ECPR General Conference September, Bordeaux

Transcription

1 ECPR General Conference September, Bordeaux Section S50: The Consequences of Crisis for Southern Europe (Chair: Susannah Verney) Panel P240: Party Politics in a Crisis Zone : Party System Change in Southern Europe (Chair: Marco Lisi; Discussant: Myrto Tsakatika) Title: Did the conditions of government accountability survive to the financial crisis? A comparative analysis among the Southern European s PIGS. Author: Stefano Rombi University of Pavia Abstract. Most of the world s democratic governments are party government. Political parties control the decision-making process although often they are not the only ones to do so and, in a high-quality democracy, parties executive control is subjected to a golden rule: the existence of a governed-governors relationship based on electoral accountability. With the onset of the economic crisis, the Southern European s political systems but not just them have experienced some changes. In particular, party (or parties) governments and accountable governments two important pillars of the democratic politics appear under pressure. And someone could say that they are breaking down. This paper aims to provide an assessment of some essential conditions of government electoral accountability. The analysis encompasses four Southern European democracies: Portugal, Italy, Greece and Spain. To contextualize the political impact of economic crisis, each country is examined also in the long run. Assuming that patterns in party government are crucial for the effective work of electoral accountability, the paper focuses on three dimensions: government identifiability, parliament fragmentation, and identifiability of alternatives in government. 1

2 1. Introduction When analysing government accountability, the main concern is the opportunity for the electors to hold the government accountable. The question is: in a certain country, can we find conditions institutional, political, and social which allow voters to oust a government that governs badly? In order to answer this question, we can take into consideration several issues. For instance, one can analyse the possibility of alternation in government, the mass-media system quality or the quality of the electorate (number of switcher voters, interest in politics, political competence). Another, crucial, factor is represented by the level of transparency in government. This will be our focus. Let us first begin with what we mean by transparency. From a negative side, a thing is transparent when it is not opaque, obscure. From a figurative and affirmative point of view, something is transparent when it is characterized by extreme clarity and simplicity. It is something without any ambiguity. Generally speaking, a government is transparent when people can understand what it does and how it does it. Although this meaning cannot be dismissed, it can be integrated. In particular, within a democratic political system, the concept of transparency in government should also imply the possibility for the voters to look over the wall between them and the rulers. In this sense, transparency is not a kind of government s concession, but it is a structural attribute of the political system. Already the presence of this feature may allow voters to identify with relative ease the responsible of a certain political decision. As Przeworski, Stokes and Manin wrote, «governments are accountable if voters can discern whether governments are acting in their interests and sanction them appropriately, so that those incumbents who act in the best interest of citizens win reelection and those who do not lose them» (1999, 40). The possibility of discerning, which means the possibility of properly assign the decisional responsibility, should induce rulers to a responsive behaviour. In a nutshell, from our point of view, a government is transparent if those responsible for the policy choices are easily identifiable, and if, to the extent that they are identifiable, voters will be able to hold them accountable for those choices. In this context, adopting a long-term perspective, we will try to understand if the economic crisis that began in 2008 and is still in progress have an impact on the conditions of transparency in government. Our hypothesis is that there is an impact, in 2

3 the sense of a tendency to a responsibility obscuration in order to escape accountability mechanisms. The analysis, which is mainly descriptive, covers four countries affected, to a similar but not identical extent, from the economic crisis: Greece, Italy, Portugal and Spain. The examination of the conditions of transparency in government will begins from the first democratic election (after the Second World War) and will terminates with the last one. For these reasons, Greece will be studied from 1974 to 2012, Italy from 1946 to 2013, Portugal from 1976 to 2011, and Spain from 1977 to The paper addresses three issues: first, the level of incumbent government identifiability; second, the clarity of the parliamentary processes; third, the electoral supply configuration and its link with the voters political efficacy. In each section, there will be a comparison between the performance of the indicators before and after the crisis. Conclusions follow. 2. The incumbent government identifiability Voters capacity to assign responsibilities for the conditions of their countries is certainly related to individual factors (interest in politics and/or political competence). Nevertheless, the adoption of a micro approach could hide some crucial element. Therefore, it seems appropriate to adopt the perspective employed by many scholars engaged with the economic voting. In this field of studies, individual characteristics are only one side of the coin. The other side is represented by the nature of the incumbent government (Powell e Whitten 1993; Nishizawa 2009; Fisher e Hobolt 2010). In general, the process of responsibilities assignment is positively affected by: first, majority governments; second, governments able to stay in office for the entire duration of the legislature; third, governments that consist of a number of partners as limited as possible. The first condition limits the dispersion of responsibilities and allows voters to easily individuate the relevant political decision makers. The exact opposite happens in the case of minority governments, when the policy decisions are typically the product of parliamentary agreements. They are the result of negotiations often characterized by obscure political exchanges between the government and the parties supporting it from the outside. The second condition prevents that, at the election time, voters have difficulties in identifying which of the various governments should be held accountable 3

4 for the situation of the country. The third condition, limits the phenomenon of inter party buck-passing. Table 1 shows, for each country, some useful indicators that might help us evaluate the level of incumbent government identifiability in their entire democratic history. Table 1. Indicators of government identifiability. Average values. Political system Number of gov Weighted number gov a Number of gov s partners b Effective number of gov s partners Percentage of minority governments Gov crisis length b Spain Greece Portugal Italy II Italy I Source: own elaboration Note: Greece ( ); Italy I ( ); Italy II ( ); Portugal ( ); Spain ( ). a) This is the average of the ratio between the number of governments of a given legislatures and the duration of it. The result of each ratio is multiplied by 1000 to facilitate the reading. b) From the calculus of the number of government s partners are excluded non-partisan governments. Specifically, the figure does not include: the Greek government Grivas (12 October November 1989); the Portuguese Nobre da Costa government (28 August September 1978), Mota Pinto government (November 22, June 7, 1979), Pintasilgo I government (August 1, 1979 to 2 December 1979); the Dini government (17 January January 1996) and the Monti government (16 November December 2012) in Italy. c) The duration of the crisis is calculated from the date of the resignation of the previous government. Regarding the average number of government in each legislature, Spain and Greece show the lowest values, with an index slightly greater than 1.0. These two countries, therefore, seem to represent the ideal situation. They have never exceeded the two governments for legislature. In particular, in the Spanish case all the legislatures, with the exception of that between 1979 and 1982, have a single government. As expected, the first phase of the Italian Republic has the highest values. From 1946 to 1992, Italy had, on average, 4.2 governments for legislature. From 1994 to 2013, however, Italy improved its performances. The average number of governments for legislature fell, in fact, to 2.2. In Portugal, finally, the average index is equal to 1.7. This value seems to be negatively affected by two specific legislatures: and , when, respectively, five and four governments were formed. 1 In formal terms, the second Portuguese legislature does not proceed from 1976 to 1979, but from 1976 to The elections of 1979 (so-called eleições intercalares), in fact, were called by the President of the Republic Eanes immediately after the resignation of the government led by Mota Pinto, with the sole aim of completing the legislature that, according to provisional standard of the Constitution, must necessarily be completed in October

5 The mere number of government, though relevant, it does not seem sufficient to discover which countries have the highest level of transparency in government. It is quite reasonable, in fact, to argue that the identifiability of the responsible government also depends on the length of the legislature. A significantly short legislature, in fact, does not allow the government in charge to complete its policy program and, consequently, complicates the process of vertical accountability. In the light of this argument, it could be very interesting to look at the average number of government weighted for the length of the legislature (third column Table 1). Regarding this indicator, Spain confirms its vigorous tendency to produce identifiable government. The index value is noticeably limited (0.954), and is simply the product of the interaction between stable governments and long-lived legislatures. More precisely, the Spanish legislature, whose duration is four years, had an early termination for three times. However, only the first legislature ( ) had a significantly short life (623 days). Portugal, instead, shows a much more consistent weighted number of governments (1.916). This is due to the great systemic instability typical of the first phase of Portuguese democracy and, specifically, to the period between 1979 and 1987, when the four legislatures lasted in, on average, 623 days, with a minimum of 276 days ( ) and a maximum of 747 days ( ). But the highest weighted number of governments regards the legislature (4.296): in 931 days there were four governments, all supported by the majority composed by the Social Democrats and the Social Democratic Centre (CDS). Greece has the most pronounced discrepancy between the average normal number and the average weighted number of governments. The first one is equal to 1.3, while the second is considerably higher: 2.7. Why is it so? The main reason seems very simple. We have to think of the exceptional shortness of the last two legislatures in the 1980s. More specifically, in the electoral cycle begun in June 1989 and ended in November of the same year (126 days) the weighted number of governments was (real number: 2) while in the electoral cycle begun in November 1989 and ended in April 1990 (136 days) this value reached (real number: 1). Of course, all this affects the average value. Excluding that turbulent period, the weighted average number of governments falls to However, such moments of the Greek political system cannot be neglected. All the more that, as we shall see later, in the late spring of 2012 there was a similar situation. 5

6 In Italy, finally, there is a net improvement between the first and second phase of the political system. During the so-called First Republic, the index stood at 3.018, while between 1994 and 2013 it fell, on average, to The dynamics behind this significant improvement was, however, asymmetrical. This means that only one of the two competing coalitions the center-right has been able to produce stable governments without the need to anticipate the conclusion of the legislature. In the phase, for example, only in the case of the legislature, there was only one government: the second government headed by Romano Prodi. Moreover, it seems appropriate to point out that the government Prodi II, having operated within a short legislature (731 days), cannot be described as a government of legislature. As well known, in fact, only those governments able to remain in office until their formal expiry date may deserve this name. The number of parties in government is another useful indicator for our purposes. How many parties do compose the Council of Ministers? If the answer to this question is one, few or many, the picture changes dramatically and, with it, changes the level of clarity of responsibility. This point, however, must be integrated. The number of partners, though essential, seems not to be sufficient. In fact, it should also take into account the possible presence of parties that support the government from the outside. The distinction is crucial. Indeed, it is reasonable to assume that a party government supported by a solid parliamentary majority allows a better identification of those responsible than a party government that cannot count on a compact majority. The latter, in fact, has to negotiate, more or less frequently, the content of each policy (Powell 2000). Therefore, it seems appropriate to discuss simultaneously these two issues. If we consider only the number of partners in government, we would argue that, even according to this indicator, Spain and Greece have the same level of government identifiability. Their average number of governmental partners is equal to 1.1 and 1.5. In short, the governments in these countries are, with very rare and unexpected exceptions, single party governments. However, also considering the nature of the government important differences emerge. In fact, as showed in the last column of Table 1, the percentage of minority governments is limited in Greece (16.7%) and very high in Spain (55.5%). In the Iberian country, minority governments are more frequent than majority governments. Therefore, in many cases the decision-making process is much more complex and much less easily intelligible than the presence of a single-party government would suggest. 6

7 In an intermediate position fully compatible with the presence of identifiable governments we find Portugal. Its average number of government s partners is 2.3. Even if the number of partners is limited, it is necessary to consider the substantial weight of the non-partisan ministers. In Portugal the presence of a group of non-partisan ministers occurred in sixteen out of twenty governments (three of which were not considered because completely composed by non-partisan ministers). In addition, in four cases government Cavaco Silva II (PSD), Cavaco Silva III (PSD), Guterres II (PS), and Socrates I (PS) the group formed by technical ministers was the sole partner of the party in government. In these cases, it was a one party government assisted by the expertise of some specific technical ministers. Ultimately, Lusitanian governments are less transparent than the Spanish and Greek ones, mainly because of the extensive use of technical ministers, which are free from any relationship with the voters, although they operate within a largely partisan executive. Furthermore, we have to take into account the relative importance of minority governments, which represents 40% of the Portuguese executives. Also regarding the government s composition, Italy confirms its poor performance. Looking at the number of parties in government, an increase in the transition from the first to the second republican phase emerges (the index has almost doubled from 3.5 to 6.7). But this trend has been accompanied by a vertical drop of minority governments from 46% to 9.1%. If from a clarity of responsibility perspective the decrease of minority governments is a positive factor, the fragmentation of governments is increasingly worrying. Between 1946 and 1992, for example, the maximum executives extension was represented by the Pentapartito governments formed by Christian Democrats, Socialists, Social Democrats, Republicans, and Liberals. But the highest number has been reached by the Prodi II government ( ), which was made up by nine parties and one non-partisan minister: a level of widespread responsibility and, therefore, hidden unbeaten, in Italy as elsewhere. The number of normal government does not represent carefully the governmental fragmentation. Two governments, in fact, though composed by the same number of parties, would work very differently if a preponderant number of ministries was assigned to only one party or, on the contrary, if there was not a dominant party and the ministers distribution was egalitarian. The first kind of government would probably be identified with the major and, at least apparently, more responsible for the governmental choices political forces. Therefore, the decision-making process would 7

8 be clearer. And, of course, this clearness would allow to easily identify the culprits or, if any, the deserving. The fifth column of Table 1 shows an index able to express in a condensed form, as well as the number of parties in government, the distribution of the various ministries among them: the effective number of government s partners, calculated using the Golosov index (2010) 2. Looking at the fourth and fifth column of Table 1, you can discover that the effective number of partners is, sometimes, very similar to the normal number and, sometimes, more limited. In particular, in Spain and Greece, the two values are analogous. This pattern is typical of systems in which a single party usually composes governments. In all other cases, however, there are significant differences. In Portugal the effective number of partners is equal to 1.4. Even in Italy, the effective number is much smaller than the unweighted number, both in the period (1.7) and from 1994 to 2013 (3.1). Despite this general trend, however, an in-depth data analysis reveals some differences. Examining the two centreleft parliamentary majority in the so-called Second Republic ( and ), it should be emphasized that the average number of normal government partners has grown from 7.5 to 10.0, while the weighted number has decreased from 4.2 to 3.9. This trend seems to be the effect of the concentration of sixteen out of twenty-five ministries in the two major parties of the coalition, the Left Democrats (DS) and the Marguerite (DL). This indicator, however, reveals important differences in terms of government solidity between centre-left and centre-right governments. The index, in fact, takes the lowest values in the two legislatures occupied entirely or in large part by the Berlusconi governments (3.0 in and 1.2 in ), reflecting the pivotal role held by Go Italy (FI) from 2001 to 2006 and by the People of Freedom (PDL) from 2008 to In addition, we must also consider the influence of partners with a limited number of ministers. In fact, they usually have the ability to exercise a veto power on the governmental choices. It is a crucial element of difference between the majoritarian democracies based on one-party governments, and the majoritarian democracies based on coalition governments. These two types of systems, in fact, work very differently because of the blackmail power exercised by the minor partners of the coalition 3. 2 The index of Golosov seems preferable to the better-known index of Laakso and Taagepera (1979) for its better interpretation of party systems characterized by a dominant party. 3 From this point of view, Silvio Berlusconi s discourse of 28 April 2005, during the parliamentary vote of confidence to his third government, cited in Massari (2005), appears very significant: «where is the flaw? [...] The lack of our system is that, within a coalition, there isn t the democratic rule of majority and minority [...]. It happens that we have a meeting, we present our proposals, and, if only one of the 8

9 However, in relative terms, the existence of some big parties leading the coalition, limiting the dispersion of decision-making process seems to allow a better identifiability of government responsibilities, compared to a situation in which there is not a party significantly more representative than the others. In addition, a possible symptom of the lack of transparency in government is the duration of the crisis of government. It seems reasonable to argue that: the shorter the crisis, the greater the effectiveness of the elections in determining the government structure and the lower the complexity of inter-party relations. Regarding this indicator, the case of Greece does not arouse great surprises: the crises are very short as proof of the absolute domination of the electoral arena. The only real exception is the legislature, when the formation of a large post-election coalition between ND, PASOK and SIN (Coalition of the Left) took 18 days of negotiations. The longest crisis took place between 9 September 1993 the day of the resignation of the leader of ND Mitsokatis and 10 October 1993, when, three days after the election, the third Papandreou government was formed. Regarding Italy, in the phase crises are, on average, longer than the previous phase: 49 days and 34 days. As we shall see, the figure of the phase is very negatively affected by the recent political events. In Portugal and Spain the distance between the fall of the previous government and the investiture of the new head of government is about a month in the second case and nearly 50 days in the first. In the case of Spain, the time does not seem excessive, but it is unexpectedly similar to that of the first phase of Italian democracy ( ). In this context, there are some important exceptions, especially in the Spanish case. We have to look, in particular, to the formation of the Aznar government, which took office on 5 May 1996, after 63 days of negotiations between the Popular Party (PP) led by Aznar and the Catalan party Convergence and Union (CiU) led by Jordi Pujol. The first Aznar government, despite being a minority government, lasted an entire legislature like almost all Spanish elections, the 2000 elections were anticipated for a few months confirming the robustness achieved by the outside support granted by the Catalan, Basques and Canary Islanders nationalists. The Portuguese case, instead, is negatively coalition parties do not agree, we cannot proceed in that direction. There are parties representing perhaps 6-7 percent of the coalition and if they use their veto powers you cannot go forward. Thus, the 94 percent or so of the coalition gives up to a 6 percent or so of the coalition, whose votes in Parliament it is necessary to have a majority able to approve a bill». 9

10 affected by the three legislatures of the period , when the average duration of the crisis was 138 days. After having analyzed the situation of our countries in a long-term perspective, we still have to see if and what has changed over the years of the economic crisis (2007/ ). Table 2 shows the same Table 1 indicators for the crisis years. Table 2. Indicators of government identifiability. Crisis years. Average data. Political system Number of gov Weighted number of gov Number of gov s partners Effective number of gov s partners Percentage of minority governments Gov crisis length Spain Greece Portugal Italy Source: own elaboration Note: Greece ( ); Italy ( ); Portugal ( ); Spain ( ). Data are updated to 1 st August Compared to the general trends of each country, in the course of the ongoing Southern European economic crisis the level of incumbent government identifiability does not seem to have suffered big jolts. There is, however, a sharp deterioration in the performance of some indicators. First, in Greece the weighted number of governments has increased from an average figure of during to considering only the period Other indicators, instead, are consistent with the overall trends. An important exception has to do, however, with the average duration of the government crisis that, excluding Greece, has increased exponentially in all countries examined. In particular, it went from 38 to 143 days in Spain, from 48 to 60 days in Portugal and from 49 (or from 34 considering the phase) to 78 days in Italy. The increase of the government crisis length is closely related to the impact of the economic crisis, though its causes are different from country to country. In Spain, for example, the 5 months of political crisis are due to the very early resignation of the Zapatero government, which took place in June 2011, as the socialist executive could not resist to the opposite pressures on both the domestic and the international front. In short, the political crisis comes to a full-blown inability to tackle the economic crisis. In Italy, instead, the length of the average government crisis is not the result of the three days process begins with the resignation of the government Berlusconi IV and ending with the birth of the caretaker government led by Monti. Rather, it depends on the time required for the formation of the post-electoral grande coalizione government. The 10

11 government led by the democrat Letta has the mission to continue the policies undertaken by the Monti government and the specific task of limiting the impact of the mechanisms of electoral accountability. 3. Parties in parliament and government accountability As part of an examination of government electoral accountability, the analysis of the parliament can only be partial and focused on its structural aspects. By focusing on the behavior of parliamentarians and their method of election, in fact, would shift the focus of the analysis from the government accountability to that of the single MP. The topic, however, though not part of our focus, is certainly worthy of study and analysis. In general, two factors affect the level of accountability of the single elected official: first, the level of trasformismo; second, the nature direct or indirect of the election. The four democracies of Southern Europe seem not to vary significantly along these two dimensions. Italian (at least since 2006), Spanish and Portuguese MPs are elected within closed proportional lists, which, at least in Italy, are very long and do not even appear on the ballot. This, of course, depresses the relationship between elected officials and voters, with the latter pledged to be accountable to the leaders of the party rather than to their electoral constituency. In Greece, however, there is the preferential voting and, at least in theory, the voters-elected relationship is more direct. Regarding the first dimension, in a depressing general context, Italy seems to be the country with the lowest performance. The trasformismo that is, the change of partisan affiliation during the legislature not only represents a repudiation of the voters mandate but, at the elections, also makes extremely complicated the evaluation of the switcher MP, who, probably, will be presented within the list of his new party. Apart from these general considerations, which should be explored more carefully, in this paper we are interested in the parliaments structural aspects. In particular, it will be necessary to identify some indicators that might furnish relevant information about the complexity of parliamentary decision-making process and, therefore, about the possibility for voters to understand it. Reasonably, it can be argued that a simplified parliament, which consists of a small number of components, is easier to monitor and thus can be able to offer the electorate the best opportunity to express an evaluation at the elections. If so, we have to consider the parliamentary fragmentation. 11

12 Modern parliaments are articulated into parliamentary groups that, very often, reproduce the dividing lines between the various political parties. However, it may happen that different political forces share the same parliamentary group, so that the same choices of the group might be the result of bargaining, far from transparency and publicity, among the parties that compose it. 4 This situation can occur for two reasons: first, the rules governing the formation of groups oblige political forces unable to elect a sufficient number of MPs to enter together a particular parliamentary group (i.e. the Italian gruppo misto ); second, electoral agreements between parties can lead to the formation of a unitary group between political forces that, de facto, remain independent. 5 For these reasons, the most useful way to compare parliamentary fragmentation among our countries is to consider the number of parties that are able to obtain at least one parliamentary seat. More specifically, we analyse the average number 6 of parties obtaining seats. Also in this case, the Golosov index is employed to calculate the effective number. 4 Moreover, although this practice is mainly Italian, it may happen that during a legislature some groups are formed without any connection with the election results. This can happen for several reasons. As an example, we can refer to two very different cases. The first case occurred in the XV Italian legislature in the Senate where, according to the internal rules, the groups that, because of their small size, could not appoint thirteen different senators in each of the thirteen standing committees, may appoint the same senator in three commissions. Because of its numerical instability, the centre-left tried to take advantage of this rule, composing the group For Autonomies through the fictitious switching of some Ulivo senators. With the same mechanism, the centre-right formed the group DC for Autonomies-Italian Republican Party-Movement for Autonomy. The second case occurred during the XVI Italian legislature and could not be attributed to the rules incentives. It was due to real and specific political differences. This is the birth, between July and August 2010, of the group Future and Freedom for Italy, made up by 33 deputies and 10 senators, all elected within the lists of the People of Freedom. 5 Considering the Italian case, it seems relevant the recent example of the Radical Party candidates, which in 2008 were elected within the lists of the Democratic Party. The elected radicals went in the same parliamentary group of the Democrats, but they continued to maintain their independence. 6 Disaggregated data are available contacting the author. 12

13 Table 3. Parliamentary fragmentation of party system. Average data. Political Normal number of Effective number of System a Plenum parliamentary parties parliamentary parties Greece Spain Portugal Italy I b Italy II Source: own elaboration. Notes: a) When there are two chambers, data refers to the low one. b) The average number of MPs is equal to 611 because until the constitutional law of 9 February 1963, Italian Constitution did not state a fixed number of MPs. On the contrary, article 56 stated: the Chamber of Deputies is elected by direct and universal suffrage, with one deputy for eighty thousand inhabitants or for a fraction exceeding forty thousand. In each country, the third column of Table 3 shows the average number of parties that manage to get at least one seat. Fragmentation is very limited in Greece and Portugal, where the parliament is composed, on average, by a number of parties that oscillates between four and five. Although the last two Greek elections caused a parliament with seven parties, this figure equaled the record of the second legislature ( ). In that case, the three parties received less than five seats, and one of these small parties the Neoliberal Party (KN) was incorporated by New Democracy (ND) since The successive legislature ( ), however, was the one with the least number of parties in parliament: the conservative ND, the social democratic PASOK, and, finally, the communist KKE. The Portuguese case, instead, does not show exceptions. In none of the cases, not even in the years immediately after the authoritarian regime of Salazar and Caetano, more than five parties get seats: in nine out of thirteen legislatures (one of which is still in progress) the normal number of parliamentary parties is equal to five, while in the remaining four legislature it is equal to four. In Spain and Italy (both in the First and in the Second Republic), the number of parties in parliament is on average much higher than in the other cases. In the Italian case, between 1946 and 1992, there was an average of 11.7 parties able to get seats. In spite of this remarkable consistency, during the phase the average reached 13.8 parties. Therefore, the largely majority electoral system adopted in 1993 has in no way reached the goal of reducing the number of parties. Indeed, it contributed to the fragmentation by assigning a consistent blackmail power to the small parties. In Spain the average number of parties with seats is This high level of fragmentation is 13

14 solely due to the success, now decreasing, of the partidos de ámbito no estatal (Rombi 2012). These data acquire greater importance considering them in effective terms. The fourth column of Table 3 shows the effective number of parliamentary parties. Looking at this indicator, which takes into account the relative weight of the parties in parliament, emerges that Greece and Spain have an average effective number of parties in parliament equal respectively to 2.1 and 2.3. The Greek data ranges from 1.4 (first legislature, ) to 3.8 of the legislature between May and June However, it should be considered that the exceptionally low value of the first democratic elections is an anomaly due to the enormous success of ND (220 seats out of 300). If we put aside this circumstance, in fact, the minimum value is represented by the 1.9 parties reached in the legislatures and In diachronic terms, the indicator suggests a slow, few significant, but apparently unstoppable growing trend. This pattern, as mentioned, exploded after the two elections held in May and June of In Spain, the average number of parties varies between 2.9 ( ) to 1.9 ( ). In the latter case, the PSOE controlled 58% of the seats (202 out of 350). Portugal occupies a mid-range position. Its effective number of parliamentary parties is, on average, equal to 2.4. Less than two and a half parties do not produce poorly understandable parliamentary dynamics. Moreover, as we already know, the average normal number of parties, equal to 4.7, is the lowest among our cases. The Portuguese data do not show any particular trend: the two highest values (2.8 and 3.4), in fact, characterized, respectively, the legislature but also the ones from 1983 to 1985 and from 2009 to 2011 and the legislature. In 1976, it was the success of the Portuguese Communist Party (PCP) causing a high number of effective parties. Instead, in 1985 the figure is due, above all, to the exploit of the Renewal Democratic Party (PRD), founded by the former General and President of the Republic Ramalho Eanes. However, starting with the 1987 elections, there was a consistent erosion of the PRD parliamentary base (went from 45 to 7 seats) and the Portuguese parliament become highly simplified. After a long period in which the effective number was between 1.8 and 2.4, with the 2009 elections it returned, as mentioned, to 2.8 parties as a consequence of the rise of the Leftist Bloc (BE), which, obtaining 16 seats, doubled the result of 2005 elections. Quite different is the situation of the Italian parliament. In Italy, as well as for the normal number, the worst performance is during the so-called Second Republic. If 14

15 during the Italian parliament was constituted, on average, by 3.2 effective parties, during this number, including the results of the elections of February 2013, is equal to 4.6: not far from the exceptionally high figure of the famous French Fourth Republic (5.4). Turning to a diachronic evaluation of the data, it is possible to isolate at least three political phases. In the first one, between 1946 and 1972, the number of parties represented in parliament varies from nine to twelve, with an effective number on average equal to 2.9. In the phase the number of parliamentary parties rise up to a range of ten-sixteen. The effective number is equal to 3.6, with a significant growth 4.7 during the last legislature of this phase ( ). From 1994 to 2013 the number of parliamentary parties varies between twelve and eighteen, with a further increase in the average effective number. The highest number is the one of the legislature : 8.2 effective parties. Bearing this in mind, now we have to examine the indicators during the years of the economic crisis. Table 4 shows the data. Table 4. Parliamentary fragmentation. Crisis years. Average data. Political Normal number of Effective number of Plenum System parliamentary parties parliamentary parties Spain Portugal Italy Greece Source: own elaboration. Note: when there are two chambers, data refers to the low one. Data indicates three main elements. First, Spanish indicators are similar to those before the crisis. Second, during the economic crisis, Italian parliamentary fragmentation is decreased: the normal number of parties fall from 15.2 to 11.0 (the overall number is 13.8) and the effective number decreased from 5.6 to 2.7 (the overall number is 4.6). Third, Portugal and especially Greece are characterized by an increasing parliamentary fragmentation and, consequently, by a significant decrease in the intelligibility of the parliamentary processes. In Portugal this pattern seems to be limited. The average normal number of parties, in fact, increases just from 4.6, in the pre-crisis era, to 5.0 after the elections of 2009 and 2011 (the overall index is equal to 4.7). The average effective number, instead, rises from 2.3 to 2.6 (the overall index is equal to 2.4). But the most profound and troubling change occurred in Greece. In the Hellenic country, on average, the number of parliamentary parties shift from 4.4 to 6.0 (the overall number is 4.8). Moreover, there is a significant increase in the effective number of parties, which 15

16 rises from 1.9, in the pre-crisis era, to 2.8 after the four general elections of the crisis (2007, 2009, May 2012, and June 2012), with a peak of 3.8 after the elections of May 2012 (the overall index is 2.1). 4. The alternatives in government identifiability The examination of the incumbent government identifiability, on which we focused up to now, is directly related to the voters possibility to vote retrospectively, taking into account the appreciation of the public policies implemented during the past legislature (Fiorina 1981). However, the highly complicated cognitive process that leads to the electoral choice seems to be also related to prospective factors. Considerations about the past and predictions about the future work simultaneously and, in this case, what count is: the programs presented in the electoral campaign, the composition of the future executive, and, last but not least, the designated Premier (Downs 1957). The prospective evaluations may be correct only if the elections promises are as clear as possible. But what determines their intelligibility? The essential variable is represented by the type of political supply. The electoral supply configuration, which is largely influenced by both the electoral system and the party system, has a major impact on the degree of the voters political efficacy (Nardulli 2005). In other words, in those countries where two clear alternatives in government constitutes the supply, the legislatures tend to be occupied by only one government. On the contrary, where this is not the case, the weight of the first government will tend to decrease, preventing that the successive elections may operate as a punishment/reward mechanism. In this sense, the relationship between the characteristics of the electoral supply related to prospective considerations and the government identifiability linked to retrospective considerations looks very tight. Table 5 shows some indicators in order to indirectly measure both the voters political efficacy and the configuration of the electoral supply. Specifically, we focus on four elements: first, the percentage of governments born immediately after the elections (it is irrelevant if they are the consequence of postelectoral bargaining or if they derive from clear pre-electoral alternatives); second, the length (measured in days) of the first government of the legislatures; third, the relevance of the first government within the legislature; fourth, the level of clearness of the preelectoral alternatives in government. 16

17 Table 5 Voters political efficacy and alternatives in government identifiability. Average values. Political System Percentage of first governments Length of the first government Relevance of the first government Identifiability of alternatives in government (IAG) Spain Greece Italy II Portugal Italy I Source: own elaboration. Note: for the calculation of the IAG is also considered the political supply of the last election: Spain 2011, Portugal 2011, Greece June 2012 and Italy Generally speaking, in order to have a high level of voters political efficacy, politicians have to strictly follow the election outcomes. When this is not the case, or it is rarely, the will of the voters may be easily bypassed, worsening the quality of democracy. Of course, an electoral competition between two alternatives in government, where voters may know in advance the configuration of the government of their country, represents the ideal situation. However, even the first post-electoral government seems more tied to the electoral outcomes than the successive governments of the same legislature. For these reasons, the growing importance of governments born immediately after the election, being a symptom of a growing clarity of responsibility, seems a very important condition for the functioning of accountability mechanisms. As we can see from the second column of Table 5, 95% of the Spanish governments are born immediately after the elections. The sole exception is the government led by Leopoldo Calvo-Sotelo, which was appointed president of the Centre Democratic Union (UCD) majority government on 25 February 1981 (immediately after the resignation of Adolfo Suárez), and remained in office for 611 days. In Greece and Portugal more than 80% of governments derived, directly or indirectly, by elections; more exactly: 84.6% in the first case, and 82.9% in the second. In Italy, between 1994 and 2013, the first governments constitute 55% of the total. If we consider the period this percentage fall to 26.9%. Although the improvement appears undeniable, it should signaled the usual exception of the legislature, when the centre-left majority produced four different governments. However, it should be also stressed that the transition between those two phases produced a great change due, principally, to electoral reforms. Until 1992, government coalitions were formed ex-post in parliament. On the contrary, in the successive phase the electoral coalitions would-be governmental coalitions presented, ex-ante, to the 17

18 voters the leader and the government manifesto. With the 2013 elections, however, the system seems to be back to a situation more similar to the one of the so-called First Republic. Sorting the countries according to the length of the first government showed in the third column once again the excellent performance of Spain (1138 days) and Greece (970 days) emerges. But the Greek case requires a clarification. Its average value, in fact, is negatively affected by the convulsive phase begun in the summer of 1989 and completed at the end of That phase was characterized by two very short governments: 101 and 136 days. If they were not considered, the average length of the first Greek governments would rise up to 1125 days. In diachronic terms, in these years of economic crisis, the Hellenic country is going through a phase of serious political instability. As will be discussed in more detail later, even excluding the troubled phase began in May 2012, it should be noted that since 2007 the duration of the first government has decreased dramatically. Both the government led by Karamanlis and those led by Papandreou, in fact, lasted about two and a half years: 746 days, the first one, and 765 days the second one. In Portugal and during the second phase of Italian republic, first governments lengths is, on average, slightly more than 850 days. Which means, for Italy, a very significant step forward. Comparing and data, a clear positive trend emerges, though we have to point out that, unlike the other indicators, in this case is the centre-right, with the Berlusconi I government which lasted only 226 days that represents the most relevant exception to the rule. The opposite exception in the First Republic, however, is represented by the Craxi government that, in the years , was in office for over 1038 days. A very high longevity, especially considering that, between 1946 and 1992, the average length of the first government of the legislature was only 333 days. As we have already pointed out, it is useful to divide the Portuguese political system into two periods: e Even regarding the length of the first government this kind of temporization seems to be relevant. In particular, there is a clear increasing trend in the duration of this type of governments. Between 1976 and 1987, in fact, their average life was equal to 425 days, with a peak of 747 represented by the Soares III government: the so-called governo do Bloco Central, born as a consequence of a large post-electoral coalition between the Socialist Party (PS) and the Social Democratic Party (PSD). This government, however, aimed precisely to produce a spreading in the decision-making responsibilities, in order to allow parties to avoid a 18

19 likely negative electoral judgment due to unpopular policy choices (Magone 1999). In other words, the formation of such a government was intended, although this aim was not declared, to reduce the electoral accountability and, therefore, the level of responsiveness. Between 1987 and 2011 the average was 1178 days, with a maximum of 1660 days occupied by the Socialist government lead by the former social democrat José Sócrates ( ). The duration of the first governments of the legislature helps understand whether and to what extent the electoral arena is more relevant than the parliamentary one. However, you have to look also at their relevance within each legislature. How is the weight of the first government? To answer is sufficient to relate its length with that of the legislature. To avoid paradoxical situations, 7 we related the longevity of the first government not to the actual duration of the legislature but to its potential duration prescribed by the Constitution. 8 The index varies from a theoretical minimum of 0 to a maximum of 1. According to this indicator, the differences among our countries seems to be identical to those emerged according to the percentage of first governments. Therefore, on average, Spain and Greece are largely dominated by the first governments of the legislature. The Spanish index is equal to 0.779, while the Greek one is equal to In these countries, very frequently, there are governments formed immediately after the elections and able to stay in office for the whole legislature. Portugal is close to the first two countries, with an average index equal to During the Portuguese democratic history, there has been periods in which the first government occupied the whole legislature and phases in which this has not happened. As for all the indicators proposed, even in this case there is a clear difference between the pre-1987 legislature and the post-1987 ones. From 1976 to 1987 the average index stops at Despite the limited relevance of the Sócrates II government (0,351) , from 1987 to 2011, instead, the average index reaches to Italy ranks, once again, in the back rows. The average index stood at during the period and at during the successive phase ( ). Also in this 7 Such as, for example, to assign to the Prodi II government the same level of relevance attributed to the Spanish government of legislature. 8 Specifically, a legislature lasts four years (1460 days) in Greece and Spain, five years (1825 days) in Italy. In Portugal, the duration of a legislature was fixed in four years. However, its duration may also be higher in the case in which the elections are held during a legislative session (between mid-october and mid-june). For this reason, the legislature , for which voting took place in March 2005, lasted four and a half years. Beyond these aspects, it is still reasonable to consider the duration equal to four years, interpreting Sócrates I government, the only one that has exceeded this limit, as a government of legislature, which, in numerical terms, means to attribute it the value 1. 19