. For all other variables, the source is Istat. All the variables are at the
provincial level except variable 25, which is at the regional level.
26
Figure A3.1 - Correlation Heatmap
27
Table 3.2 – Correlation matrix – Part I
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
1.
Arsons
1.000
0.247*
-0.085
-0.302*
-0.005
0.218*
-0.019
0.085
0.199*
-0.277*
-0.013
-0.223*
-0.122*
-0.049
-0.263*
-0.243*
0.274*
-0.172*
-0.197*
-0.287*
0.214*
(0.000)
(0.207)
(0.000)
(0.943)
(0.001)
(0.777)
(0.209)
(0.003)
(0.000)
(0.844)
(0.001)
(0.071)
(0.469)
(0.000)
(0.000)
(0.000)
(0.011)
(0.003)
(0.000)
(0.001)
2.
Attempted homicides
0.247*
1.000
0.120*
-0.261*
0.180*
0.343*
0.219*
0.640*
0.501*
-0.131*
-0.050
-0.115*
0.132*
0.178*
-0.162*
-0.166*
0.361*
-0.298*
-0.043
-0.302*
0.273*
(0.000)
(0.076)
(0.000)
(0.007)
(0.000)
(0.001)
(0.000)
(0.000)
(0.052)
(0.461)
(0.090)
(0.051)
(0.008)
(0.016)
(0.014)
(0.000)
(0.000)
(0.525)
(0.000)
(0.000)
3.
Bag theft
-0.085
0.120*
1.000
0.242*
0.280*
0.273*
0.535*
0.090
0.194*
0.638*
0.196*
0.141*
0.870*
0.020
0.242*
0.232*
0.120*
-0.248*
0.467*
0.095
-0.298*
(0.207)
(0.076)
(0.000)
(0.000)
(0.000)
(0.000)
(0.183)
(0.004)
(0.000)
(0.003)
(0.036)
(0.000)
(0.770)
(0.000)
(0.001)
(0.075)
(0.000)
(0.000)
(0.159)
(0.000)
4.
Home burglaries
-0.302*
-0.261*
0.242*
1.000
-0.001
-0.248*
0.333*
-0.257*
-0.247*
0.693*
0.319*
0.248*
0.204*
-0.108
0.404*
0.469*
-0.412*
0.302*
0.114*
0.589*
-0.404*
(0.000)
(0.000)
(0.000)
(0.984)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.002)
(0.111)
(0.000)
(0.000)
(0.000)
(0.000)
(0.093)
(0.000)
(0.000)
5.
Drug-related crimes
-0.005
0.180*
0.280*
-0.001
1.000
0.270*
0.060
-0.009
-0.034
0.269*
0.120*
0.317*
0.229*
0.088
-0.194*
-0.000
-0.137*
0.199*
0.093
0.093
0.010
(0.943)
(0.007)
(0.000)
(0.984)
(0.000)
(0.375)
(0.896)
(0.617)
(0.000)
(0.076)
(0.000)
(0.001)
(0.191)
(0.004)
(0.996)
(0.043)
(0.003)
(0.168)
(0.169)
(0.887)
6.
Extortions
0.218*
0.343*
0.273*
-0.248*
0.270*
1.000
0.137*
0.193*
0.250*
-0.000
-0.034
-0.032
0.276*
-0.044
-0.201*
-0.311*
0.153*
-0.157*
0.085
-0.371*
-0.014
(0.001)
(0.000)
(0.000)
(0.000)
(0.000)
(0.043)
(0.004)
(0.000)
(0.998)
(0.618)
(0.639)
(0.000)
(0.516)
(0.003)
(0.000)
(0.024)
(0.020)
(0.208)
(0.000)
(0.837)
7.
House robberies
-0.019
0.219*
0.535*
0.333*
0.060
0.137*
1.000
0.200*
0.127*
0.420*
0.222*
0.067
0.555*
0.094
0.129*
0.212*
0.203*
-0.215*
0.107
0.130*
-0.202*
(0.777)
(0.001)
(0.000)
(0.000)
(0.375)
(0.043)
(0.003)
(0.060)
(0.000)
(0.001)
(0.320)
(0.000)
(0.163)
(0.056)
(0.002)
(0.003)
(0.001)
(0.113)
(0.054)
(0.003)
8.
Intentional homicides
0.085
0.640*
0.090
-0.257*
-0.009
0.193*
0.200*
1.000
0.655*
-0.147*
-0.129*
-0.201*
0.140*
0.266*
-0.153*
-0.187*
0.316*
-0.275*
-0.032
-0.322*
0.173*
(0.209)
(0.000)
(0.183)
(0.000)
(0.896)
(0.004)
(0.003)
(0.000)
(0.030)
(0.056)
(0.003)
(0.038)
(0.000)
(0.023)
(0.006)
(0.000)
(0.000)
(0.636)
(0.000)
(0.010)
9.
Mafia homicides
0.199*
0.501*
0.194*
-0.247*
-0.034
0.250*
0.127*
0.655*
1.000
-0.099
-0.052
-0.212*
0.241*
-0.019
0.033
-0.156*
0.237*
-0.306*
0.146*
-0.315*
0.047
(0.003)
(0.000)
(0.004)
(0.000)
(0.617)
(0.000)
(0.060)
(0.000)
(0.143)
(0.441)
(0.002)
(0.000)
(0.777)
(0.629)
(0.021)
(0.000)
(0.000)
(0.030)
(0.000)
(0.490)
10.
Micro criminality
-0.277*
-0.131*
0.638*
0.693*
0.269*
-0.000
0.420*
-0.147*
-0.099
1.000
0.352*
0.397*
0.618*
-0.070
0.426*
0.555*
-0.205*
0.068
0.373*
0.551*
-0.445*
(0.000)
(0.052)
(0.000)
(0.000)
(0.000)
(0.998)
(0.000)
(0.030)
(0.143)
(0.000)
(0.000)
(0.000)
(0.301)
(0.000)
(0.000)
(0.002)
(0.318)
(0.000)
(0.000)
(0.000)
11.
Prostitution-related
crimes
-0.013
-0.050
0.196*
0.319*
0.120*
-0.034
0.222*
-0.129*
-0.052
0.352*
1.000
0.241*
0.182*
0.045
0.235*
0.295*
-0.162*
0.117*
0.065
0.341*
-0.286*
(0.844)
(0.461)
(0.003)
(0.000)
(0.076)
(0.618)
(0.001)
(0.056)
(0.441)
(0.000)
(0.000)
(0.007)
(0.504)
(0.000)
(0.000)
(0.016)
(0.084)
(0.338)
(0.000)
(0.000)
12.
Sexual violence
-0.223*
-0.115*
0.141*
0.248*
0.317*
-0.032
0.067
-0.201*
-0.212*
0.397*
0.241*
1.000
0.149*
0.088
0.168*
0.339*
-0.271*
0.262*
0.191*
0.457*
-0.117*
(0.001)
(0.090)
(0.036)
(0.000)
(0.000)
(0.639)
(0.320)
(0.003)
(0.002)
(0.000)
(0.000)
(0.027)
(0.196)
(0.012)
(0.000)
(0.000)
(0.000)
(0.004)
(0.000)
(0.084)
13.
Robbery
-0.122*
0.132*
0.870*
0.204*
0.229*
0.276*
0.555*
0.140*
0.241*
0.618*
0.182*
0.149*
1.000
0.023
0.288*
0.324*
0.221*
-0.346*
0.623*
0.148*
-0.310*
(0.071)
(0.051)
(0.000)
(0.002)
(0.001)
(0.000)
(0.000)
(0.038)
(0.000)
(0.000)
(0.007)
(0.027)
(0.733)
(0.000)
(0.000)
(0.001)
(0.000)
(0.000)
(0.028)
(0.000)
14.
Robbery homicides
-0.049
0.178*
0.020
-0.108
0.088
-0.044
0.094
0.266*
-0.019
-0.070
0.045
0.088
0.023
1.000
-0.163*
0.057
0.153*
-0.061
0.015
0.050
0.156*
(0.469)
(0.008)
(0.770)
(0.111)
(0.191)
(0.516)
(0.163)
(0.000)
(0.777)
(0.301)
(0.504)
(0.196)
(0.733)
(0.015)
(0.398)
(0.023)
(0.370)
(0.826)
(0.465)
(0.021)
15.
Fertility rate
-0.263*
-0.162*
0.242*
0.404*
-0.194*
-0.201*
0.129*
-0.153*
0.033
0.426*
0.235*
0.168*
0.288*
-0.163*
1.000
0.640*
-0.042
-0.277*
0.262*
0.435*
-0.234*
(0.000)
(0.016)
(0.000)
(0.000)
(0.004)
(0.003)
(0.056)
(0.023)
(0.629)
(0.000)
(0.000)
(0.012)
(0.000)
(0.015)
(0.000)
(0.537)
(0.000)
(0.000)
(0.000)
(0.000)
16.
Total growth rate of
population
-0.243*
-0.166*
0.232*
0.469*
-0.000
-0.311*
0.212*
-0.187*
-0.156*
0.555*
0.295*
0.339*
0.324*
0.057
0.640*
1.000
0.108
-0.232*
0.290*
0.857*
-0.160*
(0.000)
(0.014)
(0.001)
(0.000)
(0.996)
(0.000)
(0.002)
(0.006)
(0.021)
(0.000)
(0.000)
(0.000)
(0.000)
(0.398)
(0.000)
(0.111)
(0.001)
(0.000)
(0.000)
(0.018)
17.
Population between 15
and 64 years
0.274*
0.361*
0.120*
-0.412*
-0.137*
0.153*
0.203*
0.316*
0.237*
-0.205*
-0.162*
-0.271*
0.221*
0.153*
-0.042
0.108
1.000
-0.911*
0.045
-0.319*
0.252*
(0.000)
(0.000)
(0.075)
(0.000)
(0.043)
(0.024)
(0.003)
(0.000)
(0.000)
(0.002)
(0.016)
(0.000)
(0.001)
(0.023)
(0.537)
(0.111)
(0.000)
(0.509)
(0.000)
(0.000)
18.
Population over 64 years
-0.172*
-0.298*
-0.248*
0.302*
0.199*
-0.157*
-0.215*
-0.275*
-0.306*
0.068
0.117*
0.262*
-0.346*
-0.061
-0.277*
-0.232*
-0.911*
1.000
-0.167*
0.272*
-0.141*
(0.011)
(0.000)
(0.000)
(0.000)
(0.003)
(0.020)
(0.001)
(0.000)
(0.000)
(0.318)
(0.084)
(0.000)
(0.000)
(0.370)
(0.000)
(0.001)
(0.000)
(0.013)
(0.000)
(0.037)
19.
19. Population density
-0.197*
-0.043
0.467*
0.114*
0.093
0.085
0.107
-0.032
0.146*
0.373*
0.065
0.191*
0.623*
0.015
0.262*
0.290*
0.045
-0.167*
1.000
0.181*
-0.327*
(0.003)
(0.525)
(0.000)
(0.093)
(0.168)
(0.208)
(0.113)
(0.636)
(0.030)
(0.000)
(0.338)
(0.004)
(0.000)
(0.826)
(0.000)
(0.000)
(0.509)
(0.013)
(0.007)
(0.000)
20.
Total immigration
-0.287*
-0.302*
0.095
0.589*
0.093
-0.371*
0.130*
-0.322*
-0.315*
0.551*
0.341*
0.457*
0.148*
0.050
0.435*
0.857*
-0.319*
0.272*
0.181*
1.000
-0.228*
(0.000)
(0.000)
(0.159)
(0.000)
(0.169)
(0.000)
(0.054)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.028)
(0.465)
(0.000)
(0.000)
(0.000)
(0.000)
(0.007)
(0.001)
21.
Isolation (highways,
airports, and ports)
0.214*
0.273*
-0.298*
-0.404*
0.010
-0.014
-0.202*
0.173*
0.047
-0.445*
-0.286*
-0.117*
-0.310*
0.156*
-0.234*
-0.160*
0.252*
-0.141*
-0.327*
-0.228*
1.000
(0.001)
(0.000)
(0.000)
(0.000)
(0.887)
(0.837)
(0.003)
(0.010)
(0.490)
(0.000)
(0.000)
(0.084)
(0.000)
(0.021)
(0.000)
(0.018)
(0.000)
(0.037)
(0.000)
(0.001)
22.
Participation to labor
market
-0.467*
-0.545*
-0.145*
0.552*
0.053
-0.383*
-0.253*
-0.448*
-0.408*
0.368*
0.190*
0.368*
-0.165*
-0.112*
0.357*
0.446*
-0.649*
0.550*
0.058
0.681*
-0.232*
(0.000)
(0.000)
(0.032)
(0.000)
(0.435)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.005)
(0.000)
(0.014)
(0.098)
(0.000)
(0.000)
(0.000)
(0.000)
(0.391)
(0.000)
(0.001)
23.
Participation to labor
market: difference
between men and
women
0.407*
0.353*
0.131*
-0.463*
-0.156*
0.328*
0.161*
0.275*
0.294*
-0.331*
-0.206*
-0.386*
0.129*
-0.014
-0.188*
-0.352*
0.665*
-0.633*
-0.063
-0.645*
0.134*
(0.000)
(0.000)
(0.052)
(0.000)
(0.020)
(0.000)
(0.017)
(0.000)
(0.000)
(0.000)
(0.002)
(0.000)
(0.057)
(0.832)
(0.005)
(0.000)
(0.000)
(0.000)
(0.351)
(0.000)
(0.047)
24.
Exports per capita
-0.408*
-0.475*
-0.074
0.377*
-0.182*
-0.261*
-0.143*
-0.326*
-0.244*
0.244*
0.077
0.163*
-0.067
-0.112*
0.397*
0.323*
-0.366*
0.225*
0.132*
0.389*
-0.303*
(0.000)
(0.000)
(0.273)
(0.000)
(0.007)
(0.000)
(0.034)
(0.000)
(0.000)
(0.000)
(0.253)
(0.016)
(0.322)
(0.098)
(0.000)
(0.000)
(0.000)
(0.001)
(0.050)
(0.000)
(0.000)
25.
Income inequality:
0.430*
0.431*
0.170*
-0.382*
0.078
0.377*
0.137*
0.257*
0.240*
-0.289*
-0.198*
-0.235*
0.159*
0.054
-0.281*
-0.348*
0.455*
-0.389*
0.055
-0.487*
0.190*
(0.000)
(0.000)
(0.012)
(0.000)
(0.249)
(0.000)
(0.043)
(0.000)
(0.000)
(0.000)
(0.003)
(0.000)
(0.018)
(0.427)
(0.000)
(0.000)
(0.000)
(0.000)
(0.418)
(0.000)
(0.005)
26.
Non-performing entry
rate of loans to
households
0.199*
0.182*
0.172*
-0.105
-0.095
0.253*
0.197*
0.035
0.196*
-0.048
0.044
-0.321*
0.143*
-0.061
0.007
-0.171*
0.283*
-0.332*
0.018
-0.304*
-0.097
(0.003)
(0.007)
(0.011)
(0.121)
(0.162)
(0.000)
(0.003)
(0.601)
(0.004)
(0.474)
(0.519)
(0.000)
(0.034)
(0.366)
(0.917)
(0.011)
(0.000)
(0.000)
(0.791)
(0.000)
(0.151)
28
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
27.
Unemployment: job
seekers aged 15 and over
0.364*
0.475*
0.076
-0.429*
0.141*
0.412*
0.069
0.311*
0.274*
-0.334*
-0.192*
-0.235*
0.051
0.113*
-0.502*
-0.478*
0.410*
-0.252*
-0.076
-0.570*
0.285*
(0.000)
(0.000)
(0.264)
(0.000)
(0.036)
(0.000)
(0.310)
(0.000)
(0.000)
(0.000)
(0.004)
(0.000)
(0.450)
(0.093)
(0.000)
(0.000)
(0.000)
(0.000)
(0.261)
(0.000)
(0.000)
28.
Value added:
manufacturing
-0.410*
-0.542*
-0.138*
0.417*
-0.317*
-0.294*
-0.189*
-0.346*
-0.241*
0.173*
0.074
-0.043
-0.119*
-0.134*
0.403*
0.305*
-0.359*
0.196*
0.094
0.357*
-0.349*
(0.000)
(0.000)
(0.041)
(0.000)
(0.000)
(0.000)
(0.005)
(0.000)
(0.000)
(0.010)
(0.276)
(0.524)
(0.077)
(0.047)
(0.000)
(0.000)
(0.000)
(0.004)
(0.166)
(0.000)
(0.000)
29.
Value added: public
sector
0.434*
0.529*
-0.053
-0.606*
0.085
0.296*
0.138*
0.421*
0.258*
-0.485*
-0.220*
-0.183*
-0.063
0.154*
-0.499*
-0.476*
0.500*
-0.311*
-0.230*
-0.575*
0.363*
(0.000)
(0.000)
(0.435)
(0.000)
(0.211)
(0.000)
(0.040)
(0.000)
(0.000)
(0.000)
(0.001)
(0.006)
(0.350)
(0.022)
(0.000)
(0.000)
(0.000)
(0.000)
(0.001)
(0.000)
(0.000)
30.
Value added: per capita
-0.488*
-0.395*
0.066
0.448*
0.150*
-0.264*
-0.143*
-0.330*
-0.294*
0.501*
0.117*
0.493*
0.083
-0.043
0.470*
0.572*
-0.439*
0.292*
0.260*
0.660*
-0.206*
(0.000)
(0.000)
(0.327)
(0.000)
(0.026)
(0.000)
(0.035)
(0.000)
(0.000)
(0.000)
(0.083)
(0.000)
(0.221)
(0.529)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.002)
31.
Median gross hourly
wage of employees born
abroad
-0.520*
-0.439*
-0.246*
0.197*
-0.023
-0.283*
-0.320*
-0.358*
-0.331*
0.054
-0.059
0.286*
-0.258*
-0.094
0.236*
0.128*
-0.494*
0.407*
-0.052
0.287*
-0.015
(0.000)
(0.000)
(0.000)
(0.003)
(0.738)
(0.000)
(0.000)
(0.000)
(0.000)
(0.422)
(0.384)
(0.000)
(0.000)
(0.167)
(0.000)
(0.059)
(0.000)
(0.000)
(0.445)
(0.000)
(0.826)
32.
Median gross hourly
wage of employees born
in Italy
-0.517*
-0.476*
0.002
0.496*
-0.041
-0.334*
-0.164*
-0.364*
-0.283*
0.431*
0.052
0.400*
0.038
-0.114*
0.515*
0.473*
-0.526*
0.357*
0.298*
0.595*
-0.274*
(0.000)
(0.000)
(0.979)
(0.000)
(0.546)
(0.000)
(0.015)
(0.000)
(0.000)
(0.000)
(0.447)
(0.000)
(0.574)
(0.091)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
33.
Mean wage of
employees
-0.514*
-0.515*
0.028
0.500*
-0.069
-0.312*
-0.149*
-0.395*
-0.288*
0.442*
0.092
0.374*
0.079
-0.127*
0.499*
0.469*
-0.507*
0.332*
0.324*
0.585*
-0.360*
(0.000)
(0.000)
(0.682)
(0.000)
(0.311)
(0.000)
(0.027)
(0.000)
(0.000)
(0.000)
(0.174)
(0.000)
(0.244)
(0.061)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
34.
Mean wealth per capita
-0.470*
-0.481*
0.021
0.564*
0.007
-0.309*
-0.102
-0.397*
-0.318*
0.475*
0.207*
0.450*
0.035
-0.131*
0.486*
0.454*
-0.622*
0.468*
0.206*
0.645*
-0.297*
(0.000)
(0.000)
(0.757)
(0.000)
(0.922)
(0.000)
(0.130)
(0.000)
(0.000)
(0.000)
(0.002)
(0.000)
(0.608)
(0.051)
(0.000)
(0.000)
(0.000)
(0.000)
(0.002)
(0.000)
(0.000)
35.
Population having at
least a secondary degree
-0.171*
-0.467*
-0.074
0.272*
0.085
-0.168*
-0.278*
-0.414*
-0.275*
0.271*
0.116*
0.339*
-0.131*
-0.159*
0.182*
0.233*
-0.561*
0.489*
0.110
0.466*
-0.323*
(0.011)
(0.000)
(0.274)
(0.000)
(0.209)
(0.013)
(0.000)
(0.000)
(0.000)
(0.000)
(0.085)
(0.000)
(0.052)
(0.018)
(0.007)
(0.000)
(0.000)
(0.000)
(0.104)
(0.000)
(0.000)
36.
Immigration of
graduates between 25
and 39 years
-0.455*
-0.453*
0.161*
0.605*
0.115*
-0.281*
0.014
-0.403*
-0.366*
0.608*
0.257*
0.455*
0.156*
-0.060
0.414*
0.619*
-0.443*
0.343*
0.266*
0.774*
-0.360*
(0.000)
(0.000)
(0.017)
(0.000)
(0.089)
(0.000)
(0.841)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.021)
(0.372)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
37.
Foreign residents
-0.384*
-0.437*
0.015
0.555*
0.063
-0.227*
-0.140*
-0.375*
-0.247*
0.453*
0.204*
0.278*
0.004
-0.141*
0.486*
0.428*
-0.595*
0.394*
0.130*
0.590*
-0.380*
(0.000)
(0.000)
(0.831)
(0.000)
(0.354)
(0.001)
(0.038)
(0.000)
(0.000)
(0.000)
(0.002)
(0.000)
(0.957)
(0.037)
(0.000)
(0.000)
(0.000)
(0.000)
(0.054)
(0.000)
(0.000)
38.
Emigration to other
Italian regions
0.512*
0.360*
0.006
-0.174*
-0.062
0.163*
0.140*
0.253*
0.353*
-0.128*
0.068
-0.169*
0.019
-0.001
-0.116*
-0.198*
0.107
-0.052
-0.158*
-0.178*
0.039
(0.000)
(0.000)
(0.931)
(0.010)
(0.361)
(0.015)
(0.038)
(0.000)
(0.000)
(0.058)
(0.318)
(0.012)
(0.781)
(0.987)
(0.085)
(0.003)
(0.114)
(0.440)
(0.019)
(0.008)
(0.566)
39.
Emigration abroad
-0.304*
-0.315*
-0.266*
0.005
0.053
-0.101
-0.456*
-0.280*
-0.176*
-0.081
-0.100
0.247*
-0.324*
-0.155*
0.141*
-0.099
-0.529*
0.415*
-0.006
0.057
-0.025
(0.000)
(0.000)
(0.000)
(0.945)
(0.434)
(0.136)
(0.000)
(0.000)
(0.009)
(0.233)
(0.139)
(0.000)
(0.000)
(0.021)
(0.037)
(0.145)
(0.000)
(0.000)
(0.931)
(0.403)
(0.717)
40.
Beds in emergency
residences for migrants
-0.005
-0.024
0.059
-0.057
0.050
0.033
-0.008
-0.069
-0.011
0.042
-0.031
0.199*
0.023
-0.102
0.164*
-0.007
-0.053
-0.016
0.055
-0.039
-0.065
(0.943)
(0.718)
(0.381)
(0.401)
(0.460)
(0.621)
(0.911)
(0.310)
(0.866)
(0.537)
(0.645)
(0.003)
(0.736)
(0.131)
(0.015)
(0.914)
(0.438)
(0.816)
(0.418)
(0.563)
(0.339)
41.
Newspaper circulation
-0.353*
-0.107
-0.057
0.307*
0.034
-0.453*
0.101
-0.019
-0.175*
0.306*
0.208*
0.376*
0.002
0.137*
0.268*
0.445*
-0.205*
0.207*
-0.048
0.528*
0.035
(0.000)
(0.112)
(0.404)
(0.000)
(0.619)
(0.000)
(0.137)
(0.780)
(0.009)
(0.000)
(0.002)
(0.000)
(0.979)
(0.043)
(0.000)
(0.000)
(0.002)
(0.002)
(0.479)
(0.000)
(0.603)
29
Table 3.3 – Correlation matrix – Part II
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
1.
Arsons
-0.467*
0.407*
-0.408*
0.430*
0.199*
0.364*
-0.410*
0.434*
-0.488*
-0.520*
-0.517*
-0.514*
-0.470*
-0.171*
-0.455*
-0.384*
0.512*
-0.304*
-0.005
-0.353*
(0.000)
(0.000)
(0.000)
(0.000)
(0.003)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.011)
(0.000)
(0.000)
(0.000)
(0.000)
(0.943)
(0.000)
2.
Attempted homicides
-0.545*
0.353*
-0.475*
0.431*
0.182*
0.475*
-0.542*
0.529*
-0.395*
-0.439*
-0.476*
-0.515*
-0.481*
-0.467*
-0.453*
-0.437*
0.360*
-0.315*
-0.024
-0.107
(0.000)
(0.000)
(0.000)
(0.000)
(0.007)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.718)
(0.112)
3.
Bag theft
-0.145*
0.131*
-0.074
0.170*
0.172*
0.076
-0.138*
-0.053
0.066
-0.246*
0.002
0.028
0.021
-0.074
0.161*
0.015
0.006
-0.266*
0.059
-0.057
(0.032)
(0.052)
(0.273)
(0.012)
(0.011)
(0.264)
(0.041)
(0.435)
(0.327)
(0.000)
(0.979)
(0.682)
(0.757)
(0.274)
(0.017)
(0.831)
(0.931)
(0.000)
(0.381)
(0.404)
4.
Home burglaries
0.552*
-0.463*
0.377*
-0.382*
-0.105
-0.429*
0.417*
-0.606*
0.448*
0.197*
0.496*
0.500*
0.564*
0.272*
0.605*
0.555*
-0.174*
0.005
-0.057
0.307*
(0.000)
(0.000)
(0.000)
(0.000)
(0.121)
(0.000)
(0.000)
(0.000)
(0.000)
(0.003)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.010)
(0.945)
(0.401)
(0.000)
5.
Drug-related crimes
0.053
-0.156*
-0.182*
0.078
-0.095
0.141*
-0.317*
0.085
0.150*
-0.023
-0.041
-0.069
0.007
0.085
0.115*
0.063
-0.062
0.053
0.050
0.034
(0.435)
(0.020)
(0.007)
(0.249)
(0.162)
(0.036)
(0.000)
(0.211)
(0.026)
(0.738)
(0.546)
(0.311)
(0.922)
(0.209)
(0.089)
(0.354)
(0.361)
(0.434)
(0.460)
(0.619)
6.
Extortions
-0.383*
0.328*
-0.261*
0.377*
0.253*
0.412*
-0.294*
0.296*
-0.264*
-0.283*
-0.334*
-0.312*
-0.309*
-0.168*
-0.281*
-0.227*
0.163*
-0.101
0.033
-0.453*
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.013)
(0.000)
(0.001)
(0.015)
(0.136)
(0.621)
(0.000)
7.
House robberies
-0.253*
0.161*
-0.143*
0.137*
0.197*
0.069
-0.189*
0.138*
-0.143*
-0.320*
-0.164*
-0.149*
-0.102
-0.278*
0.014
-0.140*
0.140*
-0.456*
-0.008
0.101
(0.000)
(0.017)
(0.034)
(0.043)
(0.003)
(0.310)
(0.005)
(0.040)
(0.035)
(0.000)
(0.015)
(0.027)
(0.130)
(0.000)
(0.841)
(0.038)
(0.038)
(0.000)
(0.911)
(0.137)
8.
Intentional homicides
-0.448*
0.275*
-0.326*
0.257*
0.035
0.311*
-0.346*
0.421*
-0.330*
-0.358*
-0.364*
-0.395*
-0.397*
-0.414*
-0.403*
-0.375*
0.253*
-0.280*
-0.069
-0.019
(0.000)
(0.000)
(0.000)
(0.000)
(0.601)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.310)
(0.780)
9.
Mafia homicides
-0.408*
0.294*
-0.244*
0.240*
0.196*
0.274*
-0.241*
0.258*
-0.294*
-0.331*
-0.283*
-0.288*
-0.318*
-0.275*
-0.366*
-0.247*
0.353*
-0.176*
-0.011
-0.175*
(0.000)
(0.000)
(0.000)
(0.000)
(0.004)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.009)
(0.866)
(0.009)
10.
Micro criminality
0.368*
-0.331*
0.244*
-0.289*
-0.048
-0.334*
0.173*
-0.485*
0.501*
0.054
0.431*
0.442*
0.475*
0.271*
0.608*
0.453*
-0.128*
-0.081
0.042
0.306*
(0.000)
(0.000)
(0.000)
(0.000)
(0.474)
(0.000)
(0.010)
(0.000)
(0.000)
(0.422)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.058)
(0.233)
(0.537)
(0.000)
11.
Prostitution-related
crimes
0.190*
-0.206*
0.077
-0.198*
0.044
-0.192*
0.074
-0.220*
0.117*
-0.059
0.052
0.092
0.207*
0.116*
0.257*
0.204*
0.068
-0.100
-0.031
0.208*
(0.005)
(0.002)
(0.253)
(0.003)
(0.519)
(0.004)
(0.276)
(0.001)
(0.083)
(0.384)
(0.447)
(0.174)
(0.002)
(0.085)
(0.000)
(0.002)
(0.318)
(0.139)
(0.645)
(0.002)
12.
Sexual violence
0.368*
-0.386*
0.163*
-0.235*
-0.321*
-0.235*
-0.043
-0.183*
0.493*
0.286*
0.400*
0.374*
0.450*
0.339*
0.455*
0.278*
-0.169*
0.247*
0.199*
0.376*
(0.000)
(0.000)
(0.016)
(0.000)
(0.000)
(0.000)
(0.524)
(0.006)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.012)
(0.000)
(0.003)
(0.000)
13.
Robbery
-0.165*
0.129*
-0.067
0.159*
0.143*
0.051
-0.119*
-0.063
0.083
-0.258*
0.038
0.079
0.035
-0.131*
0.156*
0.004
0.019
-0.324*
0.023
0.002
(0.014)
(0.057)
(0.322)
(0.018)
(0.034)
(0.450)
(0.077)
(0.350)
(0.221)
(0.000)
(0.574)
(0.244)
(0.608)
(0.052)
(0.021)
(0.957)
(0.781)
(0.000)
(0.736)
(0.979)
14.
Robbery homicides
-0.112*
-0.014
-0.112*
0.054
-0.061
0.113*
-0.134*
0.154*
-0.043
-0.094
-0.114*
-0.127*
-0.131*
-0.159*
-0.060
-0.141*
-0.001
-0.155*
-0.102
0.137*
(0.098)
(0.832)
(0.098)
(0.427)
(0.366)
(0.093)
(0.047)
(0.022)
(0.529)
(0.167)
(0.091)
(0.061)
(0.051)
(0.018)
(0.372)
(0.037)
(0.987)
(0.021)
(0.131)
(0.043)
15.
Fertility rate
0.357*
-0.188*
0.397*
-0.281*
0.007
-0.502*
0.403*
-0.499*
0.470*
0.236*
0.515*
0.499*
0.486*
0.182*
0.414*
0.486*
-0.116*
0.141*
0.164*
0.268*
(0.000)
(0.005)
(0.000)
(0.000)
(0.917)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.007)
(0.000)
(0.000)
(0.085)
(0.037)
(0.015)
(0.000)
16.
Total growth rate of
population
0.446*
-0.352*
0.323*
-0.348*
-0.171*
-0.478*
0.305*
-0.476*
0.572*
0.128*
0.473*
0.469*
0.454*
0.233*
0.619*
0.428*
-0.198*
-0.099
-0.007
0.445*
(0.000)
(0.000)
(0.000)
(0.000)
(0.011)
(0.000)
(0.000)
(0.000)
(0.000)
(0.059)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.003)
(0.145)
(0.914)
(0.000)
17.
Population between 15
and 64 years
-0.649*
0.665*
-0.366*
0.455*
0.283*
0.410*
-0.359*
0.500*
-0.439*
-0.494*
-0.526*
-0.507*
-0.622*
-0.561*
-0.443*
-0.595*
0.107
-0.529*
-0.053
-0.205*
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.114)
(0.000)
(0.438)
(0.002)
18.
Population over 64 years
0.550*
-0.633*
0.225*
-0.389*
-0.332*
-0.252*
0.196*
-0.311*
0.292*
0.407*
0.357*
0.332*
0.468*
0.489*
0.343*
0.394*
-0.052
0.415*
-0.016
0.207*
(0.000)
(0.000)
(0.001)
(0.000)
(0.000)
(0.000)
(0.004)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.440)
(0.000)
(0.816)
(0.002)
19.
19. Population density
0.058
-0.063
0.132*
0.055
0.018
-0.076
0.094
-0.230*
0.260*
-0.052
0.298*
0.324*
0.206*
0.110
0.266*
0.130*
-0.158*
-0.006
0.055
-0.048
(0.391)
(0.351)
(0.050)
(0.418)
(0.791)
(0.261)
(0.166)
(0.001)
(0.000)
(0.445)
(0.000)
(0.000)
(0.002)
(0.104)
(0.000)
(0.054)
(0.019)
(0.931)
(0.418)
(0.479)
20.
Total immigration
0.681*
-0.645*
0.389*
-0.487*
-0.304*
-0.570*
0.357*
-0.575*
0.660*
0.287*
0.595*
0.585*
0.645*
0.466*
0.774*
0.590*
-0.178*
0.057
-0.039
0.528*
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.008)
(0.403)
(0.563)
(0.000)
21.
Isolation (highways,
airports, and ports)
-0.232*
0.134*
-0.303*
0.190*
-0.097
0.285*
-0.349*
0.363*
-0.206*
-0.015
-0.274*
-0.360*
-0.297*
-0.323*
-0.360*
-0.380*
0.039
-0.025
-0.065
0.035
(0.001)
(0.047)
(0.000)
(0.005)
(0.151)
(0.000)
(0.000)
(0.000)
(0.002)
(0.826)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.566)
(0.717)
(0.339)
(0.603)
22.
Participation to labor
market
1.000
-0.823*
0.630*
-0.707*
-0.405*
-0.666*
0.649*
-0.800*
0.808*
0.655*
0.804*
0.787*
0.844*
0.659*
0.776*
0.810*
-0.346*
0.505*
-0.015
0.450*
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.830)
(0.000)
23.
Participation to labor
market: difference
between men and
women
-0.823*
1.000
-0.464*
0.582*
0.440*
0.581*
-0.433*
0.590*
-0.714*
-0.524*
-0.676*
-0.638*
-0.729*
-0.594*
-0.711*
-0.653*
0.206*
-0.410*
0.030
-0.478*
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.002)
(0.000)
(0.655)
(0.000)
24.
Exports per capita
0.630*
-0.464*
1.000
-0.493*
-0.148*
-0.551*
0.809*
-0.733*
0.596*
0.505*
0.711*
0.739*
0.590*
0.445*
0.544*
0.616*
-0.347*
0.363*
0.041
0.213*
(0.000)
(0.000)
(0.000)
(0.028)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.543)
(0.002)
25.
Income inequality:
-0.707*
0.582*
-0.493*
1.000
0.358*
0.607*
-0.559*
0.589*
-0.570*
-0.492*
-0.560*
-0.555*
-0.590*
-0.382*
-0.535*
-0.524*
0.199*
-0.384*
0.072
-0.469*
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.003)
(0.000)
(0.288)
(0.000)
26.
Non-performing entry
rate of loans to
households
-0.405*
0.440*
-0.148*
0.358*
1.000
0.379*
-0.064
0.174*
-0.475*
-0.416*
-0.412*
-0.381*
-0.465*
-0.253*
-0.324*
-0.138*
0.235*
-0.231*
0.038
-0.513*
(0.000)
(0.000)
(0.028)
(0.000)
(0.000)
(0.345)
(0.010)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.041)
(0.000)
(0.001)
(0.572)
(0.000)
27.
Unemployment: job
seekers aged 15 and over
-0.666*
0.581*
-0.551*
0.607*
0.379*
1.000
-0.620*
0.666*
-0.654*
-0.484*
-0.689*
-0.694*
-0.728*
-0.512*
-0.641*
-0.633*
0.293*
-0.308*
0.071
-0.412*
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.293)
(0.000)
30
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
28.
Value added:
manufacturing
0.649*
-0.433*
0.809*
-0.559*
-0.064
-0.620*
1.000
-0.816*
0.500*
0.456*
0.651*
0.698*
0.538*
0.371*
0.479*
0.633*
-0.373*
0.346*
-0.060
0.092
(0.000)
(0.000)
(0.000)
(0.000)
(0.345)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.379)
(0.173)
29.
Value added: public
sector
-0.800*
0.590*
-0.733*
0.589*
0.174*
0.666*
-0.816*
1.000
-0.695*
-0.484*
-0.764*
-0.787*
-0.766*
-0.464*
-0.684*
-0.808*
0.407*
-0.393*
0.058
-0.229*
(0.000)
(0.000)
(0.000)
(0.000)
(0.010)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.389)
(0.001)
30.
Value added: per capita
0.808*
-0.714*
0.596*
-0.570*
-0.475*
-0.654*
0.500*
-0.695*
1.000
0.622*
0.861*
0.841*
0.842*
0.594*
0.777*
0.708*
-0.363*
0.454*
0.062
0.508*
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.363)
(0.000)
31.
Median gross hourly
wage of employees born
abroad
0.655*
-0.524*
0.505*
-0.492*
-0.416*
-0.484*
0.456*
-0.484*
0.622*
1.000
0.712*
0.647*
0.630*
0.497*
0.472*
0.429*
-0.467*
0.629*
0.093
0.381*
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.169)
(0.000)
32.
Median gross hourly
wage of employees born
in Italy
0.804*
-0.676*
0.711*
-0.560*
-0.412*
-0.689*
0.651*
-0.764*
0.861*
0.712*
1.000
0.953*
0.863*
0.596*
0.745*
0.729*
-0.393*
0.498*
0.096
0.412*
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.155)
(0.000)
33.
Mean wage of
employees
0.787*
-0.638*
0.739*
-0.555*
-0.381*
-0.694*
0.698*
-0.787*
0.841*
0.647*
0.953*
1.000
0.845*
0.596*
0.734*
0.723*
-0.403*
0.439*
0.053
0.344*
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.432)
(0.000)
34.
Mean wealth per capita
0.844*
-0.729*
0.590*
-0.590*
-0.465*
-0.728*
0.538*
-0.766*
0.842*
0.630*
0.863*
0.845*
1.000
0.570*
0.764*
0.730*
-0.304*
0.478*
0.005
0.489*
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.947)
(0.000)
35.
Population having at
least a secondary degree
0.659*
-0.594*
0.445*
-0.382*
-0.253*
-0.512*
0.371*
-0.464*
0.594*
0.497*
0.596*
0.596*
0.570*
1.000
0.631*
0.577*
-0.108
0.386*
0.025
0.171*
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.112)
(0.000)
(0.712)
(0.011)
36.
Immigration of
graduates between 25
and 39 years
0.776*
-0.711*
0.544*
-0.535*
-0.324*
-0.641*
0.479*
-0.684*
0.777*
0.472*
0.745*
0.734*
0.764*
0.631*
1.000
0.705*
-0.337*
0.271*
-0.009
0.448*
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.899)
(0.000)
37.
Foreign residents
0.810*
-0.653*
0.616*
-0.524*
-0.138*
-0.633*
0.633*
-0.808*
0.708*
0.429*
0.729*
0.723*
0.730*
0.577*
0.705*
1.000
-0.215*
0.490*
0.089
0.235*
(0.000)
(0.000)
(0.000)
(0.000)
(0.041)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.001)
(0.000)
(0.188)
(0.000)
38.
Emigration to other
Italian regions
-0.346*
0.206*
-0.347*
0.199*
0.235*
0.293*
-0.373*
0.407*
-0.363*
-0.467*
-0.393*
-0.403*
-0.304*
-0.108
-0.337*
-0.215*
1.000
-0.299*
-0.080
-0.131*
(0.000)
(0.002)
(0.000)
(0.003)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.112)
(0.000)
(0.001)
(0.000)
(0.237)
(0.052)
39.
Emigration abroad
0.505*
-0.410*
0.363*
-0.384*
-0.231*
-0.308*
0.346*
-0.393*
0.454*
0.629*
0.498*
0.439*
0.478*
0.386*
0.271*
0.490*
-0.299*
1.000
0.201*
0.109
(0.000)
(0.000)
(0.000)
(0.000)
(0.001)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.003)
(0.107)
40.
Beds in emergency
residences for migrants
-0.015
0.030
0.041
0.072
0.038
0.071
-0.060
0.058
0.062
0.093
0.096
0.053
0.005
0.025
-0.009
0.089
-0.080
0.201*
1.000
-0.025
(0.830)
(0.655)
(0.543)
(0.288)
(0.572)
(0.293)
(0.379)
(0.389)
(0.363)
(0.169)
(0.155)
(0.432)
(0.947)
(0.712)
(0.899)
(0.188)
(0.237)
(0.003)
(0.717)
41.
Newspaper circulation
0.450*
-0.478*
0.213*
-0.469*
-0.513*
-0.412*
0.092
-0.229*
0.508*
0.381*
0.412*
0.344*
0.489*
0.171*
0.448*
0.235*
-0.131*
0.109
-0.025
1.000
(0.000)
(0.000)
(0.002)
(0.000)
(0.000)
(0.000)
(0.173)
(0.001)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.011)
(0.000)
(0.000)
(0.052)
(0.107)
(0.717)
31
Table A3.4 – Italian provinces in 2018
Province
Acronym
Province
Acronym
Province
Acronym
Agrigento
AG
Genova
GE
Pordenone
PN
Alessandria
AL
Gorizia
GO
Potenza
PZ
Ancona
AN
Grosseto
GR
Prato
PO
Aosta
AO
Imperia
IM
Ragusa
RG
Arezzo
AR
Isernia
IS
Ravenna
RA
Ascoli Piceno
AP
La Spezia
SP
Reggio di Calabria
RC
Asti
AT
L'Aquila
AQ
Reggio nell'Emilia
RE
Avellino
AV
Latina
LT
Rieti
RI
Bari
BA
Lecce
LE
Rimini
RN
Barletta-Andria-Trani
BT
Lecco
LC
Roma
RM
Belluno
BL
Livorno
LI
Rovigo
RO
Benevento
BN
Lodi
LO
Salerno
SA
Bergamo
BG
Lucca
LU
Sassari
SS
Biella
BI
Macerata
MC
Savona
SV
Bologna
BO
Mantova
MN
Siena
SI
Bolzano/Bozen
BZ
Massa-Carrara
MS
Siracusa
SR
Brescia
BS
Matera
MT
Sondrio
SO
Brindisi
BR
Medio Campidano
VS
Sud Sardegna
SU
Cagliari
CA
Messina
ME
Taranto
TA
Caltanissetta
CL
Milano
MI
Teramo
TE
Campobasso
CB
Modena
MO
Terni
TR
Carbonia-Iglesias
CI
Monza e della Brianza
MB
Torino
TO
Caserta
CE
Napoli
NA
Trapani
TP
Catania
CT
Novara
NO
Trento
TN
Catanzaro
CZ
Nuoro
NU
Treviso
TV
Chieti
CH
Ogliastra
OG
Trieste
TS
Como
CO
Olbia-Tempio
OT
Udine
UD
Cosenza
CS
Oristano
OR
Varese
VA
Cremona
CR
Padova
PD
Venezia
VE
Crotone
KR
Palermo
PA
Verbano-Cusio-Ossola
VB
Cuneo
CN
Parma
PR
Vercelli
VC
Enna
EN
Pavia
PV
Verona
VR
Fermo
FM
Perugia
PG
Vibo Valentia
VV
Ferrara
FE
Pesaro e Urbino
PU
Vicenza
VI
Firenze
FI
Pescara
PE
Viterbo
VT
Foggia
FG
Piacenza
PC
Forlì-Cesena
FC
Pisa
PI
Frosinone
FR
Pistoia
PT
32
Appendix 4 Factor analysis
Table A4.1 – Factor analysis, loading factors greater than 0.3
Variable
Factor 1
Factor 2
Factor 3
Factor 4
Factor 5
Factor 6
Factor 7
Factor 8
Factor 9
Communality
Uniqueness
1. Arsons
-0.384
0.682
0.709
0.291
2. Attempted homicides
-0.380
0.328
0.593
0.680
0.320
3. Bag theft
0.917
0.868
0.132
4. Home burglaries
0.517
0.339
-0.314
0.428
0.767
0.233
5. Drug-related crimes
0.344
0.682
0.644
0.356
6. Extortions
0.311
0.319
0.379
0.329
0.564
0.436
7. House robberies
0.482
0.671
0.750
0.250
8. Intentional homicides
0.804
0.778
0.222
9. Mafia homicides
0.345
0.606
0.734
0.266
10. Micro criminality
0.403
0.796
0.904
0.096
11. Prostitution-related crimes
0.757
0.697
0.303
12. Sexual violence
0.369
0.474
0.340
0.471
0.773
0.227
13. Robbery
0.901
0.890
0.110
14. Robbery homicides
0.384
0.474
-0.323
0.398
0.692
0.308
15. Fertility rate
0.481
0.418
0.529
0.809
0.191
16. Total growth rate of population
0.781
0.456
0.893
0.107
17. Population between 15 and 64 years
-0.505
0.790
0.908
0.092
18. Population over 64 years
0.319
-0.875
0.933
0.067
19. Population density
0.655
-0.335
0.738
0.262
20. Total immigration
0.869
0.887
0.113
21. Isolation (highways, airports, and ports)
-0.491
0.421
-0.329
0.645
0.355
22. Participation to labor market
0.882
0.911
0.089
23. Participation to labor market: difference between men
and women
-0.804
0.325
0.812
0.188
24. Exports per capita
0.668
-0.482
0.751
0.249
25. Income inequality: Gini concentration index on
equivalent net household income
-0.656
0.391
0.697
0.303
26. Non-performing entry rate of loans to households
-0.648
0.646
0.354
27. Unemployment: job seekers aged 15 and over
-0.805
0.717
0.283
33
Variable
Factor 1
Factor 2
Factor 3
Factor 4
Factor 5
Factor 6
Factor 7
Factor 8
Factor 9
Communality
Uniqueness
28. Value added: manufacturing
0.601
-0.668
0.881
0.119
29. Value added: public sector
-0.761
0.426
0.878
0.122
30. Value added: per capita
0.907
0.895
0.105
31. Median gross hourly wage of employees born abroad
0.639
-0.364
0.352
0.747
0.253
32. Median gross hourly wage of employees born in Italy
0.869
0.893
0.107
33. Mean wage of employees
0.833
0.874
0.126
34. Mean wealth per capita
0.870
0.868
0.132
35. Population having at least a secondary degree
0.699
-0.323
0.762
0.238
36. Immigration of graduates between 25 and 39 years
0.848
0.844
0.156
37. Foreign residents
0.763
0.768
0.232
38. Emigration to other Italian regions
-0.335
0.765
0.745
0.255
39. Emigration abroad
0.413
0.579
0.656
0.344
40. Beds in emergency residences for migrants
0.671
0.557
0.443
41. Newspaper circulation
0.708
0.747
0.253
Eigenvalues
15.569
4.517
2.981
2.473
1.607
1.352
1.330
1.056
1.027
Difference
11.051
1.537
0.508
0.866
0.255
0.022
0.273
0.030
Proportion
0.380
0.110
0.073
0.060
0.039
0.033
0.032
0.026
0.025
Cumulative proportion
0.380
0.490
0.563
0.623
0.662
0.695
0.728
0.753
0.778
Explained variance
13.468
4.209
2.890
2.691
2.354
1.903
1.640
1.453
1.302
Number of variables
41.000
Number of retained factors
9.000
Notes. Factor loadings below 0.3 are omitted.
34
Table A4.2 – Factor analysis – Full table
Variable
Factor 1
Factor 2
Factor 3
Factor 4
Factor 5
Factor 6
Factor 7
Factor 8
Factor 9
Communality
Uniqueness
1. Arsons
-0.384
-0.183
0.117
0.117
-0.120
0.682
-0.142
0.024
0.027
0.709
0.291
2. Attempted homicides
-0.380
0.018
0.128
0.328
0.593
0.122
0.055
-0.154
0.137
0.680
0.320
3. Bag theft
0.090
0.917
0.048
0.083
0.014
-0.039
0.049
-0.030
0.073
0.868
0.132
4. Home burglaries
0.517
0.339
-0.314
-0.147
-0.245
-0.069
-0.030
-0.124
0.428
0.767
0.233
5. Drug-related crimes
-0.036
0.344
-0.217
0.682
0.017
-0.030
-0.090
0.057
0.007
0.644
0.356
6. Extortions
-0.268
0.311
0.123
0.106
0.319
0.379
0.027
0.124
0.329
0.564
0.436
7. House robberies
-0.063
0.482
-0.022
-0.045
0.010
-0.034
-0.011
0.243
0.671
0.750
0.250
8. Intentional homicides
-0.277
-0.054
0.059
0.094
0.804
-0.071
-0.160
-0.083
-0.050
0.778
0.222
9. Mafia homicides
-0.279
0.345
0.071
-0.262
0.606
0.286
0.041
-0.009
-0.114
0.734
0.266
10. Micro criminality
0.403
0.796
-0.114
0.082
-0.140
-0.011
0.013
-0.062
0.256
0.904
0.096
11. Prostitution-related crimes
0.133
0.161
-0.205
-0.035
-0.104
0.020
0.083
0.757
0.139
0.697
0.303
12. Sexual violence
0.369
0.191
-0.150
0.474
-0.078
0.001
0.340
0.471
-0.097
0.773
0.227
13. Robbery
0.052
0.901
0.204
0.045
0.126
-0.019
-0.018
0.112
0.054
0.890
0.110
14. Robbery homicides
0.063
-0.022
0.037
0.384
0.474
-0.323
-0.200
0.398
0.106
0.692
0.308
15. Fertility rate
0.481
0.190
0.418
-0.274
-0.082
0.040
0.529
-0.034
-0.042
0.809
0.191
16. Total growth rate of population
0.781
0.164
0.456
0.078
-0.055
-0.074
-0.040
0.144
0.105
0.893
0.107
17. Population between 15 and 64 years
-0.505
0.024
0.790
0.058
0.100
-0.011
-0.089
-0.077
-0.022
0.908
0.092
18. Population over 64 years
0.319
-0.168
-0.875
0.106
-0.062
0.013
-0.082
0.120
0.028
0.933
0.067
19. Population density
0.172
0.655
0.280
-0.049
0.030
-0.112
-0.085
0.255
-0.335
0.738
0.262
20. Total immigration
0.869
0.036
-0.073
0.132
-0.101
-0.012
-0.145
0.233
0.149
0.887
0.113
21. Isolation (highways, airports, and ports)
-0.154
-0.491
0.210
0.421
0.214
-0.018
0.015
-0.329
0.065
0.645
0.355
22. Participation to labor market
0.882
-0.002
-0.290
-0.109
-0.172
-0.087
0.022
-0.003
-0.012
0.911
0.089
23. Participation to labor market: difference between men
and women
-0.804
0.038
0.325
-0.140
0.071
0.124
0.104
-0.053
-0.070
0.812
0.188
24. Exports per capita
0.668
-0.024
-0.038
-0.482
-0.102
-0.195
0.111
0.094
-0.015
0.751
0.249
25. Income inequality: Gini concentration index on
equivalent net household income
-0.656
0.059
0.391
0.285
0.019
0.061
-0.080
-0.129
0.043
0.697
0.303
26. Non-performing entry rate of loans to households
-0.648
0.173
0.103
-0.257
-0.298
0.044
0.026
0.056
0.160
0.646
0.354
27. Unemployment: job seekers aged 15 and over
-0.805
0.051
0.041
0.200
0.123
0.087
-0.022
-0.008
0.034
0.717
0.283
28. Value added: manufacturing
0.601
-0.069
-0.045
-0.668
-0.131
-0.212
-0.008
0.012
0.065
0.881
0.119
29. Value added: public sector
-0.761
-0.190
0.107
0.426
0.195
0.153
0.001
0.035
-0.085
0.878
0.122
35
Variable
Factor 1
Factor 2
Factor 3
Factor 4
Factor 5
Factor 6
Factor 7
Factor 8
Factor 9
Communality
Uniqueness
30. Value added: per capita
0.907
0.170
-0.023
0.054
-0.025
-0.135
0.120
-0.014
-0.078
0.895
0.105
31. Median gross hourly wage of employees born abroad
0.639
-0.153
-0.163
-0.042
0.008
-0.364
0.352
-0.088
-0.149
0.747
0.253
32. Median gross hourly wage of employees born in Italy
0.869
0.142
-0.067
-0.153
-0.040
-0.191
0.198
-0.012
-0.108
0.893
0.107
33. Mean wage of employees
0.833
0.171
-0.079
-0.261
-0.106
-0.171
0.120
0.050
-0.137
0.874
0.126
34. Mean wealth per capita
0.870
0.131
-0.211
-0.066
-0.079
-0.083
0.169
0.018
-0.047
0.868
0.132
35. Population having at least a secondary degree
0.699
0.052
-0.254
-0.007
-0.323
0.163
-0.059
0.086
-0.252
0.762
0.238
36. Immigration of graduates between 25 and 39 years
0.848
0.241
-0.091
0.051
-0.158
-0.112
-0.100
0.067
0.065
0.844
0.156
37. Foreign residents
0.763
0.208
-0.143
-0.186
-0.206
0.099
0.102
0.025
0.156
0.768
0.232
38. Emigration to other Italian regions
-0.335
-0.080
-0.152
0.054
0.097
0.765
0.015
-0.041
-0.057
0.745
0.255
39. Emigration abroad
0.413
-0.231
-0.200
-0.050
-0.062
-0.136
0.579
0.021
-0.177
0.656
0.344
40. Beds in emergency residences for migrants
-0.130
0.038
0.098
0.075
-0.148
-0.037
0.671
0.156
0.163
0.557
0.443
41. Newspaper circulation
0.708
-0.130
-0.220
0.286
0.154
-0.195
0.172
0.080
-0.003
0.747
0.253
Eigenvalues
15.569
4.517
2.981
2.473
1.607
1.352
1.330
1.056
1.027
Difference
11.051
1.537
0.508
0.866
0.255
0.022
0.273
0.030
Proportion
0.380
0.110
0.073
0.060
0.039
0.033
0.032
0.026
0.025
Cumulative proportion
0.380
0.490
0.563
0.623
0.662
0.695
0.728
0.753
0.778
Explained variance
13.468
4.209
2.890
2.691
2.354
1.903
1.640
1.453
1.302
Number of variables
41.000
Number of retained factors
9.000
36
Figure A4.1 - Scree plot of eigenvalues after factors
Table A4.3 – Factor scores correlation matrix, 2017
Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Factor 6 Factor 7 Factor 8 Factor 9
Factor 1
1.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Factor 2
0.00
1.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Factor 3
0.00
0.00
1.00
0.00
0.00
0.00
0.00
0.00
0.00
Factor 4
0.00
0.00
0.00
1.00
0.00
0.00
0.00
0.00
0.00
Factor 5
0.00
0.00
0.00
0.00
1.00
0.00
0.00
0.00
0.00
Factor 6
0.00
0.00
0.00
0.00
0.00
1.00
0.00
0.00
0.00
Factor 7
0.00
0.00
0.00
0.00
0.00
0.00
1.00
0.00
0.00
Factor 8
0.00
0.00
0.00
0.00
0.00
0.00
0.00
1.00
0.00
Factor 9
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
1.00
Table A4.4 – Factor scores correlation matrix, 2012 and 2017
Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Factor 6 Factor 7 Factor 8 Factor 9
Factor 1
1.00
-0.05
0.08
0.09
-0.13
0.01
0.06
0.06
0.07
Factor 2
-0.05
1.00
0.03
-0.03
0.09
0.06
0.06
-0.03
0.09
Factor 3
0.08
0.03
1.00
0.04
-0.01
0.04
-0.19
0.16
0.21
Factor 4
0.09
-0.03
0.04
1.00
-0.16
-0.09
-0.06
0.01
0.06
Factor 5
-0.13
0.09
-0.01
-0.16
1.00
0.15
0.07
0.03
0.02
Factor 6
0.01
0.06
0.04
-0.09
0.15
1.00
0.01
0.08
-0.01
Factor 7
0.06
0.06
-0.19
-0.06
0.07
0.01
1.00
-0.22
-0.17
Factor 8
0.06
-0.03
0.16
0.01
0.03
0.08
-0.22
1.00
0.22
Factor 9
0.07
0.09
0.21
0.06
0.02
-0.01
-0.17
0.22
1.00
37
Figure A4.2 - Factor 2 Crime in densely populated areas vs. Factor 1 Economic well-being
Figure A4.3 – Factor 3 Demographic growth vs. Factor 1 Economic well-being
38
Figure A4.4 - Factor 4 Crime in less industrialized areas vs. Factor 1 Economic well-being
Figure A4.5 - Factor 5 Organized crime violence vs. Factor 1 Economic well-being
39
Figure A4.6 - Factor 6 Arsons and extortions in areas with high emigration vs. Factor 1 Economic well-being
Figure A4.7 - Factor 7 Government management of uncontrolled immigration vs. Factor 1 Economic well-being
40
Figure A4.8 - Factor 8 Crimes against women vs. Factor 1 Economic well-being
Figure A4.9 - Factor 9 House robberies vs. Factor 1 Economic well-being
41
Figure A4.10 – Factor 1 Economic well-being, 2012 and 2017
Figure A4.11 – Factor 2 Crime in densely populated areas, 2012 and 2017
42
Figure A4.12 – Factor 3 Demographic growth, 2012 and 2017
Figure A4.13 – Factor 4 Crime in less industrialized areas, 2012 and 2017
43
Figure A4.14 – Factor 5 Organized crime violence, 2012 and 2017
Figure A4.15 – Factor 6 Arsons and extortions in areas with high emigration, 2012 and 2017
44
Figure A4.16 – Factor 7 Government management of uncontrolled immigration, 2012 and 2017
Figure A4.17 – Factor 8 Crimes against women, 2012 and 2017
45
Figure A4.18 – Factor 9 House robberies, 2012 and 2017
46
Appendix 5 Regression analysis on factor scores
Table A5.1 – Political elections 2018: OLS regressions on factor scores
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
(13)
(14)
(15)
Abstention
Abstention
Abstention
M5S
M5S
M5S
Lega
Lega
Lega
PD
PD
PD
FI
FI
FI
F1 Economic well-being
-3.973***
-0.320
-0.473
-4.531***
-4.327***
-4.551***
5.106***
2.722***
2.751***
2.962***
0.825***
0.796***
-1.311***
-0.784***
-0.621***
(0.266)
(0.265)
(0.357)
(0.266)
(0.251)
(0.294)
(0.437)
(0.259)
(0.314)
(0.258)
(0.176)
(0.243)
(0.156)
(0.126)
(0.133)
F2 Crime in densely
populated areas
-0.438*
-0.025
-0.019
0.697***
0.423**
0.760***
-0.862**
-0.222
-0.435
0.796***
0.254**
0.252
0.252
-0.109
-0.142
(0.237)
(0.146)
(0.142)
(0.199)
(0.212)
(0.241)
(0.341)
(0.230)
(0.282)
(0.236)
(0.113)
(0.159)
(0.193)
(0.136)
(0.151)
F3 Demographic growth
1.146***
0.297
0.385
0.900***
1.868***
1.897***
-1.170***
-1.966***
-1.870***
-1.286***
-0.049
-0.064
0.472***
0.315***
0.224**
(0.239)
(0.223)
(0.252)
(0.222)
(0.230)
(0.259)
(0.427)
(0.214)
(0.206)
(0.280)
(0.129)
(0.153)
(0.163)
(0.107)
(0.108)
F4 Crime in non-
industrialized areas
2.028***
1.067***
1.130***
0.600**
0.572**
0.789**
-2.496***
-0.865***
-0.922***
-0.447*
-0.490***
-0.573***
-0.400**
0.090
0.029
(0.233)
(0.230)
(0.268)
(0.262)
(0.247)
(0.307)
(0.450)
(0.233)
(0.272)
(0.227)
(0.102)
(0.129)
(0.162)
(0.127)
(0.137)
F5 Organized crime violence
1.667
***
0.494
***
0.553
***
-0.175
0.422
0.392
-0.910
***
-0.808
***
-0.653
***
-0.734
***
-0.393
***
-0.460
***
0.217
0.435
**
0.373
***
(0.282)
(0.174)
(0.152)
(0.385)
(0.317)
(0.302)
(0.185)
(0.139)
(0.166)
(0.226)
(0.131)
(0.144)
(0.172)
(0.184)
(0.112)
F6 Arsons and extortions in
areas with high emigration
0.446*
-0.186
-0.127
0.313
0.678***
0.580**
-1.327***
-0.315*
-0.299
-0.091
-0.128
-0.143
0.419**
0.350**
0.374***
(0.238)
(0.145)
(0.165)
(0.253)
(0.225)
(0.244)
(0.270)
(0.177)
(0.190)
(0.204)
(0.112)
(0.142)
(0.175)
(0.135)
(0.121)
F7 Government management
of uncontrolled immigration
1.185***
0.908***
0.972**
-1.116***
-1.016***
-1.308***
0.356
-0.436**
-0.269
-0.837***
0.042
0.195
-0.174
0.161
0.071
(0.253)
(0.287)
(0.426)
(0.222)
(0.236)
(0.347)
(0.584)
(0.180)
(0.269)
(0.285)
(0.135)
(0.204)
(0.164)
(0.104)
(0.183)
F8 Crimes against women
-0.001
-0.113
-0.143
0.435*
0.238
-0.164
-0.311
0.024
0.499**
-0.373*
-0.080
0.029
0.061
-0.129
-0.238*
(0.255)
(0.110)
(0.158)
(0.227)
(0.194)
(0.312)
(0.282)
(0.160)
(0.208)
(0.214)
(0.108)
(0.137)
(0.115)
(0.081)
(0.124)
F9 House robberies
-0.268
-0.048
0.036
0.351
0.068
-0.082
-0.310
0.333*
0.313
0.747***
0.076
0.043
0.093
-0.010
0.030
(0.252)
(0.144)
(0.137)
(0.255)
(0.239)
(0.240)
(0.330)
(0.195)
(0.233)
(0.254)
(0.119)
(0.152)
(0.153)
(0.106)
(0.117)
Abstention in 2013
0.720***
0.708***
(0.054)
(0.065)
M5S in 2013
0.559***
0.611***
(0.092)
(0.096)
Lega in 2013
1.185***
1.209***
(0.076)
(0.077)
PD in 2013
0.653***
0.652***
(0.049)
(0.053)
FI in 2013
0.522***
0.537***
(0.046)
(0.049)
(...)
47
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
(13)
(14)
(15)
Abstention
Abstention
Abstention
M5S
M5S
M5S
Lega
Lega
Lega
PD
PD
PD
FI
FI
FI
F1 Change 2012-2017
0.834
3.108
-2.726
-0.572
0.163
(1.225)
(2.260)
(1.660)
(1.052)
(0.790)
F2 Change 2012-2017
0.500
1.407*
-0.408
-0.364
-0.575*
(0.485)
(0.826)
(0.655)
(0.499)
(0.332)
F3 Change 2012-2017
-0.251
0.803
0.457
-0.301
-0.308
(0.665)
(1.161)
(0.969)
(0.578)
(0.472)
F4 Change 2012-2017
-0.376
-1.071
-0.019
0.746
0.729*
(0.478)
(0.874)
(0.514)
(0.473)
(0.430)
F5 Change 2012-2017
-0.147
-0.028
0.214
0.238
-0.061
(0.249)
(0.474)
(0.296)
(0.231)
(0.182)
F6 Change 2012-2017
-0.080
-0.948
0.128
0.274
0.316
(0.329)
(0.708)
(0.472)
(0.386)
(0.331)
F7 Change 2012-2017
-0.080
1.350***
-0.586
-0.309
-0.001
(0.355)
(0.503)
(0.431)
(0.328)
(0.254)
F8 Change 2012-2017
0.247
0.372
-0.747**
-0.293*
0.165
(0.203)
(0.370)
(0.298)
(0.163)
(0.143)
F9 Change 2012-2017
-0.368*
0.135
0.038
0.283
-0.071
(0.188)
(0.332)
(0.276)
(0.191)
(0.157)
Constant
29.684***
9.535***
9.661***
21.819***
11.312***
11.636***
12.278***
9.161***
8.469***
12.531***
0.334
0.306
9.507***
1.596**
1.260*
(0.247)
(1.554)
(2.032)
(0.284)
(1.746)
(1.961)
(0.364)
(0.294)
(0.765)
(0.272)
(0.873)
(0.954)
(0.152)
(0.688)
(0.678)
No. of observations
110
110
110
110
110
110
110
110
110
109
109
109
109
109
109
R-squared
.808
.938
.942
.746
.817
.851
.736
.914
.925
.645
.929
.935
.514
.764
.815
F test
51.6***
255***
145***
57.3***
72.3***
44.8***
59.9***
145***
95.5***
20.1***
77.9***
58.3***
11.3***
27.7***
21.5***
Notes. Robust standard errors in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01.
48
Appendix 6 Regression analysis on selected variables
Table A6.1 - Political elections 2018: Regressions on selected variables
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
(13)
(14)
(15)
M5S+Lega
M5S+Lega
M5S+Lega
M5S
M5S
M5S
Lega
Lega
Lega
PD
PD
PD
FI
FI
FI
Value added: per capita
change 2012-2017
0.322
0.415
0.628
-1.382***
-1.316***
-0.660
1.704***
1.732***
1.288**
0.815**
0.866**
0.461
-0.559***
-0.538***
-0.237
(0.460)
(0.428)
(0.393)
(0.433)
(0.419)
(0.426)
(0.603)
(0.574)
(0.570)
(0.340)
(0.341)
(0.389)
(0.166)
(0.156)
(0.162)
Population density
-0.362
-0.098
0.137
0.646
0.419
1.143
-1.008
-0.517
-1.006
0.063
0.029
-0.394
0.354
0.292
0.607
(1.022)
(0.859)
(0.902)
(1.082)
(1.105)
(1.046)
(1.423)
(1.341)
(1.488)
(0.769)
(0.724)
(0.894)
(0.470)
(0.481)
(0.690)
Isolation
-0.059
-0.062*
-0.059*
-0.039
-0.019
-0.008
-0.020
-0.043
-0.050
-0.008
-0.001
-0.008
-0.038***
-0.037***
-0.032***
(0.036)
(0.035)
(0.034)
(0.030)
(0.029)
(0.029)
(0.047)
(0.046)
(0.046)
(0.025)
(0.024)
(0.025)
(0.010)
(0.011)
(0.010)
Elderly population
0.010
0.024
-0.015
-0.194
-0.207
-0.327**
0.204
0.231
0.312
0.287
0.285
0.348*
-0.102
-0.111
-0.158**
(0.295)
(0.282)
(0.253)
(0.197)
(0.177)
(0.158)
(0.301)
(0.284)
(0.281)
(0.196)
(0.190)
(0.189)
(0.077)
(0.073)
(0.068)
Education
0.008
-0.063
-0.037
-0.148
**
-0.095
-0.015
0.156
*
0.032
-0.023
0.119
*
0.125
*
0.079
-0.068
**
-0.054
*
-0.020
(0.072)
(0.073)
(0.071)
(0.065)
(0.071)
(0.076)
(0.081)
(0.082)
(0.086)
(0.061)
(0.065)
(0.061)
(0.026)
(0.029)
(0.030)
Net migration
0.075
0.058
0.178
-0.643***
-0.748***
-0.376**
0.718***
0.806***
0.554**
0.439***
0.391***
0.200
-0.192***
-0.214***
-0.071
(0.159)
(0.172)
(0.173)
(0.161)
(0.158)
(0.161)
(0.187)
(0.197)
(0.223)
(0.125)
(0.137)
(0.175)
(0.070)
(0.069)
(0.074)
Management of uncontrolled
immigration
-0.038
-0.037
-0.017
-0.014
-0.021
-0.023
-0.017
-0.020
-0.014*
-0.013*
(0.027)
(0.025)
(0.020)
(0.015)
(0.023)
(0.023)
(0.014)
(0.014)
(0.008)
(0.007)
House robberies
-2.082
-2.480
6.509**
5.280**
-8.592**
-7.760**
2.235
2.639
1.323
1.023
(2.628)
(2.409)
(2.491)
(2.167)
(3.384)
(3.291)
(2.590)
(2.487)
(1.029)
(0.956)
Intentional homicides
-13.551*
-14.000*
0.284
-1.102
-13.835**
-12.898**
-2.840
-2.534
1.617
1.389
(7.187)
(7.520)
(7.700)
(8.628)
(5.828)
(5.619)
(3.478)
(3.721)
(2.849)
(2.312)
Value added: per capita 2017
-0.125
-0.387***
0.262*
0.209*
-0.155***
(0.133)
(0.094)
(0.149)
(0.118)
(0.057)
Constant
36.189***
42.228***
44.138***
38.892***
32.951***
38.852***
-2.703
9.277
5.285
-2.013
-3.078
-5.827
18.597***
17.462***
19.507***
(5.073)
(6.252)
(5.763)
(5.484)
(5.569)
(5.144)
(7.286)
(8.100)
(7.869)
(5.217)
(6.339)
(6.376)
(2.042)
(2.199)
(2.472)
No. of observations
110
110
110
110
110
110
110
110
110
109
109
109
109
109
109
R-squared
.0851
.158
.17
.506
.532
.593
.456
.5
.517
.485
.495
.519
.48
.501
.56
F test
2.16*
2.07**
2.05**
22.6***
17.3***
21.2***
19.5***
19.4***
19.5***
16.5***
13.5***
13.3***
19.9***
14.7***
13.5***
Notes. The economic variables are expressed in thousands of Euros. Robust standard errors in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01
49
Appendix 7 Panel regressions on selected variables
Table A7.1 - Political elections 2013 and 2018: Panel regressions on selected variables
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
Abstention
Abstention
M5S
M5S
Lega
Lega
PD
PD
FI
FI
(FE)
(RE)
(FE)
(RE)
(FE)
(RE)
(FE)
(RE)
(FE)
(RE)
Value added: per capita
0.503**
-0.356***
-2.138***
-0.433***
1.840***
0.529***
-0.106
0.290***
-0.240
-0.285***
(0.206)
(0.074)
(0.461)
(0.079)
(0.356)
(0.097)
(0.196)
(0.074)
(0.178)
(0.038)
Population density
3.311
1.467
34.373
0.308
-40.417
-1.107
22.323
0.019
-7.296
0.822*
(14.427)
(1.107)
(32.186)
(0.996)
(24.890)
(1.362)
(13.578)
(1.105)
(12.333)
(0.491)
Isolation
0.000
0.095***
0.000
-0.043*
0.000
-0.019
0.000
-0.019
0.000
-0.044***
(.)
(0.026)
(.)
(0.023)
(.)
(0.031)
(.)
(0.025)
(.)
(0.011)
Elderly population
-0.703*
-0.403***
2.056**
0.114
-1.163
-0.084
1.021***
0.734***
-0.080
-0.216***
(0.407)
(0.155)
(0.909)
(0.152)
(0.703)
(0.198)
(0.383)
(0.150)
(0.348)
(0.073)
Education
0.008
-0.069
-0.375***
-0.106*
0.216**
0.155**
0.028
0.024
-0.011
-0.011
(0.061)
(0.047)
(0.135)
(0.057)
(0.105)
(0.067)
(0.057)
(0.044)
(0.052)
(0.026)
Net migration
-0.168**
-0.236***
0.551***
0.272**
-0.395***
-0.198*
0.026
0.055
0.177**
0.173***
(0.081)
(0.076)
(0.180)
(0.123)
(0.139)
(0.119)
(0.076)
(0.069)
(0.069)
(0.051)
year=2013
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
(.)
(.)
(.)
(.)
(.)
(.)
(.)
(.)
(.)
(.)
year=2018
1.925**
2.421***
4.341**
4.326***
8.174***
8.201***
-7.868***
-7.599***
-4.799***
-4.582***
(0.876)
(0.433)
(1.954)
(0.698)
(1.511)
(0.647)
(0.824)
(0.401)
(0.748)
(0.274)
Constant
31.001**
43.968***
32.955
33.331***
-14.171
-14.207***
-8.446
-4.202
24.248**
28.423***
(12.104)
(4.202)
(27.002)
(4.235)
(20.882)
(5.394)
(11.389)
(4.103)
(10.344)
(1.999)
No. of observations
220
220
220
220
220
220
218
218
218
218
No. of provinces
110
110
110
110
110
110
109
109
109
109
R-squared: within
.457
.311
.499
.335
.885
.858
.908
.9
.91
.91
R-squared: between
.116
.635
.136
.298
.0646
.307
.0355
.461
.0121
.492
R-squared: overall
.0767
.615
.119
.306
.141
.582
.11
.59
.31
.756
Hausman test chi-squared
838
42.8
35.1
17.4
1.13
Hausman test p-value for the chi-squared
8.8e-178
1.30e-07
4.18e-06
.00795
.98
Notes. The economic variables are expressed in thousands of Euros. Standard errors in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01
50
Table A7.2 - Political elections 2013 and 2018: Panel regressions on selected variables
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
Abstention
Abstention
M5S
M5S
Lega
Lega
PD
PD
FI
FI
(FE)
(RE)
(FE)
(RE)
(FE)
(RE)
(FE)
(RE)
(FE)
(RE)
Value added: per capita
0.437*
-0.145**
-1.832***
-0.354***
1.496***
0.390***
-0.359*
0.144*
-0.052
-0.279***
(0.233)
(0.073)
(0.515)
(0.085)
(0.386)
(0.107)
(0.214)
(0.083)
(0.195)
(0.044)
Population density
4.012
1.015
25.238
0.073
-34.692
-0.640
36.668***
0.387
-9.729
0.829*
(15.512)
(0.911)
(34.276)
(0.971)
(25.723)
(1.290)
(13.955)
(1.063)
(12.675)
(0.497)
Isolation
0.000
0.060***
0.000
-0.059**
0.000
0.010
0.000
0.010
0.000
-0.047***
(.)
(0.022)
(.)
(0.025)
(.)
(0.032)
(.)
(0.026)
(.)
(0.012)
Elderly population
-0.821*
-0.213
2.712**
0.113
-1.713**
-0.092
0.538
0.687***
0.164
-0.225***
(0.474)
(0.131)
(1.048)
(0.147)
(0.786)
(0.189)
(0.428)
(0.146)
(0.388)
(0.073)
Education
0.007
-0.054
-0.374***
-0.121**
0.190*
0.140**
0.039
0.019
-0.005
-0.022
(0.062)
(0.043)
(0.137)
(0.058)
(0.103)
(0.066)
(0.056)
(0.044)
(0.051)
(0.027)
Net migration
-0.165**
-0.267***
0.551***
0.302**
-0.424***
-0.237*
-0.013
0.062
0.189***
0.172***
(0.083)
(0.074)
(0.184)
(0.129)
(0.138)
(0.122)
(0.075)
(0.070)
(0.068)
(0.053)
Foreign residents
-0.035
-0.422***
0.392
-0.208**
-0.328
0.316***
-0.572***
0.285***
0.116
-0.017
(0.216)
(0.076)
(0.477)
(0.093)
(0.358)
(0.112)
(0.201)
(0.082)
(0.182)
(0.045)
Management of uncontrolled
immigration
0.001
0.021*
-0.010
0.006
-0.017
-0.037*
0.017
0.002
-0.001
-0.007
(0.013)
(0.012)
(0.028)
(0.021)
(0.021)
(0.019)
(0.011)
(0.011)
(0.010)
(0.008)
House robberies
0.184
2.160**
-0.405
-0.356
2.338
1.669
0.790
0.318
-1.406*
-0.391
(0.964)
(0.854)
(2.130)
(1.417)
(1.599)
(1.387)
(0.866)
(0.809)
(0.787)
(0.591)
Intentional homicides
3.393
5.711**
-10.665
-9.768**
8.439
3.143
-3.217
-1.334
-5.264**
-3.095*
(3.086)
(2.593)
(6.820)
(3.880)
(5.118)
(4.116)
(2.779)
(2.514)
(2.524)
(1.717)
year=2013
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
(.)
(.)
(.)
(.)
(.)
(.)
(.)
(.)
(.)
(.)
year=2018
2.370*
3.104***
1.914
4.507***
10.817***
8.206***
-5.817***
-7.917***
-5.992***
-4.607***
(1.259)
(0.423)
(2.781)
(0.740)
(2.087)
(0.680)
(1.145)
(0.416)
(1.040)
(0.292)
Constant
34.861**
38.656***
11.508
36.292***
6.882
-15.726***
8.844
-4.166
15.114
29.933***
(14.536)
(3.744)
(32.120)
(4.667)
(24.106)
(5.528)
(13.191)
(4.107)
(11.981)
(2.193)
No. of observations
220
220
220
220
220
220
218
218
218
218
No. of provinces
110
110
110
110
110
110
109
109
109
109
R-squared: within
.465
.341
.515
.346
.895
.86
.917
.897
.919
.914
R-squared: between
.0218
.756
.0949
.357
.00709
.375
.0188
.54
.0659
.49
R-squared: overall
.0091
.732
.104
.345
.095
.619
.0000525
.645
.108
.757
Hausman test chi-squared
43.5
47.7
24.3
19.3
20.2
Hausman test p-value for the chi-
squared
1.71e-06
6.98e-07
.0069
.0369
.027
Notes. The economic variables are expressed in thousands of Euros. Standard errors in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01
51
Table A7.3 - Political elections 2013 and 2018: Fixed effects panel regressions on selected variables
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
Abstention
Abstention
M5S
M5S
Lega
Lega
PD
PD
FI
FI
(FE)
(FE)
(FE)
(FE)
(FE)
(FE)
(FE)
(FE)
(FE)
(FE)
Value added: per capita
0.503**
0.437*
-2.138***
-1.832***
1.840***
1.496***
-0.106
-0.359*
-0.240
-0.052
(0.206)
(0.233)
(0.461)
(0.515)
(0.356)
(0.386)
(0.196)
(0.214)
(0.178)
(0.195)
Population density
3.311
4.012
34.373
25.238
-40.417
-34.692
22.323
36.668***
-7.296
-9.729
(14.427)
(15.512)
(32.186)
(34.276)
(24.890)
(25.723)
(13.578)
(13.955)
(12.333)
(12.675)
Isolation
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
(.)
(.)
(.)
(.)
(.)
(.)
(.)
(.)
(.)
(.)
Elderly population
-0.703*
-0.821*
2.056**
2.712**
-1.163
-1.713**
1.021***
0.538
-0.080
0.164
(0.407)
(0.474)
(0.909)
(1.048)
(0.703)
(0.786)
(0.383)
(0.428)
(0.348)
(0.388)
Education
0.008
0.007
-0.375***
-0.374***
0.216**
0.190*
0.028
0.039
-0.011
-0.005
(0.061)
(0.062)
(0.135)
(0.137)
(0.105)
(0.103)
(0.057)
(0.056)
(0.052)
(0.051)
Net migration
-0.168**
-0.165**
0.551***
0.551***
-0.395***
-0.424***
0.026
-0.013
0.177**
0.189***
(0.081)
(0.083)
(0.180)
(0.184)
(0.139)
(0.138)
(0.076)
(0.075)
(0.069)
(0.068)
year=2013
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
(.)
(.)
(.)
(.)
(.)
(.)
(.)
(.)
(.)
(.)
year=2018
1.925**
2.370*
4.341**
1.914
8.174***
10.817***
-7.868***
-5.817***
-4.799***
-5.992***
(0.876)
(1.259)
(1.954)
(2.781)
(1.511)
(2.087)
(0.824)
(1.145)
(0.748)
(1.040)
Foreign residents
-0.035
0.392
-0.328
-0.572***
0.116
(0.216)
(0.477)
(0.358)
(0.201)
(0.182)
Management of
uncontrolled immigration
0.001
-0.010
-0.017
0.017
-0.001
(0.013)
(0.028)
(0.021)
(0.011)
(0.010)
House robberies
0.184
-0.405
2.338
0.790
-1.406*
(0.964)
(2.130)
(1.599)
(0.866)
(0.787)
Intentional homicides
3.393
-10.665
8.439
-3.217
-5.264**
(3.086)
(6.820)
(5.118)
(2.779)
(2.524)
Constant
31.001**
34.861**
32.955
11.508
-14.171
6.882
-8.446
8.844
24.248**
15.114
(12.104)
(14.536)
(27.002)
(32.120)
(20.882)
(24.106)
(11.389)
(13.191)
(10.344)
(11.981)
No. of observations
220
220
220
220
220
220
218
218
218
218
No. of provinces
110
110
110
110
110
110
109
109
109
109
R-squared: within
.457
.465
.499
.515
.885
.895
.908
.917
.91
.919
R-squared: between
.116
.0218
.136
.0949
.0646
.00709
.0355
.0188
.0121
.0659
R-squared: overall
.0767
.0091
.119
.104
.141
.095
.11
.0000525
.31
.108
Hausman test chi-
squared
838
43.5
42.8
47.7
35.1
24.3
17.4
19.3
1.13
20.2
Hausman test p-value for
the chi-squared
8.8e-178
1.71e-06
1.30e-07
6.98e-07
4.18e-06
.0069
.00795
.0369
.98
.027
Notes. The economic variables are expressed in thousands of Euros. Standard errors in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01
52
Table A7.4 - Political elections 2013 and 2018: Random effects panel regressions on selected variables
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
Abstention
Abstention
M5S
M5S
Lega
Lega
PD
PD
FI
FI
(RE)
(RE)
(RE)
(RE)
(RE)
(RE)
(RE)
(RE)
(RE)
(RE)
Value added: per capita
-0.356***
-0.145**
-0.433***
-0.354***
0.529***
0.390***
0.290***
0.144*
-0.285***
-0.279***
(0.074)
(0.073)
(0.079)
(0.085)
(0.097)
(0.107)
(0.074)
(0.083)
(0.038)
(0.044)
Population density
1.467
1.015
0.308
0.073
-1.107
-0.640
0.019
0.387
0.822*
0.829*
(1.107)
(0.911)
(0.996)
(0.971)
(1.362)
(1.290)
(1.105)
(1.063)
(0.491)
(0.497)
Isolation
0.095***
0.060***
-0.043*
-0.059**
-0.019
0.010
-0.019
0.010
-0.044***
-0.047***
(0.026)
(0.022)
(0.023)
(0.025)
(0.031)
(0.032)
(0.025)
(0.026)
(0.011)
(0.012)
Elderly population
-0.403***
-0.213
0.114
0.113
-0.084
-0.092
0.734***
0.687***
-0.216***
-0.225***
(0.155)
(0.131)
(0.152)
(0.147)
(0.198)
(0.189)
(0.150)
(0.146)
(0.073)
(0.073)
Education
-0.069
-0.054
-0.106*
-0.121**
0.155**
0.140**
0.024
0.019
-0.011
-0.022
(0.047)
(0.043)
(0.057)
(0.058)
(0.067)
(0.066)
(0.044)
(0.044)
(0.026)
(0.027)
Net migration
-0.236***
-0.267***
0.272**
0.302**
-0.198*
-0.237*
0.055
0.062
0.173***
0.172***
(0.076)
(0.074)
(0.123)
(0.129)
(0.119)
(0.122)
(0.069)
(0.070)
(0.051)
(0.053)
year=2013
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
(.)
(.)
(.)
(.)
(.)
(.)
(.)
(.)
(.)
(.)
year=2018
2.421***
3.104***
4.326***
4.507***
8.201***
8.206***
-7.599***
-7.917***
-4.582***
-4.607***
(0.433)
(0.423)
(0.698)
(0.740)
(0.647)
(0.680)
(0.401)
(0.416)
(0.274)
(0.292)
Foreign residents
-0.422***
-0.208**
0.316***
0.285***
-0.017
(0.076)
(0.093)
(0.112)
(0.082)
(0.045)
Management of
uncontrolled immigration
0.021*
0.006
-0.037*
0.002
-0.007
(0.012)
(0.021)
(0.019)
(0.011)
(0.008)
House robberies
2.160**
-0.356
1.669
0.318
-0.391
(0.854)
(1.417)
(1.387)
(0.809)
(0.591)
Intentional homicides
5.711**
-9.768**
3.143
-1.334
-3.095*
(2.593)
(3.880)
(4.116)
(2.514)
(1.717)
Constant
43.968***
38.656***
33.331***
36.292***
-14.207***
-15.726***
-4.202
-4.166
28.423***
29.933***
(4.202)
(3.744)
(4.235)
(4.667)
(5.394)
(5.528)
(4.103)
(4.107)
(1.999)
(2.193)
No. of observations
220
220
220
220
220
220
218
218
218
218
No. of provinces
110
110
110
110
110
110
109
109
109
109
R-squared: within
.311
.341
.335
.346
.858
.86
.9
.897
.91
.914
R-squared: between
.635
.756
.298
.357
.307
.375
.461
.54
.492
.49
R-squared: overall
.615
.732
.306
.345
.582
.619
.59
.645
.756
.757
Notes. The economic variables are expressed in thousands of Euros. Standard errors in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01
53
Appendix 8 Selected social and economic data for Italy
Figure A8.1 – Total Population
Figure A8.2 – Real Gross Domestic Product
54
Figure A8.3 – Constant GDP per Capita
Figure A8.4 – General Government Gross Debt
55
Figure A8.5 – Long−Term Government Bond Yields
Figure A8.6 – Spread between Italian and German Long−Term Government Bond Yields
56
Figure A8.7 – General Government Net Lending−Borrowing
Figure A8.8 – Current Account Balance
57
Figure A8.9 – Unemployment Rate
Figure A8.10 – Value Added of the Industrial Sector
58
Figure A8.11 – Migrants Disembarked
