Presidential power and bargaining complexity in the party selection of Italian heads of state, 1948-2022

ABSTRACT

Among political scientists, presidents in parliamentary democracies have received little attention as compared to their popularly elected counterparts. Yet, there is evidence of influential heads of state beyond semi-presidential and presidential systems, and the Italian one is a case in point. Scholars agree that the ‘informal power’ of Italian presidents has grown substantially since the early 1990s, due to the combination of weak party organisations, the personalisation of politics, and the mediatisation of the presidency. While the literature shows that the choice of the president has become more salient for parties, hardly anything is known about the impact that the increased presidential power has on the complexity of the selection process. This article argues that, when presidents are powerful, parties face high adverse selection costs and, therefore, party leaders will be less likely to compromise on candidates. This, in turn, can lead to political stalemates. Using a novel measure of bargaining complexity, the empirical analysis supports this argument, which holds also after controlling for the contingent features of the parliamentary party set-up. The findings have implications for the study of political leaders and party behaviour at a time of party government decline.

KEYWORDS:

• President
• Italian head of state
• parliamentary democracy
• party bargaining
• presidential selection
• presidential power

1. Introduction

On 22 January 2020, a vast majority of the Greek members of parliament voted for Katerina Sakellaropoulou as new president of the republic; only a few abstained. Similarly, almost 80% of the German Bundestag’s members, together with state delegates, elected Frank-Walter Steinmeier for his second term in office at the first try on 13 February 2022. In contrast, it took six days and eight rounds of voting for the Italian parliament to re-elect Sergio Mattarella in January 2022; in this case, profound intra-party disagreements affected the process. In light of this, one could ask why, in parliamentary democracies,¹ some presidential elections are smooth processes, while others turn out to present serious problems of collective action.

This article focusses on the selection² of presidents of the Italian parliamentary republic over more than seven decades (1948–2022). In particular, it investigates the relationship between the political influence³ of the head of state and the complexity of party bargaining during the selection process. It finds that, when informal presidential power increases, inter-party negotiations become more complex.

Due to their limited political power, presidents in parliamentary democracies have generally attracted little attention from political scientists, especially if compared to their counterparts in presidential and semi-presidential systems (e.g. Blondel 2015; Raunio and Sedelius 2020). A couple of studies on the impact of presidential party affiliation on the choice of prime ministers (see Kang 2009; Schleiter and Morgan-Jones 2009; Bucur and Cheibub 2017) as well as power-sharing in dual executives (Anckar 2022) are remarkable exceptions. Yet, the Italian case is deviant in that it has been the subject of several meaningful single-country studies (cf. Cassese, Galasso, and Melloni 2018).

In this regard, it is worth noting that the number of these studies has considerably increased since the early 2000s, after the strengthening of Italian heads of state at the expense of political parties and cabinets. Both the early 1990s breakdown of the first republican party system and then the increasing personalisation of politics proved to be crucial factors in fostering such change (e.g. Amoretti and Giannone 2014),⁴ in spite of the absence of any relevant constitutional reform of the presidency. Most importantly, signs of a growing difficulty in finding suitable candidates for the presidency have emerged (e.g. Tebaldi 2005; Passarelli 2022). The overall takeaway message of these works is that the Italian political system has been experiencing a long-term change in the way in which political parties relate to heads of state and tackle their selection. Against this background, one can speculate about the presidential selection process: should it become more complex (if conflictual), it may in fact be conducive to more frequent political stalemates and poor parliamentary performance.

In spite of this topic’s relevance, systematic investigations of the effect of increasing presidential power on the political parties’ ability to act as efficient selectors of the head of state are missing, to say nothing of the causal mechanisms linking the two. This article seeks to fill this gap and aims to assess the impact of presidents’ political influence on the degree of complexity in inter-party negotiations in Italian presidential elections. It addresses two questions: Why are political parties less likely to agree on the presidential candidate when the president is a consequential actor? Does the party system affect the course of the selection process?

The next section reviews the debate about the growing political power of Italian presidents and describes the constitutional and political facets of their election over the decades. Subsequently, the article presents its core theoretical argument and the relevant expectations. Some methodological clarifications and an empirical analysis lead to the discussion of the findings and suggestions for further research.

2. The Italian president: power and selection

2.1. The strengthening of the president

The Italian constitution endows the head of state with little formal political power, if compared to the parliament and the executive.⁵ Besides specific rights to appoint senior civil servants and other monocratic prerogatives, key formal powers are those of promulgating laws, dissolving Parliament, calling fresh elections, and appointing cabinet ministers (Clementi 2016, 623–624). But the constitution only provides a legal framework (e.g. Grimaldi 2021) for the actions of the president, whose actual power also depends on other factors, such political parties’ capacity to fulfil their basic functions (Dalton, Farrell, and McAllister 2011). Building on this premise, this article looks beyond the formal rules at the ‘real’ influence of the head of state within the political system, defined by the political opportunity structure. I assume that presidents are strong inasmuch as they have the potential to affect political outcomes; moreover, I posit that he or she will translate this potential power into actual behaviour whenever possible.

In this regard, the power of Italian presidents – as well as the related public expectations concerning their role – has substantially changed since the establishment of the presidency in 1948.⁶ This has mostly occurred because the political context in which the head of state operates has been transformed, opening new room for presidential manoeuvre within the same constitutional framework.⁷ In particular, the scope of presidents’ outreach has expanded since the disappearance – or the significant redefinition – of the parties, which, until 1993, had for long dominated Italian politics. One major follow-up of this political event was the emergence of new parties, organisationally weak and poorly rooted in society. In the eyes of voters, presidents were resilient and reliable institutional figures, who could legitimately take the lead in filling the representation voids left by a discredited political class (Pasquino 2012; Amoretti and Giannone 2014: 446, Amoretti and Giannone 2016, 71). According to some observers, the ultimate watershed came with the first Napolitano presidency (2006–2013), which was conducive to a de facto long-term reshaping of inter-institutional power relations to the benefit of the head of state and to the detriment of both Parliament and the executive (Osservatorio di Pavia 2013, 271–272). In a nutshell, formal presidential powers have remained the same, but the head of state has become a more politically important actor over the years.

Increasing presidential influence has been most apparent in two fields: cabinet formation and policy-making.⁸ With regard to the former, presidential support has turned into a key resource for those seeking to be selected as prime minister (Barbieri and Vercesi 2022). At the same time, presidents have become more likely to veto the appointment of cabinet ministers (Tebaldi 2005, 159–161). Sergio Mattarella’s rejection of the formateur’s proposal of Paolo Savona as minister for the Economy in the future Conte I cabinet is merely an illustrative example of a relatively recent trend (Valbruzzi 2018, 474; Stancanelli and Arsì 2022, 1696–1699). With regard to policy-making, presidents have taken advantage of extensive media exposure to affect policy outcomes prior to their definition⁹ (Grimaldi 2015, 80–85).

For the most part, Italian presidents have affected policy-making through so-called ‘moral suasion’, whereby they use their public authority and credibility to modify other political actors’ behaviour. This type of intervention gained momentum under President Carlo Azeglio Ciampi (1999–2006) and became established in subsequent years (Amoretti and Giannone 2014, 448). Moral suasion typically occurs through public statements (esternazioni), whose main addressee is public opinion.

In light of these developments, it is not surprising that the media have increasingly focussed on presidential actions. For example, while the daily number of articles in the Italian newspaper, La Stampa, mentioning the president’s name ranged from zero to one between 1948 and the late 1970s, it increased to three between the 1980s and the early 1990s. Finally, it stabilised at around four in the following decades (Amoretti and Giannone 2014, 444).

2.2. Changes in presidential selection

In Italy, a joint session of Parliament and three delegates from each region (one from the Valle d’Aosta) chosen by their regional councils elect the head of state (who must be at least 50 years old). The election is conducted by means of a series of secret ballots, and voting rounds continue until a candidate wins. While a qualified majority of two thirds is required at the first three rounds, an absolute majority suffices thereafter.

The term of office lasts seven years, being longer than the constitutionally defined legislative term of five years.¹⁰ The constitution is silent about the electability of the same president for a second term, which means that it does not explicitly prohibit it. Based on this formal framework, 11 presidents have been elected in 13 elections since 1948. Table 1 shows their names, their duration in office, and their party and institutional backgrounds.

Table 1. Italian presidents, 1948–2022.

President	Year of election	Prior party affiliation	Duration in office (years)	Highest post held in Parliament	Highest post held in cabinet
1. Luigi Einaudi	1948	UDN	7	Member of the Senate	Deputy PM
2. Giovanni Gronchi	1955	DC	7	Chamber speaker	Minister
3. Antonio Segni	1962	DC	3.6	Committee chair	PM
4. Giuseppe Saragat	1964	PSDI	7	Committee chair	Deputy PM
5. Giovanni Leone	1971	DC	6.5	Chamber speaker	PM
6. Sandro Pertini	1978	PSI	7	Chamber speaker	-
7. Francesco Cossiga	1985	DC	6.8	Senate speaker	PM
8. Oscar Luigi Scalfaro	1992	DC	7	Chamber speaker	Minister
9. Carlo Azeglio Ciampi	1999	-	7	-	PM
10. Giorgio Napolitano	2006	DS	6.9	Chamber speaker	Minister
11. Giorgio Napolitano	2013		1.7
12. Sergio Mattarella	2015	PD (until 2009)	7	Committee chair	Deputy PM
13. Sergio Mattarella	2022		1 (as at February 2023)

Parties are as follows: UDN, Unione democratica nazionale (National Democratic Union); DC, Democrazia cristiana (Christian Democracy); PSDI, Partito socialista democratico italiano (Italian Democratic Socialist Party); PSI, Partito socialista italiano (Italian Socialist Party); DS, Democratici di sinistra (Left Democrats); PD, Partito democratico (Democratic Party).

Source: Tebaldi (2005, 118) and own updates based on information from www.presidenti.quirinale.it.

Scholars understand the president’s political background as a possible indicator of the (changing) relationship between the head of state and the political parties, in that the election of presidents with a party affiliation and experience within political institutions would indicate firm party grasp of the selection process (e.g. Tebaldi 2005).¹¹ Table 1 indicates little over-time variation in this respect. All but one president had been a party member before entering office and all of them had occupied prestigious institutional positions, either in Parliament or the cabinet (or both). However, Mattarella left his party in 2009, well before being elected, to become a Constitutional Court judge. Ciampi – prime minister from 1993 to 1994 and a cabinet minister from 1996 to 1999 – is the president who most deviates from this general pattern, in that he never had a party affiliation or a seat in Parliament before entering office.¹²

Irrespective of their political backgrounds, Italian heads of state are expected to behave as non-partisan, independent actors, representing the whole polity. Nevertheless, this does not mean that it is easy for majority and opposition parties in Parliament to agree on the name of such an impartial figure. Five (71%) of the seven presidents who were elected by both the majority and the opposition (i.e. Gronchi, Saragat, Pertini, Cossiga, Scalfaro, Ciampi, Napolitano II) entered office before 1994 (the year of the first general election with the new party system). Among the seven, the election of Napolitano in 2013 was more the outcome of a serious political stalemate in Parliament than of genuine consensus (Clementi 2014). Six presidents (Einaudi, Segni, Leone, Napolitano I, Mattarella I, and Mattarella II), in turn, were elected with the votes of the majority parties only, and three of them (50%) entered office before the 1990s. The empirical evidence suggests that over time the presidency has become more politically contested. Why?

3. The nexus between presidential power and party bargaining complexity

In parliamentary democracies, political parties are the key actors that make the chain of delegation work (Müller 2000): they recruit candidates to the legislature, select the prime minister and the cabinet, and define government policy. In other words, they are the principal vehicles that voters have for keeping their representatives accountable (Andeweg 2020). Nevertheless, the personalisation and mediatisation of politics have increased political leaders’ public visibility and their political clout at the expense of parties since the late 1980s. Popular approval has become more important than party support for individual leaders seeking to perform well in office; this also means that leaders enjoy broader leeway in shaping policy and running electoral campaigns (Poguntke and Webb 2018).

In this regard, Italy has long been a text-book case in Western Europe (Calise 2005; Marino, Martocchia Diodati, and Verzichelli 2022). The president of the republic is undoubtably one of the actors to have benefited most from these trends, in a context of severe party government decline (Mair 2008). Building upon the extant literature, this article claims that this process has changed both the way in which political parties¹³ deal with the figure of the head of state and, in turn, their strategic behaviours during the selection process.

There is empirical evidence that political parties pay particular attention to selecting trustworthy presidents, when presidential political influence is high and the prime minister has limited control of the cabinet agenda (Müller-Rommel, Vercesi, and Berz 2022, 166–167). The explanation is simple: political parties – which have a gate-keeping function – will try to select reliable agents, especially when the potential costs of later agency loss are high (Lupia 2003). In Italy, the strengthening of the president has been conducive to a greater presidential power to control – directly or indirectly – the political agenda of the government. In other words, heads of state have become more effective veto players, reducing parties’ abilities to change the policy status quo against their preferences (Zucchini 2013).

Parties, which want to maximize office and policy payoffs (Strøm and Müller 1999), will thus implement a thorough ex-ante screening of any presidential candidate to reduce the likelihood of adverse selection. This screening is even more crucial in Italy, where the duration of the presidential term does not coincide with the cabinet’s, and ‘cohabitation’ is frequent (Tavits 2008). In contrast, parties will avoid the costs of painstaking negotiations whenever the president’s role is merely ceremonial, because the potential loss from any subsequent moral hazard is minimal.

In this latter case, the choice of the president has relatively low salience for parties, because little is at stake. However, if parties anticipate possible conflicts with a powerful president, they will be less keen to compromise and conflicts over candidates will be more likely. As indicated by Andeweg and Timmermans (2008, 276), the potential for inter-party conflict is in fact a function of the salience of the point of contention. This implies that negotiations over a salient issue are more complex, assuming that complexity indicates the difficulty involved in reaching agreement.

Therefore, the first hypothesis reads as follows:

H1, selection process complexity:

the complexity of the presidential selection process will increase when the power of the head of state grows.

Such theoretical argument does not claim that the salience of the presidential selection increases because parties ‘perceive’ the salience of the choice and behave as if it were. Rather, the background condition is that presidents are actually stronger and rational parties simply have to behave consequentially, at least if they do not want to bear excessively high political costs.

That said, parties do not act in a vacuum and the context might have a conditional effect on how the selection process develops. All else equal, one may assume that compromises are easier to reach if parties have similar policy preferences. In that case, they will be more likely to agree on a presidential candidate who arguably embodies these orientations and who will not hamper their actions in the cabinet or the legislature. At the same time, to reach a compromise among many counterparts will be harder than agreeing among just a few parties (e.g. Leiserson 1968). Both the heterogeneity of policy preferences and the number of parties in the game can be considered defining aspects of the degree of complexity of the parliamentary bargaining environment (not of the selection in itself!).

The second hypothesis, which holds under an ‘all else equal’ condition, is that:

H2, bargaining environment complexity:

parties will be more likely to reach agreements on the selection of the head of state when the complexity of the parliamentary bargaining environment is lower.

The next section assesses these theoretical expectations.

4. Methodological choices

For the purposes of this article, one of the most substantial empirical challenges is to operationalise presidential power. Comparativists’ efforts to measure presidents’ actual influence have often been frustrated by the lack of reliable ‘objective’ indicators (Doyle 2020). In response to this problem, some have relied on expert surveys (e.g. Lowande and Shipan 2021; Coppedge et al. 2022). However, these attempts have been designed for cross-country analyses and – understandably – overlook the importance of idiosyncratic factors when it comes to single-country, fine-grained analyses.

In the following analysis, I adopt a halfway strategy, which provides enough empirical parsimony without impeding in-depth elaborations of particular events and phases. Based on the preceding discussion of the development of presidential power, I place the Italian presidents in three core categories, representing low, medium, and high power respectively. This tripartition coincides with three periods with three different opportunity structures for the exercise of presidential influence as discussed in section 2. It is important to note that the three levels of power are relative to one another rather than being absolute. Moreover, they refer to general patterns and do not capture the idiosyncratic and specific behaviour of single presidents.

The first category includes the heads of state elected before the 1980s (N = 6), that is, before the first signs of the personalisation and mediatisation of politics (Rahat and Kenig 2018). Presidents in the second group were in office during the transitional period of the presidency, which runs from Cossiga’s election in 1985 to the end of Ciampi’s mandate in 2006 (N = 3). Finally, in the third category one finds the double presidential terms of Napolitano and Mattarella (N = 4). The cut-off points coincide with those presidencies that are posited to be the most innovative relative to the past: in a nutshell, presidencies that changed the public understanding of the role of president in the years following. More precisely, Cossiga was elected after Pertini, who was the first example of a ‘popular’ (rather than party) president. Similarly, Napolitano succeeded Ciampi, under whom the notion of the president as a leader with the power (and the right) to use moral suasion to affect policy-making became dominant and widely accepted (Passarelli 2022). An important assumption behind this tripartition is that parties are rational actors that ‘learn’ from innovative presidencies and adjust their expectations and behaviours accordingly at the subsequent presidential election.

With regard to the complexity of the selection process, this concept is understood as multidimensional and defined by four aspects: the length of the process itself, the level of inter-party conflict, the ‘competitiveness’ of the race, and the difficulty involved in changing the presidential status quo. Growth along one or more of these dimensions contributes to increasing the complexity of the selection process. The length of the process is operationalised as the total number of electoral rounds before the final choice, while the percentage of votes obtained by the winner in the final round is posited to correlate negatively with the level of conflict (i.e. a high percentage indicates a high level of inter-party agreement). Third, the fragmentation of preferences in the first round indicates whether or not the election was competitive. Finally, decision-making stalemates and the ultimate re-election of the incumbent president is a simple proxy for the difficulty involved in changing the status quo (i.e. parties agree to disagree on any new name). Each election receives scores on the four dimensions, which contribute to the overall score of the index of (selection process) complexity.¹⁴ This index treats the variables as unweighted (see Musella and Vercesi 2019). For each dimension, a sub-index results from the sum of each v value on the respective i^th ranking, divided by the highest possible score n_v (see the Appendix for details of the construction of the sub-indices and the scores). Therefore, each sub-index I will be:

$I=\frac{{\sum }_i{v}_i}{n_v}$

The final selection process complexity score is the arithmetical mean of the values of the four sub-indices.

The small number of observations and the lack of random samples prevent us from performing any reliable statistical tests of the strength of the relationship between election years and selection complexity. However, the reliability of the findings is supported by the fact that the analysis is based on ‘not-repeatable data’, which ‘exhaust the population of substantive interest’ (Jackman 2009, p. xxxi). More prosaically, this means that – as long as the focus is on all the available presidents – the evidence is substantially informative for the time span under investigation.

Table 2 shows the scores for all Italian presidential elections (1948–2022) on each of the dimensions of selection process complexity as well as the overall score.

Table 2. Selection process complexity scores in Italian presidential elections, 1948–2022.

Election (year)	Length index	Confrontation index	Competitiveness index	‘Status quo’ index	Total complexity
1948	0.4	1.0	0.4	0.0	.45
1955	0.4	0.4	0.2	0.0	.25
1962	0.8	1.0	0.4	0.0	.55
1964	1.0	0.8	0.4	0.0	.55
1971	1.0	1.0	0.2	0.0	.55
1978	1.0	0.4	1.0	0.0	.60
1985	0.0	0.2	0.6	0.0	.20
1992	1.0	0.8	1.0	0.0	.70
1999	0.0	0.6	1.0	0.0	.40
2006	0.6	1.0	1.0	0.0	.65
2013	0.4	0.8	0.6	1.0	.70
2015	0.4	0.8	1.0	0.0	.55
2022	0.8	0.4	1.0	1.0	.80
Mean	0.6	0.7	0.7	0.2	0.55

Sources: own elaboration of data from Openpolis, Tebaldi (2005, 132), www.quirinale.it.

5. Findings: towards a more complex selection process

5.1. Selection process complexity

Applying the aforementioned index of selection process complexity, a first interesting finding appears: as expected, selection process complexity increases over time (Figure 1).

Figure 1. Selection process complexity over time, 1948–2022.

Points correspond to the presidential elections.

Source: see Table 2.

The highest level of complexity coincides with the second Mattarella election (2022), whose score is 0.8. Scalfaro’s election (1992) follows with 0.70, just above the value (0.65) for the first Napolitano election (2006).

However, contingent factors (such as crises or other exogenous events) may also be driving forces behind the fluctuation of the data. To minimise this possible source of ‘noise’, Figure 2 groups all scores by periods.

Figure 2. Selection complexity by presidential period.

Again, the empirical evidence is in line with the first expectation, in that the average complexity score (0.68) for the most recent period is significantly higher than those of the previous time periods (0.49 and 0.43, respectively). Moreover, the value for the 2006–2022 period is the only one above the mean for all elections (0.55).¹⁵ However, it is worth noting that the average score for the 1985–1999 period is not – contrary to expectations – higher than the one for the 1948–1978 period, but rather pretty similar. This finding is in fact strongly affected by the two elections of Cossiga (1985) and Ciampi (1999), which are outliers in the Italian context and need ad hoc explanations. In the former case, the ostensibly smooth decision-making process was the result of an intra-coalition competitive strategy on the part of the then largest majority party, the Democrazia Cristiana (Christian Democrats, DC), which looked for a direct agreement with the main opposition party, the Partito Comunista Italiano (Italian Communist Party, PCI). The DC’s coalition partners then had to accept this agreement. The latter case stemmed from the new setting of bipolar party competition, the substantial equilibrium between the two (internally fragmented) coalitions, and their decision to choose a politically neutral yet authoritative candidate (Tebaldi 2005, 143–146).

In a further step, Figure 3 classifies all presidents by level of actual power and complexity of the respective selection process. For the sake of simplicity, the three periods, from the earliest to the most recent, correspond to low, medium and high power; complexity, in turn, is low if the election score ranges between 0 and 0.4, medium if it falls between 0.4 excluded and 0.6, and high when values are higher than 0.6.

Figure 3. Presidents by power and selection process complexity.

In each cell, presidents are listed from the one corresponding to the lowest selection complexity to the one with the highest.

Source: see Table 2.

Despite the caveats, the findings are striking. No president with a ‘low’ level of power was chosen in a highly complex election and no president with a ‘high’ level of power was easy to select. All but one (Mattarella I) of the four most recent presidencies derived from a highly complex selection process. Yet, the concentration of presidents (all but Gronchi) with low power in the cell corresponding to medium complexity can be understood as a signal that, already in the first decades after the Second World War, party actors were aware of the president’s potential as a politically consequential actor. Therefore, they invested substantial energy in choosing appropriate candidates. However, when parties lost influence to the benefit of presidents, the process of selecting an appropriate candidate became even more crucial and, since then, party leaders have had greater difficulty in finding mutually satisfying figures. Finally, the particular case of Scalfaro is consistent with the transitional nature of the period in which he was elected. On the one hand, traditional parties were losing their grip on cabinet governance, new parties were still to come, and the president was likely to become an important king-maker in prime ministerial selection; on the other hand, the election started when the major anti-corruption operation, ‘Clean Hands’, had already affected governing parties (Verzichelli and Cotta 2000, 462).¹⁶

Does the specific parliamentary bargaining environment condition the complexity of the selection process?

5.2. Bargaining environment complexity

The second hypothesis is that, irrespective of the level of presidential power, it will be more difficult to achieve inter-party agreement if the bargaining environment within Parliament is complex. Following the literature on coalition politics, this article operationalises parliamentary bargaining environment complexity as a function of the number and size of parties as well as of the heterogeneity of policy preferences (e.g. Strøm, Müller, and Bergman 2008). In particular, it posits a positive correlation with the first and the third variable and a negative relation with party size. For example, having five parties, each occupying 20% of the parliamentary seats, will make it harder to find a majority to elect the president than having two parties with 35% of seats and three with 10% each. Indeed, in the latter scenario it is enough that the two largest parties reach an agreement, whereas in the former three parties are necessary.¹⁷

First, the number of parties is calculated by weighting them by their parliamentary size. In this regard, Laakso-Taagepera’s index of the effective number of parties (Laakso and Taagepera 1979) is a reliable indicator of the level of party fragmentation within Parliament. This index is the fraction between 1 and the sum of parties’ squared percentages of seats. When applied to the parliamentary distribution of seats, the index has no specific upper limit other than the number of seats itself (i.e. the theoretically highest possible level of fragmentation is when each party holds just one seat).

Second, I apply an index of bipartisanism as suggested by Chiaramonte (2015, 19) to assess further the level of concentration of power in Parliament. This index is the result of the sum of the percentages of seats occupied by the two largest parliamentary groups.

Third, I use party polarisation as a proxy for ideological heterogeneity (e.g. Keman 1997). In particular, polarisation is calculated as the range between the two extreme parties in Parliament on a left-right scale of policy preferences, which goes from 0 (extreme left) to 10 (extreme right). Although several other dimensions of party competition can characterise a political system, this analysis accepts the common argument that the left-right dimension is an effective heuristic reference for understanding party positions in a given country on all relevant issues (e.g. Sartori 1976; Benoit and Laver 2006).

Accordingly, Table 3 provides information about the complexity of the parliamentary bargaining environment for all presidential elections.

Table 3. Selection complexity and bargaining environment complexity in presidential elections, 1948–2022.

Period and president	Selection process complexity	ENP (+)	Bipartisanism index (-)	Polarisation (+)
Period 1 (1948–1978)
Einaudi	0.45	2.87	85.0	6.9
Gronchi	0.25	3.54	68.8	7.6
Segni	0.55	3.44	69.3	7.6
Saragat	0.55	3.74	67.6	7.6
Leone	0.55	3.64	70.3	7.9
Pertini	0.60	3.16	77.9	8.7
Mean	0.49	3.40	73.2	7.7
Period 2 (1985–1999)
Cossiga	0.20	4.33	67.1	8.7
Scalfaro	0.70	5.71	49.7	8.3
Ciampi	0.40	6.69	46.7	7.2
Mean	0.43	5.58	54.5	8.1
Period 3 (2006–2022)
Napolitano I	0.65	5.32	51.6	7.2
Napolitano II	0.70	3.59	61.9	7.5
Mattarella I	0.55	3.54	63.0	7.5
Mattarella II	0.80	6.45	46.0	7.5
Mean	0.68	4.73	55.6	7.4
Total (1948–2022)	0.55	4.31	63.5	7.7

Data refer to the Italian lower house (Chamber of Deputies); ENP indicates the ‘effective number of parties’ in Parliament; the expected relation with selection process complexity is shown in parentheses (+ is positive and - is negative).

Data concerning the ENP have been drawn from Zucchini and Pedrazzani (2021: 400-404) and own updates. Party size is taken from the Historical Portal of the Italian Chamber of Deputies (https://dati.camera.it/) and, from the first (Einaudi) to the 15th legislative term (Napolitano I), data refer to the beginning of the term. The level of polarisation is based on left-right party positioning as indicated in the ‘viewcalc_party_position’ dataset (updated to 29 April 2022) by Holger et al. (2022). The missing value for the Movimento Cinque Stelle (Five-star Movement) is taken from Vercesi (2021: 21). For the election of Einaudi, the value for the Fronte democratico popolare (People’s Democratic Front) is the mean of the values of its two constituent parties, the PCI and the PSI.

Interestingly, the complexity of the selection process in all three periods does not seem to be associated with any of the measures of bargaining environment complexity.18

Therefore, the results do not support the second expectation. At the same time, they further corroborate the plausibility of the first hypothesis: the actual level of presidential power in the political system seems to be conducive to a more difficult and intricate selection process, irrespective of the level of complexity of the bargaining environment within Parliament.

6. Conclusion

This article has tackled an important yet neglected aspect of the impact of heads of state in parliamentary republics: the way in which their informal power can complicate inter-party negotiations. Study of this topic is particularly important for an understanding of party behaviour in Italy, since the president has been acquiring a substantial role in cabinet formation and policy-making since the breakdown of the first republican party system. While for several decades presidents had hardly any influence in day-to-day politics and a mostly ceremonial role, the concurrence of (intertwined) country-specific events and cross-country trends such as the personalisation and mediatisation of politics has provided Italian heads of state with significantly greater actual power. This, in turn, has made their selection a key political decision, which can have major consequences for parties’ future success in terms of office, policy and electoral support.

The article’s main argument is that the growth of ex-post presidential influence on party politics is reflected in increasing complexity of the process of selecting the head of state in Parliament. Parties are less likely to move away from their ‘kernel preferences’ and to compromise, because the potential risks of a mistaken choice have increased. The empirical evidence we have gathered is in line with this argument. The analysis suggests that the theory also holds after ‘controlling’ for the effect of contextual factors, such as the characteristics of the parliamentary bargaining environment.

These findings have implications for the understanding of party leaders’ actions and party bargaining. In particular, they highlight that the reshaping of the balance of power between political institutions is important for making sense of changes in the behaviour of party actors in a constitutionally unaltered context. Comparativists have mostly addressed the impact of long-term trends on the increasing power of individual leaders vis-à-vis political parties in studies of prime ministerial ‘presidentialisation’. This article contributes to this literature as well as to the scholarship on presidential powers by introducing a different yet complementary perspective.

Future studies could extend the geographical focus of this work and systematically compare parliamentary democracies both longitudinally and cross-sectionally. Moreover, scholars could seek to compare semi-presidential and parliamentary democracies using the same conceptual framework, a research path that has so far received only limited attention.

In an era of the personalisation, mediatisation, and presidentialisation of politics, political systems are experiencing the growing power of individual leaders and monocratic institutions. Political parties, in turn, are struggling to adapt themselves to the new conditions. This article suggests that this holds also with reference to constitutional actors who have long been considered relatively powerless.

Acknowledgements

I am grateful to the anonymous reviewers as well as to Fortunato Musella and Marco Valbruzzi for their valuable comments on earlier drafts of this article. A first draft was presented at the workshop on The President of the Republic in Turbulent Times: Election and Prospects, held at the Casa Orlandi Centre (Anacapri) of the University of Naples Federico II on 27-28 June 2022. A second version was presented at the annual Conference of the Italian Political Science Association, held at Roma Tre University on 8-10 September 2022. I also thank Francesco Amoretti, Donatella Campus, Annarita Criscitiello, Domenico Fracchiolla, Luca Verzichelli, and Francesco Zucchini for their helpful suggestions.

Disclosure statement

No potential conflict of interest was reported by the author.

Additional information

Notes on contributors

Michelangelo Vercesi

Michelangelo Vercesi is Researcher in Comparative Politics at the Portuguese Institute of International Relations of the NOVA University Lisbon and Privatdozent in Political Science at the Centre for the Study of Democracy of the Leuphana University Lüneburg. His research focuses on comparative government, political elites and leadership as well as political parties. Before moving to Portugal, he researched and taught in Austria, Germany, Italy and the United Kingdom. His recent book is Prime Ministers in Europe. Changing Career Experiences and Profiles (Palgrave Macmillan), coauthored with Ferdinand Müller-Rommel and Jan Berz.

Notes

1. Parliamentary democracy is defined by a political executive accountable to a directly elected parliament, whose confidence is necessary for the prime minister and her or his cabinet to stay in office. Non-monarchical heads of state are not popularly elected.

2. For simplicity’s sake, the article uses selection and election as synonyms, to refer to the process whereby the parliament (and other relevant delegates) choose a new head of state. In fact, the ‘selection’ of presidents is fulfilled through their ‘election’.

3. Henceforth (presidential) power, influence, and strength are used as synonyms.

4. On a more specific note, contingent political and economic crises played a role as well (Grimaldi 2012).

5. But not if compared to other heads of states. According to the ‘Prespow2’ index of presidential power devised by Doyle and Elgie (2014), the constitutionally strongest head of state in the European Union is the president of Cyprus (which is a presidential system), followed by the Portuguese and French presidents. The Italian president is the strongest among those of parliamentary democracies. The ‘Prespow2’ index aggregates 28 previous indices and, relative to the ‘Prespow 1’ index by the same authors, displays a lower range of standard errors among European countries; moreover, ‘on balance the reliability of the whole set of scores is probably slightly greater’ (Doyle and Elgie 2014, 739). This article does not use this or any other index of formal power, due to its contrasting conceptualisation of power. See Section 4 for a discussion of this issue and for the operationalisation of presidential power in Italy.

6. The position was first occupied by Enrico De Nicola, who remained in office from 1 January 1948 (the date of entry into force of the republican constitution) to 12 May 1948. However, he became president automatically, after being elected by the Constituent Assembly. The first president to be elected according to the new constitution was his successor, Luigi Einaudi.

7. This statement recalls the theory of the so-called ‘presidential accordion’, advanced by Pasquino (2012, 2020, ch. 6). In a nutshell, this theory says that the (constitutional) powers of the president are like an ‘accordion’, which opens up when the political parties are weak and is shut when the parties gain ground. While this article claims that the Italian president has become more influential as the result of long-term structural changes in the political system, contingent situations can therefore make a president, under similar structural conditions, relatively stronger or weaker than others. This explains, for example, the important role played in particular circumstances in the early post-war decades by some presidents such as Gronchi and Saragat (see Passarelli 2022). In fact, even weak presidents can, in specific circumstances, play a significant role in Italian political life, and vice versa.

8. Including foreign policy. It suffices to recall the proactive role of President Giorgio Napolitano in defining the type of Italian military intervention in the Libyan civil war in 2011. See Dino Pesole, ‘Napolitano: adesso serve coerenza con gli impegni presi.’ Il Sole 24 Ore, 28 April 2011. The ability to influence foreign policy is the president’s key ‘behavioural power’ according to the well-known Siaroff (2003) power index.

9. The Italian head of state can refuse to promulgate a bill and send it back to Parliament for further deliberation. However, if the bill comes back to the presidency, then it must be promulgated without further delay (article 74 of the Constitution).

10. According to the original text of the Constitution, the legislative term of the Senate (upper house) was six years, but it was aligned with that of the Chamber (lower house) in 1963.

11. Similar arguments appear in works about presidents in parliamentary democracies characterised by strong party government (e.g. Bergmann 2022).

12. Like Einaudi from 1945 to 1948, Ciampi was Governor of the Bank of Italy between 1979 and 1993.

13. The focus is on parties because the parliamentary party groups are the main actors in presidential elections, and regional delegates vote along party lines as well. The theory is silent about the internal organisation of parties. Since the early 1990s, personal parties have mushroomed in Italy. Often, these parties have thin organisations and are dependent on their leaders in many respects (Pasquino 2014; Vercesi 2015). However, this does not undermine the viability of the theoretical argument, as long as these leaders maintain party discipline; in this case, parties can still be considered as the main actors of the selection process, although their leaders, rather than collective bodies, set the voting guidelines. Cases of voting defection and drifting away from the party’s parliamentary party group can be considered as episodes of intra-party conflict, which indicate low cohesion within the party and are covered by the third and fourth dimension of complexity as introduced in the empirical analysis (see Appendix).

14. I am aware that problems of ‘collinearity’ between the first two variables can derive from the requirement of the two thirds majority to elect the president at the first three rounds. To minimise these problems, I assign different conflict scores to the same percentages, depending on whether or not these percentages were reached before the fourth round (see Appendix).

15. Selection complexity was higher than the mean in the elections of Pertini, Scalfaro, Napolitano I, Napolitano II, and Mattarella II.

16. Scalfaro was chosen as a key candidate only in the last round (16^th), after Judge Giovanni Falcone, his wife, and bodyguards were killed by a Mafia bomb. This event prompted the parties to converge on a new name (e.Pasquino 1993).

17. Henceforth, I will take into consideration only the parliamentary party groups, excluding the regional delegates involved in presidential elections.

18. General correlations between, on the one hand, selection process complexity and, on the other hand, fragmentation, bipartisanism, and polarisation have an R of 0.258, −0.3854 and −0.1328 respectively. In the latter case, the direction is against expectations: see Table 3. None of these relations are statistically significant at p < .10.

19. Since many voters tend to cast a ‘blank ballot’ at the first round, it is not possible to apply indices of voting fragmentation – such as Rae’s (1971) – without focusing exclusively on voted candidates; this would have distortive effects on the results.

Appendix

This Appendix describes the operationalisation of the dimensions of selection process complexity in detail.

With regard to the length of the selection process, the following scores apply: one round is equal to zero; 2-3 rounds = 1; 4-5 rounds = 2; 6-7 rounds = 3; 8-9 rounds = 4; 10 or more rounds = 5. The threshold of 9 to set the highest value is based on the overall mean, which is in fact nine rounds.

The degree of conflict ranges from 0 to 5, being 0 if the percentage of votes reaches at least 90%; 1 if it is equal to 75% (i.e., 3/4) or higher by the third round included; 2 if the same percentage is achieved from the fourth round onwards; 3 if the percentage is between 66.6% (i.e., 2/3) and 75% before the fourth round or 4 if after the third; finally, the value reaches 5 if the percentage is below 2/3 of voters.

The third sub-index ranges from 0 to 5. If only one candidate obtained valid votes in the first round, the competitiveness would be null. Between 2 and 5 candidates, the score is 1; between 6 and 9 it is 2, while a score of 3 corresponds to 10-13 voted candidates. Finally, a score of 4 is attributed to a number between 13 and 16 included, and 5 for all numbers above 16.¹⁹

Finally, the difficulty in changing the presidential status quo ranges from 0 to 2: 0 means that a new incumbent is elected; 1 that a previous incumbent is elected by the third round; 2 that he or she is elected from the fourth round onwards.