1

Máster en Democracia y Gobierno

Departamento de Ciencia Política y Relaciones Internacionales Universidad Autónoma de Madrid

Working Papers Online Series http://www.uam.es/ss/Satellite/Derecho/es/1242658791834/listado

Combo/Working_Papers.htm

Estudio/Working Paper 156/2014

Availability or disaffection? How Italian citizens reacted to the

two-faced parliamentary grand coalition supporting the Monti

government

Paolo Segatti

Università degli Studi di Milano

(paolo.segatti@unimi.it)

*This is a paper also written with Federico Vegetti Central European University - Budapest

(fede.vegetti@gmail.com) and Monica Poletti Università degli Studi di Milano / London School of

Economics (monica.poletti@unimi.it)

2

Abstract

This study investigates the impact that the parliamentary grand coalition in support to the

Monti government from November 2011 to February 2013 had on the Italian electorate. In

particular, we focus on the consequences that the cohabitation of the Pd and the Pdl within

the same parliamentary coalition had on the reciprocal availability of their electorates, and on

voters' disaffection from the traditional parties. We argue that the contradictory behaviour of

the party elites, made of public quarrels and support of the technocratic government in the

institutional rooms, has led to a temporary disorientation of the electorate. This, in turn, has

temporally depolarized the electorate in respect of the two parties' evaluations, but most

importantly, it led to a persistent increase of anti-politics feelings among the public. To

support our claims, we rely on evidence based on public opinion data covering a time period

from March 2011 to February 2013.

3

1. Grand coalition, Italian style

Times may come in which democracies have to suspend electoral competition for the sake of

the general interest. Wars, divisive memories of past conflicts, a prolonged economic crisis,

but also electoral outcomes without a clear and politically viable governing majority are the

challenges that may oblige ideological distant parties to join their forces in a grand coalition

government. Although this decision might be seen as necessary, this is a difficult move for

any party, since it may alter the conditions under which voters usually make their voting

choices at the following election. When the contraposition between government and

opposition is suspended, voters might find it hard to take into account what has been done

and who is responsible for what. The risk of blurring the distinctiveness of party policy

stances is high, since voters might conclude there is no difference between parties, and all are

equally unresponsive. Moreover, when this comes together with the discovery of a vast web

of scandals in which politicians affiliated to different parties are involved, citizens’ political

disaffection is likely to further increase.

In 2011, Italian parties found themselves in such a situation. Starting in the summer, the

Italian financial state worsened to the point that, between October and November 2011, the

risk of a sovereign default was tangible. This was indicated by the increasing spread between

the Italian Buoni del Tesoro and the German Bund. Concerns about the Italian crisis and its

systemic consequences on the Eurozone were widely shared in Italy as well as abroad. “Fare

presto!” (be quick!) was the claim of the influential employers' association newspaper Il sole

24 Ore, inviting the right-wing government led by Berlusconi as well as the opposition

parties to take immediate joint action against the financial crisis. The international news

magazine The Economist conveyed a similar worrying message portraying the Euro currency

hanged loose from the Italian “boot” on the front cover. The message from inside as well as

outside the country was clear. Italy needed urgent structural reforms able to restore

confidence among its EU partners and financial investors. The expectation was that only a

grand coalition between the main parties would have been able to overcome the several veto

points that over the years have led Italy on the edge of the cliff.

Yet, the way parties decided to reply to the challenge was highly peculiar as to four aspects,

at least in comparison with the experience of grand coalitions in Germany and Austria in the

last fifty years. First, in 2011, a coalition between the main political parties was not an option

that emerged after a competitive election, where voters could have taken into account what

the incumbent government did over the previous years, as it happened in Germany and

4

Austria1. Rather, it turned out to be an alternative to calling for a new election. One might

think of several reasons for this choice: a) the need of immediate action against the

impending risk of financial bankruptcy; b) the agreement that new elections would not have

expressed a clear mandate, given the current electoral law, which favours the creation of

different majorities in the two chambers; c) the shared perception among leading political as

well as economic actors that the opposition party, the Democratic Party (Pd), was not ready to

govern; d) a simple calculus by the incumbent government to avoid the voters’ blame.

Historians would tell us which of these reasons was prominent, if any, at that time. De facto,

at the beginning of November 2011, Silvio Berlusconi resigned from its role of Prime

Minister because of the collapse of his majority. Instead of calling for a new election, the

President of the Republic nominated a highly-respected economist and former EU

Commissioner as new Prime Minister: Mario Monti2.

Second, although the main political parties were asked by the President of the Republic to

join the government, they refused to form a grand coalition. Instead, they only accepted to

support the government in Parliament. Thus, instead of a governing grand coalition, the

result of these manoeuvres turned out to be a technocratic cabinet, composed exclusively by

non-political figures (such as university professors, top bureaucrats, prefects, lawyers,

diplomats and one admiral), but supported by an oversized parliamentary coalition that

included the two major rival parties, the left-wing Democratic Party (Pd) and the right-wing

People of Freedom party (Pdl), as well as a relatively small centrist party, the Union of the

Centre (Udc).

Third, the two main parties kept on quarrelling endlessly with each other, openly and

publicly, and often criticized what the technocratic government was doing, despite the fact

that they were jointly supporting these actions in Parliament. This peculiar pattern of

contentious coalitional behaviour lasted until December 2012, when Berlusconi decided to

stop the support to the Monti cabinet, paving the way for new elections in the following

spring. To complicate the scenario, both the Pd and the Pdl were internally fragmented, with

deputies and political figures pursuing their faction’s policy or office agenda.

All in all, the management of the crucial linkage between the cabinet and the oversized

parliamentary majority was very different from the typical pattern seen with German and

Austrian grand coalitions. Miller and Müller (2011) reported that three management

1 The German Grand Coalition 1966-1969 between Cdu-Csu and Spd was the result of the collapse of the

previous coalition between the former and the Fdp.

2 Just a few days before being nominated Prime Minister he was nominated long life senator by the President

of Republic.

5

mechanisms have been frequently used in the German and Austrian experiences: public

coalition agreement, coalition committees, and ‘watchdog’ junior ministers belonging to

different parties from the “senior” ministers. All of them are helpful mechanisms to pre-empt

the constant risk of conflict between coalition parties, and not less important, to reduce the

risk that cabinet members would feel free to follow their own agenda instead of the

government platform parties agreed on. According to what was reported by the media during

those days, however, no one of these mechanisms seemed to be in place in the case of the

Monti government and its “strana maggioranza” (odd majority, as pundits labelled it)3.

All in all, what emerged was a two-faced pattern of competition, in which parties in

Parliament diligently agreed on all the unpopular reforms the Monti government promoted to

rein in the public debt crisis, while at the same time, in the public arena, made of everyday

talk shows and overflooding press statements, they kept on squabbling with one another on

whatsoever events were on the public agenda. Parties seemed to motivate such a two-faced

coalitional behaviour on the basis of the notion that their voters were unlikely to understand

and finally accept their joint responsibility in the government. It is hard to say what their

voters would have preferred. One might put forward, however, an opposite opinion from the

one that parties had. Parties’ decision to stand half-way, supporting the government in

Parliament but not taking any governing responsibility for it—or, even worse, jointly

agreeing behind the screen on what they were quarrelling on in front of the public—made it

even harder for voters to understand what was going on. In this context voters were exposed

to a chaotic flow of information in which it was hard to disentangle who was doing what. The

only accessible and robust piece of evidence for many voters was the fact that the main

parties coalesced in supporting a technocratic government, while at the same time they were

constantly fighting with each other. This might have been taken as evidence that politicians

were de facto trying to mislead and manipulate the citizens. In a climate of rampant distrust

towards parties, it is very likely the public held this opinion.

This paper follows this rationale. We argue that the two-faced coalitional behaviour adopted

by the Pd and the Pdl played an important role in building the contextual conditions under

3 Party leaders seemed unaware of the long-term management problems faced by a coalition between

opponent parties. Among other things, this unawareness was captured by the generally-proud tone of their

statements while noting that the decision to support together the Monti Government in November 2011 had

been made in a rather short time. For instance, in his speech during the confidence vote to the Monti

Government on November 18th

2011, the leader of the Pd, on. Pier Luigi Bersani, congratulated the

Parliament because a new parliamentary majority was built only after ten days from the resignation of

Berlusconi. “This means that, at the end, we are Italians, and we are still able to surprise, when we believe in

ourselves” (Bersani 2011).

6

which Italian citizens went to vote in February 2013, contributing to the great electoral

earthquake and the chaotic outcome that followed. In the next section, we discuss which

expectations we can draw from the two-faced contentious coalitional behaviour of the two

political parties. Then we briefly refer to the literature on the influence of contextual factors

in increasing the ideological grip on vote and on close determinants of party choice. Data

analysis follows and conclusions end the paper.

2. Expectations from a two-faced party competition

Literature on government coalitions and related party and voters' behaviour is vast. Large part

of it is devoted to coalition building and selection of ministers. An increasing number of

studies show that coalition is consequential on voters’ party choice. In particular, coalition

governments may hamper the clarity of responsibility that make possible for the voters to get

the rascal out (Maravall 2010, Hobolt & Karp 2010). Much less common are studies that

focus on the electoral consequences of the grand coalitions. This literature refers almost

exclusively to the German experience, and shows that the electoral price that parties pay for

joining a grand coalition government is generally rather high. Careful coalition management

may reduce it, but only up to a certain point. Scarrow (2012) shows that, in the elections

immediately following the 2005-2009 German grand coalition, turnout decreased, volatility

raised, party fragmentation increased and protest voting became more vibrant. On the same

line, Banazsak and Doerschler (2012) reports that, in elections after grand coalitions, voters

tend to move away from the coalitional parties towards opposition parties, and this movement

is more likely among the more radical voters of the parties.

Studies about the Italian elections in February 2013 showed similar phenomena, albeit larger

in magnitude. The turnout decline was unprecedented, i.e. five percentage points less than the

previous election in 2008: the greatest change since 1946. Electoral volatility was so high that

2013 election ranks first among the most volatile Italian post-WWII elections (Diamanti

2013; Itanes 2013). Although vote-switching between Pd and Pdl actually decreased since the

previous 2008 election, the two main parties that supported the Monti Government were the

most strongly affected by voters’ defection. Pd and Pdl lost together almost ten millions of

votes since 2008 (3.5 million and 6.5 million respectively). Moreover, about one out of four

valid votes went to a new party that claimed to be out of the traditional left-right ideological

divisions, the Five Star Movement (M5s), turning an almost twenty-years-old bipolar party

7

system into a three-polar setting4.

Because of these events, newspapers claimed that the pattern of competition of the Second

Republic, largely based on the left-right contraposition, had come to an end. Their analysis,

however, might have been too hasty. In fact, as De Sio and Schadee (2013) and Baldassari

(2013) show, regardless of the massive electoral change, Italian voters still think of left-right

as the predominant dimension of the national political space. Yet, they also find evidence of

the emergence of a second axis of competition that may be interpreted as a division between

the pro-establishment parties and the anti-establishment M5s. In addition, a comparison

between voters who remained loyal to the Pd and the Pdl since 2008 and those who defected

shows that the latter are pretty similar to each other regarding policy preferences and attitudes

toward politics: their policy preferences are quite moderated and different from the voters

who remained loyal to the parties, while their disaffection towards politics is stronger.

Finally, it has been shown that the Pd and Pdl defectors are not on the fringe of the left and

right continuum, but rather located around the centre (Vezzoni, 2013; Passarelli and Tuorto

2013). This is a quite different scenario from the effects of the German grand coalition

mentioned above.

Bellucci and Segatti (2013) recently argued that voters’ party choice during the 2013 election

might have been conditioned by the context in which Italians made up their decisions. They

referred to three contexts: the economic crisis, the wave of disaffection towards politics

nurtured by scandals and abuse of political money, and the joint support of the main parties to

the Monti government that made blame-attributions harder for the voters. However, no

empirical evidence has been provided about the effects of these phenomena on people's

preferences. Building on their intuition, we provide a first empirical assessment of how voters

have been reacting to the experience of an inter-block parliamentary coalition in the period

before the elections. More specifically, we have two types of expectations about such effects:

1. As the Pd and the Pdl have joined a parliamentary coalition, suspending their usual

reciprocal hostility, one should expect that the wall separating the electorate of the two

ideological blocks has been tore down, at least in part. Thus, the distinctiveness of the

two main parties is likely to have blurred, at least in the perceptions of their most

peripheral voters. This, in turn, may have increase voters openness to the appeals

coming from the opposite ideological side, increasing their electorate's availability

(Bartolini 1999). This is largely related to the fact that ideology might have been

4 Also Monti contributed to this by creating an own party and claiming to run against the bipolarism based on

left and right ideology. However, his party eventually got a little more than 10% of the valid votes.

8

losing its constraint power on voters’ preferences, even though the same voters are

still using the left-right continuum as representation of the political space (Baldassari

2013; Segatti 2013; Vegetti, Poletti and Segatti 2013).

2. As the two main parties have been undergoing a two-faced behaviour (supporting in

parliament the government and fighting in public on a daily basis), one should expect

that citizens' discouragement and disaffection about the political parties has increased,

as a reaction to the mismatch between what parties actually did in parliament and how

they behaved in front of the cameras. This outcome might have also been fuelled by

the blatant cases of corruption and abuse of public money over the period. In other

words the parties’ two-faced behaviour, far from being understood by the voters as a

way of keeping parties' reputation high, might have been interpreted as a strategy to

mislead the public from what the party elites were really doing.

Both expectations rest on the assumption that ideology has lost ground in constraining some

determinants of the voting calculus. To understand how this could have happened we refer

briefly to the literature on ideological voting.

3. Electoral Patterns of Competitions

In democracies, ideological vote is usually regarded as the most thorough, self-conscious and

rational type of vote choice. This is because ideology is often defined as a coherent value

system that provides people guidance in their search for a “just” solution on substantive

political issues. Thus, ideological vote is often assumed to be based on substantive policy

considerations, while at the same time being driven by internalized values that help people

organizing the complexity of the political reality. In trying to explain how voters process their

political decisions, scholars have been typically focused on characteristics of the individuals

related to their social groups (Campbell et al. 1960; Lipset and Rokkan 1967). In the last

decades, however, the process of electoral change in many Western democracies has seen a

decline of the role of traditional loyalties and stable social cleavages. One of the

consequences of these changes is that short-term political factors are more central now in

conditioning the determinants of voting choices, compared to the past (Thomassen 2005, Sani

and Segatti 2002, Bellucci and Segatti 2011).

When looking at the role that ideology plays for the voters, one important thing to take into

account is that people's ability to process and organize political information is limited.

Limited cognitive abilities coupled with large amount of information have the consequence of

making citizens more likely to use heuristics when deciding who to vote for, rather than

9

creating a more complex system of information and voting according to it (Chaiken 1980;

Fiske and Neuberg 1990). Previous studies have shown, however, that contextual

characteristics, such as types of party competition, might affect the complexity of this

information process. Patterns of competition might be altered by several factors such as party

fragmentation, degree of proportionality of a political system and polarization (Lachat 2011).

In this paper we consider the depolarization between the parties joining the grand coalition as

a potential factor to reduce the importance of left-right considerations for the voters. In

general, when elections are polarized it is more likely that voters engage in ideological voting

(Lachat 2008). If parties’ positions on specific issues differ more strongly, parties should

become eager to emphasize to the public where they stand on specific issues. In doing so,

issues acquire increasing salience and simplify for the citizens the task of making sense of the

large amount of political information required to make meaningful choices. This makes

ideological orientations more accessible and it becomes easier for citizens to learn how to use

them and how to translate them into votes (Lachat 2008; Knutsen and Kumlin 2005). To give

an example, Van der Eijk, Schmitt and Binder (2005) have shown how, in 21 national

elections from five European countries (Denmark, Germany, the Netherlands, Norway and

Sweden), the impact of left-right ideological orientations on citizens’ vote increased with

larger ideological distances between the parties. Moreover, Dahlberg (2009) demonstrated, in

a study of 35 elections in Germany, the Netherlands and Scandinavian countries, that voters

tend to agree more easily on where parties stand on the left-right in elections in which parties

diverge the most from each other. In these studies polarization was conceived as distance

between parties on the left-right continuum. Empirical estimation of party distance relied on

different types of measurement, which, however, have the common core in the perceived

judgement made by voters or by experts on where parties stand.

In this present study we are not able to provide this measurement for Italy in the period of

time that we analyse. However, we argue that the parliamentary support to the Monti

Government by the two main parties, from November 2011 to December 2012, has been

likely seen by the voters as a behavioural indicator of decreasing party polarization. This

expectation draws from literature on the impact of coalition governments on perceptions of

party ideologies, showing that members of the same coalitions tend to be seen as

ideologically more similar as they actually are (Fortunato and Stevenson 2013). To be sure,

resting only on this assumption does not allow us to categorically exclude that voters still

perceived the Pd and the Pdl as quite apart from each other. However, we show in a previous

study that in the months of the technocratic government, there has been indeed a decrease in

10

the correlation between people's ideological positions and their evaluations of the two parties

(Vegetti, Poletti and Segatti 2013). In other words, at least for one of the possible indicators,

the evidence seems to support our expectation of a depolarization of the party system.

Given this assumption, we formulate two expectations. First, during the time span in which

the Pd and the Pdl jointly supported the Monti Government, voters have been perceiving the

two most important parties of the two blocks, the Pd and the Pdl, as less distinct. Thus, in that

period, the level of reciprocal availability of their electorates, i.e. the degree to which the

voters of the Pd and the Pdl were open to the possibility to vote for the other party, should

have increased. Second, the continuous quarrelling among parties over any possible source of

debate may have led part of the public to think that parties were trying to fool the voters, by

secretly agreeing on the policies proposed by the government while displaying disagreement

on any possible matter at the same time in public. Thus, over the months prior to the 2013

election, the two-faced coalitional behaviour of the Pd and the Pdl might have actually

increased the disaffection felt by the citizens towards the political parties.

5. Assessing electoral availability using propensity to vote scores

We now offer an empirical exploration of the public's response to the political events taking

place between the summer of 2011 and the elections of February 2013 using data from a

repeated cross-sectional survey conducted by Ipsos5. Our sample consists of 49,901

respondents of a survey conducted every week on a fresh sample from March 2011 to

February 20136. The survey was conducted following the Computer Assisted Telephone

Interviewing (CATI) method, on a sample drawn by random digit dialling and corrected by

gender, age, region and size of municipality.

To measure party evaluations we rely on a type of rating scale known as “propensity to vote”

(PTV) scores. Similar to other variables created over the years to observe respondents'

evaluations of the major political parties, PTVs are ordered scales (in our case ranging from 1

to 10) where respondents are asked to say “how likely” it is that they will “ever” vote for

each party (see Tillie 1995 and van der Eijk et al. 2006 for a more focused discussion of the

psychological bases of PTVs and their empirical validation). In a broad sense, PTVs measure

the extent to which a respondent is considering voting for a party, without constraining this

5 Ipsos is a social research institute that collects citizens’ public opinion. Data have been bought by the

University of Milan thanks to a grant of the Fondazione Cariplo. The authors are grateful to prof. Paolo

Natale for his continuous support.

6 The value refers to the total number of cases having no missing values in any of the relevant variables. The

monthly sample size varies between a minimum of 927 in October 2011 to a maximum of 4237 in January

2013. Three months are missing from the time series: August 2011, January 2012 and August 2012.

11

consideration into a single ipsative choice, as it is the case for the more common “vote

choice” variable. To be sure, voting itself is an ipsative act, insomuch as it forces the voter to

choose for one and only one option. However, one of the aims of public opinion research is to

assess how characteristics of the external political context influence the opinions that

eventually shape individual behaviours. In this respect, PTVs are to be regarded as measures

of potential behaviour, as they capture the foundations of the choice by observing for each

individual which parties are excluded from the short-list (the ones that the respondent says

he/she would never vote for) and which parties are the real competing candidates running for

the final choice (the parties that receive a positive propensity to vote).

Because of this property, to observe a voter's set of PTVs is equivalent to measure his/her

degree of electoral availability, that is, the extent to which he/she is open to modify his/her

electoral choice (Bartolini 1999). If voters express a positive propensity to vote for one party

only, and no propensity to vote for all the others, it is likely that their choice is already made,

no matter what strategies parties will further adopt. Conversely, if voters give the same PTV

to all the parties, their choice is open to be influenced by a potentially large number of events

and last-minute considerations. In the first case, the probability that any voter will switch

between parties is essentially null, and thus party allegiances are going to be frozen, while in

the second case the potential for switching is virtually unlimited, and voting will resemble

something like a random choice7.

However, in the real world, the most of people's consideration sets should rather lie in

between these two extremes. For instance, voters can be similarly attracted by two parties,

and completely disinterested in all the others. In this situation, it is reasonable to assert that

the two parties are competing with each other for their votes. Thus, on the aggregate, this is

equivalent to say that patterns of covariation between PTVs provide a picture of who

competes with whom in the electoral arena at a given moment. If the PTVs of two parties

covary in opposite directions, it means that a positive evaluation of one party corresponds to a

negative evaluation of the other, and the other way round. In this situation, the two parties are

not competing for the same voters, as being attracted by one corresponds in the voters' mind

to being repelled by the other. In other words, the (potential) electorate of one party is

unavailable to the other. Vice versa, when the PTVs of two parties covary positively, a higher

likelihood to vote for one corresponds to a higher likelihood to vote for the other, and

7 A third hypothetical situation is the case in which voters give the lowest PTV to all parties. As we will see

later, we argue that respondents showing this type of pattern are to be regarded as complete outsiders of

party competition. In our argument, such a response style indicates a feeling of disaffection towards the

political system as a whole.

12

therefore their electorates are reciprocally available. In this case, the two parties must

compete with each other to win over the same voters.

The Italian election of 2013 has been characterized by the largest amount of vote switching in

the republican history (Itanes 2013). This could be due to the fact that the effects of the left-

right cleavage, which contributed a great deal in maintaining the voters' choice sets mostly

“frozen” into relatively stable ideological blocks during the Second Republic, have eventually

weakened. Although it has been shown that voters' movements between Pd and Pdl have

decreased from 2008 to 2013 (De Sio and Schadee, 2013), we expect that a weakening of the

left-right cleavage should be reflected, among other things, into a pattern of increasing

reciprocal availability between the electorates of the two most important parties of the two

blocks, the Pd and the Pdl..

Figure 1 shows something interesting in this regard, illustrating the monthly variation of the

polychoric correlation between the PTVs of the two parties from March 2011 to February

20138. The pattern shown in the figure is consistent with our expectation. Until

November/December 2012, the correlation between the PTVs of the Pd and the Pdl is

negative and significant. Substantively this means that, in the months within our time window

preceding the technocratic government, to have a higher propensity to vote for the Pd (Pdl)

corresponded to having a lower propensity to vote for the Pdl (Pd). This does not come as a

surprise for those who experienced the political mood of the Second Republic, made of

constant reciprocal accusations between party spokesmen in TV talk-shows and repeated

appeals to ideological labels as group flags. However, from December 2012, right after the

resignation of Silvio Berlusconi as Prime Minister and the handover to Mario Monti, the

correlation becomes increasingly weaker, reaching a level in Spring/Summer 2012 where it

looks not significantly different from zero. Normally, a zero correlation is interpreted as

“independence”, that is, knowing a person's position on one variable does not provide any

support for inferring his/her position on the other variable. In our case, this zero could be

interpreted literally, claiming that from April to September 2012 the Pdl and the Pd were

evaluated independently from each other9, or we could simply note that the predominance of

8 The polychoric correlation assumes two continuous, normally-distributed latent variables that are observed

on ordinal scales, and therefore it is the most appropriate technique to estimate the correlation between

variables such as the PTVs. Polychoric correlations are interpreted in the same way as the Pearson's r, and

like their better-known counterpart they range between -1, indicating perfect negative correlation, and +1,

indicating perfect positive correlation. The coefficients are computed via maximum likelihood estimation.

9 In the month of July 2012 the correlation seems to become slightly stronger just to drop again in September.

We believe that this effect is mainly due to an anomaly in the sampling for the month of July, probably due

to the fact that many potential respondents are on vacation. This suspicion is somewhat strengthened by the

presence of a similarly outlying observation in July 2011.

13

opposing feelings towards the two parties among the public has disappeared. This implies

that the electorates of the two main parties on the left and the right have been, for a few

months, “open” to the possibility to vote for the main opponent, although without necessarily

providing a positive evaluation of it.

FIGURE 1 ABOUT HERE

Figure 1 also shows that, from October 2012, the negative correlation between the PTVs of

the two parties gains new strength, reaching at the moment of the election in February 2013

the same level observed before the beginning of the technocratic government. This steady but

persistent return to a polarized situation, where the propensities to vote for the two parties are

essentially mutually exclusive, reflects a growing tendency of the Pd and the Pdl to “close

ranks” among their supporters. The turning point here is the beginning of the primary

campaign for the election of the candidate Prime Minister of the left-wing coalition. The

primary elections, which were announced earlier in the summer, had for the first few weeks

only one candidate running for the coalition leader Pd, the secretary of the Pd itself Pierluigi

Bersani. However, in mid-September Matteo Renzi, major of Florence, presents himself as an

alternative candidate pushing for an agenda heavily based on the renewal of the leading class

of the Pd. In the same days other two members of the coalition present their candidacy,

effectively starting the primary campaign. The primary elections, won by Bersani at the

second ballot, sign the beginning of the actual electoral campaign. On December 6th

, only a

few days after the official designation of Bersani as the candidate prime minister for the left-

wing coalition, former PM and leader of the Pdl Silvio Berlusconi announces that he will be

the candidate of the right-wing coalition. On the same day, the Pdl withdraws its support from

the Monti government, ending the temporary compromise with the Pd. From this point, the

electoral campaign carries on with the same hostile tones that the voters were used to before

the Pd and the Pdl were jointly supporting the Monti government (a period during which

reciprocal accusations among the two parties were still present, but the tones were somewhat

softened).

Figure 1 succeeds in showing one important pattern: the joint support by the two parties to

the Monti cabinet seems to have de facto depolarized the attitudes of the voters towards the

two parties. In other words, for a few months in 2012, the electorates of each of the two

parties were more available to the appeals of the other. Thus, our data seem to confirm that

the left-right cleavage has for a while weakened. This could have happened regardless of

14

whether voters might have still perceived the two parties quite far apart from each other. Yet,

the combination of signs of behavioural convergence between the two parties on the one

hand, and of quarrelling divergence on the other, may have generated also a wave of

disaffection.

Greater electoral availability or more disaffection?

The pattern observed so far, i.e. the simple correlation between the PTVs of the Pd and the

Pdl, can reflect two qualitatively different phenomena. The most obvious one, following the

rationale discussed earlier, is the variation of the competition between the two parties by

means of an increased reciprocal availability of their respective electorates. This part of the

story refers to the propensities to vote for the Pd and the Pdl becoming less (or more)

mutually exclusive, and thus the two parties becoming more (or less) appealing to the same

electorate. However, a second phenomenon that can be captured by the varying correlation

between the two PTVs is the variation over time of the proportion of respondents who state

their unwillingness to vote for any party in the system.

As we discussed above, PTVs are ordinal scales meant to allow respondents to give nuanced

answers regarding their willingness to vote for a party. Such scales are generally constructed

with two anchors, one on the lowest value (1 in our data) and one on the highest value (10).

The first proposition states that the respondent is “not at all likely” to vote for the party for

which the PTV is asked, while the second states that it is “very likely”. Thus, while a PTV of

10 still consists in a probabilistic statement, as the final choice will also depend on how the

other parties are evaluated, the lowest PTV reports an assertion of certainty, i.e. that a given

party is excluded from the choice process10

. This implies that a PTV of 1 is generated by a

different process than a PTV with a higher value, with the second process being conditional

on the first: when respondents are asked about their propensity to vote for a party, they first

decide whether or not they would take the party into consideration, and in case of a positive

response, they assess their degree of likelihood11

.

10 As we will see later, a way to assess certainty regarding the party choice using PTVs requires measuring the

distance between the highest and the second-highest PTVs. In this sense, the maximum certainty can be

achieved only when the respondent gives a PTV of 10 to one party and a PTV of 1 to all the others.

11 This way to conceive vote choice as essentially a two-step process has been explored in the past (see e.g.

Wilson 2008), although it has never been applied to the analysis of PTVs. In fact, the scholars who first

introduced the use of PTVs noted their multi-modal distribution, but they rather kept the focus on the fact

that all the categories are used, to justify the continuous nature of party evaluations (see van der Eijk et al.

2006). While the conceptualization that we propose here is by no means equivalent to assert that party

evaluations are dichotomous, we argue that a conceptualization of PTVs as mixtures of two concatenated

evaluations fits better both the format of the question and the empirical distribution of the PTVs themselves.

15

FIGURE 2 ABOUT HERE

The double process is reflected quite well in the way PTVs are distributed. To make an

example, Figure 2 shows a simple histogram of the PTV for the Pd in our sample. The choice

of the party is purely arbitrary, as all PTVs follow a very similar distribution. The figure

clearly shows the presence of a large mode on 1 next to a bell-shaped distribution over the

other values, which resembles a normal distribution (given the type of scale). To be sure, this

distribution might simply be censored, and thus piling up the observations below a certain

limit on a value around the limit itself. In this view, propensity to vote should rather be

conceived as a latent variable that is observed only within a certain range of PTV values.

However, given the way the scale is proposed to the respondent, it seems more plausible that

the lowest value captures an attitude that is separate from the rest of the scale, i.e. the

complete refusal to consider voting for a party.

Given this conceptualization, we expect a portion of respondents to give the lowest values to

all the parties. These are the citizens who do not feel attracted by any of the relevant options,

and thus refuse to even consider voting for them. These individuals are by all means

unavailable to any party, and thus completely out of party competition, since no matter how

parties change their appeals, they will simply ignore them. Because of their tendency to give

the same (lowest) PTV to all parties, these respondents show a perfect positive correlation

between the PTVs of the Pd and the Pdl. However, this is not due to the fact that they are

available to both parties, but rather that they are completely impenetrable by their appeals.

We define these respondents as disaffected from the traditional political parties, and we

categorize them using a dummy that has value of 1 if they give the lowest to all the main

parties, and 0 otherwise. In the calculation of this variable, we consider the PTV of eight

parties: Pd, Pdl, Northern League, Italy of Values (and its pre-electoral merge with other left-

wing parties, Civil Revolution), Udc, Fli, Sel and Monti's Civic Choice party. We exclude

from the calculation the PTV for the Five Star Movement (M5s). This is due both to a

theoretical and a practical reason. First, our variable measures the disaffection from the

traditional parties, i.e. the parties who describe themselves as insiders of the political system,

of which the Pd and the Pdl are the two main poles, while the M5s presents itself as

alternative to the political system. This difference is not trivial, as it implies that the space of

competition (see Sani and Sartori 1983) where the M5s seeks for the votes is essentially

different from the one where the other parties act. In fact, since its first days as a grassroots

16

movement in mid 2000s, the M5s built an image aimed to expressively capture and channel

the sentiment of disaffection, and even disgust, of the citizens towards the traditional parties,

earning the label in the media of anti-political movement. Thus, to include the PTV of the

M5s in our calculation would have excluded those respondents who give a higher PTV only

to the M5s, i.e. the voters who are already out of the space of competition where the Pd and

the Pdl contend the votes. A second and more pragmatic reason to exclude the PTV of the

M5s is that, because of the relatively sudden growth of the party, the variable is present in our

data only from June 2012. Given this constraint, including the M5s in our calculation would

make the cases classified as disaffected before and after that month not equivalent, biasing

the reliability of our operationalization.

As we mentioned earlier, the observed pattern of correlation shown in Figure 1 can be due to

a variation over time of the feeling of disaffection among the citizens. In fact, the contribution

that these voters give to the correlation between the PTVs of the Pd and the Pdl is always

positive: for them, the correlation will always be +1. To give an example, if in the months

between March 2011 and February 2013 the proportion of citizens belonging to this category

increased, the negative correlation between the two PTVs would look increasingly weaker.

However, this would not be due to a growing reciprocal availability of the two electorates,

but rather to a growing tendency among the population to refuse being available at all. Hence,

to observe how did the events occurred from 2011 up to the election date really impact on the

public opinion, it is important to isolate disaffected voters, and observe how their incidence

on the sample changed over time.

FIGURE 3 ABOUT HERE

By relying on our measure, the citizens disaffected by the traditional parties represent about

10% of our sample, and vary considerably across months, ranging from a minimum of about

6% (March 2011) to a maximum of 16% (May 2012). Figure 3 shows the time series. The

trend resembles the one in Figure 1, with some important differences. First of all, while

Figure 1 clearly showed the beginning of the decline of the correlation in December 2011,

here the bump starts rather in March 2012. Second, Figure 3 shows a sudden drop of

disaffected respondents in the months of June, July and September 2012. We attribute this

drop at least in part to the quality of the sampling used to collect our data. Despite this, Figure

3 shows a pattern of increase and subsequent decrease that follows the political events

occurred those months and discussed earlier. Thus, it seems that during the technocratic

17

government there has been a significant increase in citizens' disaffection from the traditional

parties, and that Pdl's withdrawal from the majority, coincident to the beginning of the

electoral campaign, had a negative impact on it. However, one thing worth noticing is that, at

the end of our time series, the overall level of disaffection is significantly higher than at the

beginning. This marks an important difference from the trend in Figure 1: while there the

level of correlation in February 2013 was back on the same level it had before the beginning

of the technocratic government, here the level of disaffection increase of almost five

percentage points.

Once the correlation is clean from the spurious effect given by the disaffected respondents,

the pattern looks rather different. Figure 4 shows the same correlations computed for Figure

1, but applied only to the sub-sample of respondents who are open to party competition, i.e.

those who gave a positive PTV to at least one party. The picture makes three main points.

First, all the series is shifted downwards, increasing in negative strength of 0.1 points (on a

scale from -1 to 1), and it is always significantly different from zero. This means that, once a

very specific group representing about 10% of the population is excluded from our

observation, the correlation between the PTVs of Pd and Pdl is always and inevitably

negative. Second, the range of the monthly variation reduces considerably, going from 0.4

points of Figure 1 to 0.25 points. In other words, for this population, the parliamentary

support to the technocratic government by the Pd and the Pdl seem to have had a relatively

smaller impact. Finally, at the moment of the elections in February 2013, the negative

correlation between the two PTVs has reached again the levels it had before November 2011.

Hence, the ability of the grand coalition to depolarize the electorate, and thus increase the

reciprocal electoral availability of the electorates of the two main parties of the left and the

right blocks, has been limited to the period in which the grand coalition lasted. Once the

electoral campaign started, the reciprocal availability between the electorates of the Pd and

the Pdl went back to the (low) levels that it used to have.

FIGURE 4 ABOUT HERE

Together, figures 3 and 4 effectively decompose the phenomenon observed in Figure 1, that

is, the significant reduction of the negative correlation between the propensities to vote for Pd

and the Pdl, and thus the apparent increase of the competition between the two main parties

of the left and the right blocks. Our data show that there has been indeed a moment where the

evaluations of Pd were less negatively associated with the ones of the Pdl, which lasted for

18

the most of 2012. However, the extent of this phenomenon has been rather limited, and its

occurrence only contingent to the presence of the grand coalition. At the same time, another

phenomenon, much less desirable from a normative perspective, affected the electorate to a

similar extent. Next to making the electorate less shy to cross the border between left and

right, the behaviour of the elites also contributed to an increase of the amount of citizens who

are disaffected from the traditional parties, a tendency that persisted for a portion of citizens

even after the end of the grand coalition.

The co-presence of these two types of citizens is inevitable in democratic systems. People

have indeed the right not to be interested in politics, and sometimes they may also have the

reasons to be frustrated by it. However, if the political elites, with their behaviour, can

contribute to citizens' political affection, then they should do something in order to increase

it, rather than to decrease it. Instead, what we see in our data is that, at least up to the election

of February 2013, the experience of the contentious cohabitation of the Pd and the Pdl in the

same coalition has undermined the appeal of the political parties for a considerable portion of

the population. These are the people that the Five Star Movement tried to mobilize (with

more or less success, see Itanes 2013) and that, within the last two years, seem to have been

growing in number.

Yet, our evidence suggests that for the remaining (and relatively larger) part of the electorate

there has effectively been a slight but substantive depolarization between the main left-wing

and the main right-wing party. This confirms the mechanism discussed elsewhere (see

Vegetti, Poletti and Segatti 2013), according to which the way the main parties decided to

support the Monti government has reduced the importance played by ideological differences

for the voters' judgments. In this view, the coalition between Pd and Pdl has reduced the

perceived ideological distance between the two parties (Fortunato and Stevenson 2012), in

fact depolarizing the party system (Lachat 2008).

8. Determinants of disaffection and availability at the individual level

What are the factors that influence the two phenomena that we just discussed? It is interesting

to understand what the characteristics that link voters to the disaffection for the political

parties on the one hand, and that makes them regard the Pd and the Pdl more or less equally

appealing on the other, are. Moreover, by controlling for individual characteristics, we can

clean our picture from effects given by the sample composition of each month, and make sure

that during the months of the Pd/Pdl coalition the voters have been significantly different in

their probability to be disaffected by the parties and in their joint evaluations of the Pd and

19

the Pdl. Thus, to perform a last investigation, we model these two phenomena at the

individual level using multilevel regression analysis.

The two dependent variables are straightforward. First, the disaffection by the traditional

parties is observed at the individual level by means of the dummy variable discussed before.

As a reminder, our DV here measures 1 when the respondents give the lowest PTV to each

and every party (excluding the M5s) and 0 otherwise. Second, we observe the different in

judgment between the Pd and the Pdl (and therefore their mutual exclusiveness) by taking the

absolute difference between their PTVs. This measure, that we call “PTV certainty”, has a

higher value when the Pd and the Pdl are given two very different PTVs, and a lower value

when they are evaluated similarly, ranging from 0 (PTV Pd = PTV Pdl) to 9 (one PTV is 1

and the other PTV is 10). The term “certainty” is justified by the fact that the greater the

absolute difference between the two PTVs, the more a voter's choice is predetermined, and

thus the smaller the chance that he/she will be convinced by the other party. Given the

“spurious” influence on the distribution of PTV certainty of disaffected people (i.e.

disaffected respondents give the lowest value to all PTVs, hence for them the difference

between the PTV of the Pd and the one of the Pdl will always be zero), the second model will

be estimated on the sub-sample of respondents who are open to party competition, the same

observed in Figure 4.

The social-structural predictors that we include in the model are the respondent's age

(measured in years, centred around the sample mean), the gender (a dummy where 1 =

female; 0 = male), the level of education (an ordinal variable with five ascending categories

going from low to high education, centred around the median), the degree of attendance to

the religious services (an ordinal variable with four ascending categories going from “never”

to “weekly attendance”, centred around the median), and the geo-political area of residence.

The latter is divided in five categories: North-West (used here as reference category, including

the regions of Piedmont, Aosta Valley, Lombardy, and Liguria), North-East (the so called

“white area”, including Trentino-Alto Adige/South Tyrol, Veneto and Friuli-Venezia Giulia),

Centre-North (the so called “red belt area”, including Emilia-Romagna, Tuscany, Umbria and

the Marches), Centre-South (Lazio, Abruzzi, Molise and Sardinia), and the South (Campania,

Apulia, Basilicata, Calabria, Sicily). These variables help us generate a socio-demographic

“profile” of the disaffected citizens, and of those who are more or less open towards the

competition between the Pd and the Pdl.

Other three individual-level variables that are included in the model are three dummies

indicating the ideological orientation, or self-definition, of the respondents: one for left-wing

20

voters, one for right-wing voters, and one for the voters who refuse to position themselves on

the left-right12

. These variables are useful to see whether and how disaffection and

availability to switch between blocks have affected people from different ideological

identities (including those who refuse to have such an identity at all) in different ways.

Finally, the model predicting the PTV difference between Pd and Pdl requires one additional

control variable, that is the magnitude of the highest PTV between the two considered13

.

Finally, we add to the model two macro predictors observed at the month level: a dummy

indicating the months where the Pdl and the Pd were part of the parliamentary grand coalition

jointly supporting the technocratic government14

, and the passage of time, assessed by a

progressive number associated to each month, going from 1 in March 2011 to 21 in February

2013. Both variables are rather important for our argument, as we contend that the

parliamentary coalition between the two parties is the main responsible for the loosening of

the psychological bound between ideological blocks. However, while the first variable is

meant to capture whether this effect was in place during the grand coalition, the second will

tell us whether it lasted even after the end of the agreement. Thus, to observe a significant

coefficient for these two variables while controlling for individual characteristics would

essentially prove that the coalition had an impact on the public's evaluations.

Given the hierarchical structure of our data, with individuals nested within months, we model

our dependent variables in a multilevel setting. Multilevel modelling allows to set the effect

of some variables as fixed, i.e. constant among the time points of our series, and of other

variable as random, i.e. free to vary across months. In our case, we specify a simple random

intercept model, hence controlling for each month's specific sample effect on our dependent

variables, without having this effect absorbed by other predictors, and at the same time

accounting for the (relative) non-independence between observations belonging to the same

month. Because disaffection is operationalized as a dummy variable, we model it assuming a

12 The reference category in this case consists in the voters who position themselves on the center of the left-

right scale.

13 This control is necessary as our dependent variable is related to it in a mechanical way. In fact, the

difference between the two PTVs has a theoretical maximum that depends inevitably on the range of the

largest one. Let's take as example a supporter of the Pdl who gives it a PTV of 9. In the case of this

respondent, the difference between the two PTVs can range from 0, in case he/she gives a PTV of 9 to the Pd

as well, to 8, in case he/she gives the Pd a PTV of 1. However, if another respondent, still supporter of the

Pdl, gives it a PTV of 4, then the difference can be at most of 3 points, because it is not possible to give the

Pd a PTV smaller than 1. This issue with the measurement creates artificial heterogeneity in the

measurement, which requires to be controlled for by adding among the predictors the value of the largest

PTV (among the two considered here). The effect of this variable is expected to be always positive and

highly significant, although, as we said, its influence is purely mechanical, and thus not interesting for

substantive reasons

14 The dummy has value 1 for the months from November 2011 to November 2012 included, and 0 for the

months before and after this period.

21

binomial distribution with a logit link function. For the PTV difference between Pd and Pdl

we rely on a more common linear modelling assuming a Gaussian distribution with identity

link function. The models are estimated via restricted maximum likelihood.

TABLE 1 ABOUT HERE

Table 1 shows the results of the two models. Despite being in the same range, the magnitudes

of the two sets of coefficients are not comparable to each other, as they represent in one case

variations of the linear predictor (for the logit model of disaffection) and in the other variation

of the actual dependent variable (for the linear model of PTV difference). However, we can

analyse the direction and the statistical significance of the effects in each model, in order to

assess what characterizes disaffected citizens and what determines people's availability

among the Pd and the Pdl.

Focusing on the left panel, i.e. the coefficients for disaffection, we note that more educated

people, and people who more frequently attend religious services, are significantly less likely

to feel disaffected from the traditional political parties. This makes sense if we consider that

disaffected citizens are expected to be generally more socially marginal than people who are

positively engaged, and both education and church attendance are indicator of social

integration. Moreover, all our indicators related to ideological self-labelling are significantly

associated to disaffection. First of all, both being positioned on the left and on the right is

associated with a lower probability to be disaffected, with a stronger effect for left-wing

voters than for right-wing voters. In other words, citizens who state their own ideological

identity to be either left-wing or right-wing are more likely than those on the centre to be

attracted at least by one among the (relevant) competing parties. This finding suggests a

certain degree of alienation of centre voters from the political supply, that nicely confirms

what Itanes scholars (2013) have found on the basis of a different dataset. Implicitly this is

also a confirmation of the bipolar nature of Italian political divisions as well as of the

weakening of the left-right ideology on the centrist voters. Second, people who refuse to

admit any ideological affiliation by not positioning themselves on the left-right are more

likely to be also disaffected to a rather large extent. This association in a way resembles the

one observed for left-wing and right-wing voters in respect to centre voters, namely that

ideological self-identifications in Italy are (still) strongly associated to party evaluations, and

thus refusing to be associated to such labels comes together rather frequently with the

tendency not to be attracted by any of the relevant party options.

22

Finally, our macro-level predictors, i.e. the presence of the grand coalition and the passage of

time, have both a strong positive and significant effect. This implies, first, that in the months

where the Pd and the Pdl joined a parliamentary coalition citizens' disaffection, on average,

increased; and second, that this tendency remained also in the following three months, during

the electoral campaign before the elections. This confirms what observed in Figure 3, and

holds even controlling for other potential factors of influence. Thus, we conclude that one of

the effects of the behaviour during the technocratic government of the political elites, of

which the Pd and the Pdl represent the main actors, had the effect of increasing citizens'

disaffection from the traditional political parties in an enduring way.

Moving to the second model, some effects change substantively, while others maintain a

similar profile. First of all, we note that older citizens are more “certain” about their choice

between the Pd and the Pdl, that is to say, they are less likely to switch between them. This

finding is rather intuitive, as older citizens are more likely to have developed a voting habit,

and thus to be relatively harder to be influenced by different party appeals. Moving on, we

find that education has the same effect of age in making up voters' minds about their party

preference, while the coefficient of church attendance goes in the opposite direction. In other

words, our data show that more religious people are more open to switch between the Pd and

the Pdl, holding everything else constant, while more educated people are more likely to take

a side. Finally, we find a significant tendency to be more certain among the Pd and the Pdl in

the Centre-North, and to be less certain in the South. Both these findings make sense if we

keep in mind that the “red regions” of the Centre-North are possibly the last territories that

are clearly identified with a partisan affiliation (for the Pd, or in general for left-wing parties),

while the South has lately given several signals of a generalized refusal of the two main

parties (see for instance the Sicilian regional elections of October 2012). The coefficients of

the three categories of ideological self-identification are all positive and significant,

suggesting that both left-wing and right-wing citizens, and those who do not identify

themselves using ideological categories, evaluate the Pd and the Pdl more differently than

voters at the centre. Here, again, the coefficient associated with left-wing voters is much

stronger than the other two groups, indicating that people who place themselves on the left

are more certain about their party preference (presumably for the Pd). Interestingly, those

who refuse to place themselves on the left-right tend to be more certain about their

evaluations among the Pd and the Pdl. This is somewhat counter-intuitive, as one would

expect that people who refuse ideological labels are also more likely to regard the two parties

in a similar way.

23

As for the case of disaffection, the macro-level indicators confirm what observed in the

figures. First, the Pd/Pdl coalition in support to the Monti government has a negative and

significant, although not particularly large in magnitude, effect on the PTV certainty,

indicating that during the months of the grand coalition there has been in fact a slight

depolarization among the two electorates. Second, the effect of time is not distinguishable

from zero, i.e. there is no trend of growing reciprocal availability between the supporters of

the Pd and the Pdl. In other words, after the end of the grand coalition and the beginning of

the electoral campaign, the degree of mutual-exclusiveness between the evaluations of the

two parties went back to the levels that it had before the events of 2011 forced the formation

of the technocratic cabinet.

Together with the substantively strong and statistically significant effect of the same variable

on disaffection, this result offers a rather negative picture of the long-term impact that the

parliamentary coalition in support of the technocratic government had on the citizens.

Essentially, our data show that the effect that the joint but contentious support to the Monti

Government had on voters' perceptions of the political space is an increased disaffection from

the political parties. In other words, rather than opening a “breach” between left and right

blocks, hence reducing the polarization of the electorate along the ideological cleavage, the

behaviour of the elites during the Monti government has indeed increased voters' disaffection

from the parties.

9. Discussion and conclusions

This paper investigates the electoral implications of the two-faced behaviour of the two major

Italian parties, the left-wing Democratic Party (Pd) and the right-wing People of Freedom

(Pdl), in occasion of the grand coalition in support of the Monti government. While a grand

coalition government seemed to be the most effective solution to reassure the financial

markets during the sovereign debt crisis in autumn 2011, we argue that the way in which

parties dealt with this event had unexpected consequences for the citizens. In fact, differently

from other examples of grand coalition government (Miller and Müller 2011), the Italian

parties refused to take full responsibility of their behaviour in Parliament. First, instead of

initiating a proper grand-coalition government, they decided to make a parliamentary

coalition supporting a government of technocrats. Second, instead of making any type of long

and detailed negotiations regarding the reforms to be undertaken and presenting them to their

electorates, the parties engaged essentially in a two-faced behaviour, where a wide agreement

in supporting the reforms designed by the government in Parliament went hand-in-hand with

24

a constant scuffling with each other in front of the public. This behaviour coexisted with a

pronounced degree of internal division among representatives of both parties, which made it

even harder to assess their positions on a number of important issues.

In this study we examine the impact of these behaviours on the public opinion, focusing in

particular on the degree of electoral availability, i.e. the extent to which voters are open to

switch their choice between parties in different ideological blocks (see Bartolini 1999). Italian

politics has been historically frozen by a deep left-right cleavage (Bellucci and Segatti 2011).

Given the exceptional nature of the “strana maggioranza” lasting from the end of 2011 to the

elections in 2013, one possible consequence of this could have been a substantive reduction

of the strength of this cleavage. This expectation is suggested by literature arguing that

coalition partners tend to be perceived as more ideologically similar (Fortunato and

Stevenson 2013) and by a study showing that the strength of the correlation between

ideological self-identifications and party preferences for the Pd and the Pdl has, in fact,

weakened in the period examined (Vegetti, Poletti and Segatti 2013). However, rather than

relying on ideological perceptions of party positions, this present study asked whether the

grand coalition in support of the technocratic government opened a breach in the wall

separating the two ideological blocks, establishing a pattern of direct competition between

left-wing and right-wing parties. In other words, we were seeking for a potentially positive

outcome of the recent political events: an improvement of the competitiveness of Italian

politics through an increase of voters' openness to consider appeals coming from parties

belonging to different ideological blocks.

In fact, the reality that our data show regarding the effect that the elites' behaviour had on the

voters is rather discouraging. By observing monthly changes of party evaluations on a sample

of the public opinion we find that, first, there has indeed been a slight increase of the

reciprocal availability between the electorates of the Pd and the Pdl, although this effect

vanished as soon as the Pdl withdrew its support from the technocratic government and

started the electoral campaign. On the other hand, our second finding shows that the months

of the technocratic government also brought a load of disaffection from the traditional parties

among the public. Moreover, in this case the growing trend continued even after the

beginning of the electoral campaign, suggesting a long-lasting (negative) impact on the

voters' evaluations. In other words, while an increased tendency of the voters to consider the

main left-wing and right-wing parties as similarly attractive has lasted only for the time of the

parties' temporary armistice, the tendency to feel disaffected from the traditional parties

remained to a certain extent even after that, and lasted at least until the election in February

25

2013. While the disaffected public has been mostly targeted by the mobilizing attempts of the

Five Star Movement (M5s), a new party born as an anti-establishment protest movement, its

growing importance implies essentially a failure of the political system in engaging the

citizens, and therefore comes as a rather bad news from the normative point of view. This

becomes even worse when we consider that such an outcome is to be attributed, at least in

part, to the behaviour of the traditional party elites.

In spite of the relatively limited scope of our data, given by the lack of some important

variables (e.g. the respondents' perceptions of party ideological positions), these findings

offer an important contribution to our understanding of the premises of the “political

earthquake” of the election of 2013. In particular, what our results suggest is that such turmoil

may have been produced by the choices of the parties themselves, i.e. their inability to cope

with the brand-new experience of a grand coalition brought about by necessity in a moment

of economic emergency.

Hard times can call for political responsibility. Especially in the autumn of 2011, the

economic hard times were asking to Italian parties to stop fighting with each other, and

provide a demonstration of political stability. However, the way in which parties chose to

actually provide such a demonstration ended up being effective only in part. If a government

of technocrats led by a respected and internationally-recognized character such as Mario

Monti succeeded in calming down the attacks of the financial markets, the same attempt was

not perceived by the voters as a final loosening of the long-lasting conflictuality between the

Pd and the Pdl. On the contrary, the two-faced behaviour that the two parties kept during

those months was perceived by many citizens as an attempt to fool the electorate: while

parties were being friends inside the institutional buildings on the one hand, they behaved

like foes in front of the citizens on the other. It does not come as a surprise that the slogan

“sono tutti uguali” (all parties are the same), largely promoted by the leader of the M5s

Beppe Grillo, became very popular among so many citizens. All in all, the grand coalition

that supported the Monti government was perceived by many as just a further example of the

“proverbial craftsmanship” of the Italian political class. On the one hand, fearing the

reactions of the voters given the harsh economic situation, they escaped from taking full

governmental responsibility. One the other hand, they accepted to join a parliamentary grand

coalition, thinking that their voters had an adamant confidence in them. According to the

result of the election of February 2013, they proved wrong.

An Italian reader might comment that the story we tell in this paper is trivial. The Italian

party system is completely different from the German one, both in terms of type of

26

competition and centrality of particular actors (we refer here to the twenty-year protagonism

of the Pdl leader Silvio Berlusconi). We agree: Italy is not Germany. Yet, what we argue here

is that Italian voters, as citizens and individuals, are not different from the German ones.

They both make up their mind when they have to vote on the basis of similar types of calculi.

We already pointed out that the electoral consequences of German grand coalitions looked

similar to the ones of the 2013 Italian election (i.e. negative for the coalition partners), except

for the magnitude, incredibly high, of the latter. We claim that this might be caused by the

two-faced behavior of the two main Italian parties. Thus, at the end, the decision made by Pdl

and Pd to build in November 2011 only a parliamentary coalition proved to be less

sustainable in front of their voters than to build a governing grand coalition.

27

References

Baldassarri, D. 2013. Sinistra e destra: un’Italia di moderati e conservatori, in: ITANES,

2013. Voto amaro. Disincanto e crisi economica nelle elezioni del 2013. Il Mulino, Bologna,

Italy, pp. 133-146.

Banaszak, L.A., Doerschler, P., 2012. Coalition type and voter support for parties: Grand

coalitions in German elections. Electoral Studies 31, 46–59.

Bartolini, S., 1999. Collusion, Competition and Democracy: Part I. Journal of Theoretical

Politics 11, 435–470.

Bellucci, P., Segatti, P. (Eds.), 2011. Votare in Italia: 1968-2008. Dall’appartenenza alla

scelta. Il Mulino, Bologna, Italy.

Bellucci, P. Segatti, P. 2013. Introduzione, in: ITANES, 2013. Voto amaro. Disincanto e crisi

economica nelle elezioni del 2013. Il Mulino, Bologna, Italy, pp. 7–16.

Bersani, P. L., 2011. Low chamber speech during the confidence vote to the Monti

Government, November 18th

2011. Retrieved from http://www.camera.it, 20/11/2013.

Campbell, A., Converse, P., Miller, W., Stokes, D., 1960. The American Voter. Wiley, New

York.

Chaiken, S., 1980. Heuristic versus systematic information processing and the use of source

versus message cues in persuasion. Journal of Personality and Social Psychology 39, 752–

766.

Dahlberg, S., 2009. Political parties and perceptual agreement: The influence of party related

factors on voters’ perceptions in proportional electoral systems. Electoral Studies 28, 270–

278.

De Sio, L., Schadee, H.M., 2013. I flussi di voto e lo spazio politico, in: ITANES, 2013. Voto

amaro. Disincanto e crisi economica nelle elezioni del 2013. Il Mulino, Bologna, Italy, pp.

45–56.

Diamanti, I. (2013). Destra e Sinistra Perdono il Proprio Popolo. M5S come la vecchia DC:

Interclassista. La Repubblica.

Fiske, S.T., Neuberg, S.L., 1990. A continuum of impression formation from category-based

to individuating processes: Influences of information and motivation on attention and

interpretation. Advances in Experimental Social Psychology 23, 1–74.

Fortunato, D., Stevenson, R.T., 2013. Perceptions of Partisan Ideologies: The Effect of

Coalition Participation. American Journal of Political Science 57, 459–477.

Hobolt, S.B., Karp, J.A., 2010. Voters and coalition governments. Electoral Studies 29, 299–

307.

28

ITANES, 2013. Voto amaro. Disincanto e crisi economica nelle elezioni del 2013. Il Mulino,

Bologna, Italy.

Knutsen, O., Kumlin, S., 2005. Value Orientations and Party Choice, in: Thomassen, J. (Ed.),

The European Voter. A Comparative Study of Modern Democracies. Oxford University Press,

Oxford.

Lachat, R., 2008. The impact of party polarization on ideological voting. Electoral Studies 27,

687–698.

Lachat, R., 2011. Electoral competitiveness and issue voting. Political Behavior 33, 645–663.

Lipset, S.M., Rokkan, S., 1967. Party systems and voter alignments. Free Press, New york.

Maravall, J.M., 2010. Accountability in Coalition Governments. Annual Review of Political

Science 13, 81–100.

Miller, B., Müller, W.C., 2010. Managing Grand Coalitions: Germany 2005-09. German

Politics 19, 332–352.

Passarelli, G., Tuorto D., 2013. Berlusconi tra elettori fedeli e defezioni, in: ITANES, 2013.

Voto amaro. Disincanto e crisi economica nelle elezioni del 2013. Il Mulino, Bologna, Italy,

pp. 71–82.

Sani, G., Sartori, G., 1983. Polarization, Fragmentation and Competition in Western

Democracies, in: Daadler, H., Mair, P. (Eds.), Western European Party Systems: Continuity

and Change. Sage, London, pp. 307–340.

Sani, G., Segatti, P. 2002. Fratture sociali, orientamenti politici e voto: ieri e oggi, in:

Bartolini, S. and D'Alimonte, R. (Eds.), Maggioritario Finalmente?La Transizione Elettorale

1994-2001. Il Mulino, Bologna, Italy, pp.249-282.

Scarrow, S.E., 2012. The German grand coalition of 2005–09 and party system change:

Catalyst or continuity? Electoral Studies 31, 60–71.

Segatti, P., 2013. Italia - Elezioni politiche 2013: un altro ciclo è finito. Analisi dei risultati

del voto del 24-25/2. Il Regno 8, 195–198.

Thomassen, J., 2005. The European Voter. A Comparative Study of Modern Democracies.

Oxford University Press, Oxford.

Tillie, J., 1995. Party Utility and Voting Behavior. Het Spinhuis, Amsterdam.

Van der Eijk, C., Schmitt, H., Binder, T., 2005. Left-Right Orientation and Party Choice, in:

Thomassen, J. (Ed.), The European Voter. A Comparative Study of Modern Democracies.

Oxford University Press, Oxford, pp. 167–191.

Van der Eijk, C., van der Brug, W., Kroh, M., Franklin, M., 2006. Rethinking the dependent

variable in voting behavior: On the measurement and analysis of electoral utilities. Electoral

29

Studies 25, 424–447.

Vegetti, F., Poletti, M., Segatti, P., 2013. When responsibility is blurred: Italian national

elections in times of economic crisis, technocratic government, and ever-growing populism.

Rivista Italiana di Scienza Politica, in press.

Vezzoni, C., 2013. Perché il Pd ha perso le elezioni, in: ITANES, 2013. Voto amaro.

Disincanto e crisi economica nelle elezioni del 2013. Il Mulino, Bologna, Italy, pp. 83–92.

Wilson, C.J., 2008. Consideration Sets and Political Choices: A Heterogeneous Model of Vote

Choice and Sub-national Party Strength. Political Behavior 30, 161–183.

About the authors:

Federico Vegetti is a Post-Doc Researcher at the Central European University in Budapest,

Hungary. His research interests include comparative political behaviour, political psychology,

public opinion formation, and quantitative methods for social research.

Monica Poletti is a Post-Doc Researcher at the Università degli Studi di Milano, Italy and a

Visiting Fellow at the London School of Economics, London, UK. Her research interests

include comparative political behaviour, media and public opinion analysis, Euroscepticism,

political culture and quantitative methods for social research.

Paolo Segatti is Professor of Political Sociology at the Università degli Studi di Milano,

Italy. His research interests include comparative political behaviour, public opinion analysis,

and National and European identities. He is part of the Italian Election Studies (Itanes).

30

Figure 1: Correlation between propensity to vote for the PD and the PDL over time. Source:

Own elaboration of IPSOS data.

31

Figure 2: Typical distribution of PTV variable. Source:

Own elaboration of IPSOS data.

32

Figure 3: Disaffection from the traditional political parties over time. Source: Own

elaboration of IPSOS data.

33

Figure 4: Correlation between propensity to vote for the PD and the PDL over time, for the

voters who are open to party competition. Source: Own elaboration of IPSOS data.

34

Dependent Variable:

Disaffection PTV Certainty PD/PDL

Logit OLS

Max PTV (PD & PDL) 0.881*** (0.004)

Age -0.001 (0.001) 0.009*** (0.001)

Gender (Female) 0.023 (0.033) -0.021 (0.017)

Education -0.047*** (0.013) 0.121*** (0.007)

Church Attendance -0.054*** (0.014) -0.153*** (0.008)

North-East -0.048 (0.056) 0.009 (0.030)

Center-North -0.050 (0.050) 0.057* (0.026)

Center-South 0.086 (0.045) -0.047 (0.025)

South 0.047 (0.044) -0.062* (0.025)

Left -0.783*** (0.057) 1.160*** (0.029)

Right -0.262*** (0.058) 0.315*** (0.030)

Not Positioned on L-R 1.587*** (0.053) 0.155*** (0.036)

Joint Support Government 0.330*** (0.077) -0.106* (0.050)

Time 0.033*** (0.006) 0.007 (0.004)

Intercept -2.906*** (0.098) 2.912*** (0.058)

Var (Intercept) 0.0216 0.0098

Observations 49901 43694

Groups 21 21

Log Likelihood -14378 -87416

Note: *p<0.05; **p<0.01; ***p<0.001

Table 1: Multilevel regression models for Disaffection (dummy) and PTV certainty between

PD and PDL (0 = minimum certainty, 9 = maximum certainty).