The Italian PD is well suited for exploring the above-mentioned questions since it has involved its affiliates and supporters in open primaries for almost two decades now, and the party involves both party supporters and enrolled members in internal decision-making (Pasquino and Valbruzzi 2016; Sandri and Seddone 2015). Also, the PD is the only Italian party that adopted both democratization reforms and primary elections during the past few years. The PD was created in 2007 with the merger of the leftist, post-communist party ‘Democrats of the Left’ (Democratici di Sinistra, DS, previously called ‘PDS’) and the centre-left party ‘The Daisy’ (La Margherita) (see Bordandini et al. 2008). What makes the PD case particularly interesting for the purposes of this paper is the fact that the party is the heir of both the Catholic and communist political cultures and of their mass-based organizational settings. Moreover, the PD is currently the most important left-wing Italian party. Its central offices operating at national level are organized quite traditionally, adapting the old mass-based party structures to the 21^st century context. In fact, the party is led by a national secretary, the National Assembly is a large board in charge for the main decisions, while the National Direction and the executive committee are smaller councils appointed to manage the routine party life.

If compared to other Italian parties, the PD has not only reserved a broad set of rights and privileges for its formally enrolled members, but also attributes a few important rights to the above-mentioned ‘electors’, which correspond to the category of ‘supporters’ or ‘sympathizers’ in other European parties (Scarrow 2015). While no Italian party provides the possibility for its members to directly participate in party conventions nor in the selection of party central executive bodies, the PD’s affiliates can select the delegates for the party convention. In addition, the PD’s statutes provide the possibility for party members to select - via closed primaries - representatives for the National Assembly, the legislative body of the party. Moreover, the PD’s enrolled members are endowed with the right to directly participate in the formulation of the electoral manifesto or other policy positions. The PD’s statutes also entail the possibility of holding internal referenda on specific policy issues. The party similarly gives its members the right of receiving on a regular basis information on party activities, strategies and mandates. Concerning members’ obligations, the PD also details the duties of its affiliates in its statutes, such as the obligation for all members to get involved in internal and electoral campaign activities. Comparatively speaking, while Italian parties feature significant variations in their organizational settings, especially regarding the role of members and intra-party democracy patterns, the PD is the party that gives more privileges to its members (at least until the recent revisions to the Five Star Movement’s internal statutes). Using 2015 data from Sandri and von Nostitz (2021) on five Italian parliamentary parties ¹ we evaluated the extent of the PD’s affiliated members’ rights and obligations. Figures 1 and 2 show the two indexes for measuring rights and obligations allocated to full-members: these variables are measured through two cumulative indexes of obligations imposed (0 to 4) and privileges distributed (0 to 8) to formal members (Cronbach’s alpha: 0.26 and 0.77). The higher the individual score, the more extensive rights and/or more obligations full-members have. We look at the following rights: right to stand as candidate for elected office, to attend the party congress without being a delegate, to vote at the party congress, to call the party congress, to select the party national executive organ, to select parliamentary candidates, to select presidential or prime ministerial (or, generally, chief-executive) candidates, and to select the party leader. We explored the following obligations: to pay an inscription fee; to go through a probation phase, to adhere to party rules/principles; and to be involved regularly in party activities. These indexes provide a valuable proxy concerning the level of both intraparty democracy, parties’ organizational models and the relationship between members and leadership.

While the PD’s sympathizers do not enjoy the same amount of rights and obligations of affiliates, they can participate in most party activities, they are also involved in internal ballots such as those for selecting the party candidates for election of leader, and they can stand as candidates for elected office for the party. This suggests that, while intra-party democracy measures such as open primaries can be beneficial for improving the party image and mobilizing support between elections, this kind of degree of organizational permeability and the growing evidence of recurrent pathologies related to the use of primaries by Italian parties might lead to forms of ‘plebiscitary’ intra-party democracy (Scarrow et al. 2017).

What is significantly innovative in the party’s organizational structures is the degree of internal democracy and more specifically the leadership selection method (Fasano and Natale 2017; Pasquino and Venturino 2010; 2014; De Luca and Fasano 2018; Rombi and Serricchio 2019). First, the statutes state that the access to all party offices is decided by ‘members and electors’ (Venturino 2015: 36). By so doing, the PD has gone beyond the solutions adopted by many European parties during the ‘democratic revolution’ begun in the Sixties. Since then, several parties have made increasing use of internal democracy, giving a say to their members in their internal affairs. By enfranchising both members and electors, the PD has given a say in crucial internal matters to both - very different - grassroots groups and encouraged adopting a larger than usual leadership selectorate. Moreover, the label ‘electors’ in the party statutes identifies a set of people that is larger than the number of citizens eligible to vote in parliamentary elections, as it includes citizens aged 16 years and older, as well as regularized immigrants. These ‘electors’ must register as party ‘supporters’ on primary election day, with the party keeping a more or less updated supporters’ inventory (Sandri and Seddone 2015). In practice, through these provisions the PD has institutionalised the use of open, American-style primary elections.

Open primaries have been routinely used by the PD to select both candidates for elections and officials for party roles. Referring to the latter, the national party leader - known as Segretario Nazionale - is appointed for a four-year term renewable only once. For this selection, an original mixed method is envisaged involving three different selectorates ² . First, aspirants endorsed by 10 per cent of the outgoing National Assembly, or alternatively by at least 1,500 enrolled members, are voted by the whole membership. The three most supported aspirants are admitted to the following open primaries, given that they have gained 5 per cent of the members’ votes ³ . Second, all electors are enfranchised to vote in the open primaries, which are not used to directly select the leader. Rather, primary voters cast a ballot to elect the National Assembly, a permanent party board that here has the role of an electoral college. To allow voters to express their preferences, aspirant leaders promote closed and gender-balanced list of candidates, competing through a PR electoral system. The third and final selector is the National Assembly, which can play a different role according to the results of the primaries’. Had an aspirant gained a majority of delegates, then s/he is simply acknowledged as the new party leader, while if the most voted aspirant gained less than 50 per cent of delegates then a runoff takes place between the top two aspirants. In this circumstance, the role of the National Assembly as party leader selector would be far from ceremonial, but this case has never occurred.

The activation of an inclusive selectorate of thousands or millions of potential voters requires a greater effort by the involved party. Unquestionably, this is the case with the open primaries, which need months to be planned, organized, and finally held. Thus, primaries cannot be used in emergency, and political parties based on an inclusive method of selection usually empower an alternative and smaller selectorate to quickly pick a new leader when under pressure, which is often a party council. This is the case of the PD’s statutes, which give the National Assembly an important power. Firstly, a majority of the delegates may deselect the serving leader before of the end of his/her four-year term, and in such a case a new selection is immediately called to elect a new Assembly and a new leader. Secondly, and of greater relevance here, if the serving leader gives up the Assembly, they are able to elect a new leader to conclude the current term; if the leader resigns after a disagreement with a party council, then a two-thirds majority is requested.

It should be noted that in 2019 the National Assembly changed some relevant rules concerning the party leader selection process. The new set of rules will be applied for the next leader selection scheduled in 2023 ⁴ . However, the general framework for leader selection is maintained even after the current reform.

The leaders of the PD may be then selected through open primaries, which constitutes the usual process, or alternatively by a party council under exceptional circumstances. Table 1 details the Democratic leaders who have served since the party’s launch in 2007 and their actual selectors ⁵ . In the period between 2007-2021, due to the two mandates of Matteo Renzi, nine leaders alternated for ten terms. Notably, no leader has so far completed his four-year term. As discernible in the last column, this exceptional leadership instability is due to the electoral failures upsetting many European mainstream parties in the last decade. Moreover, the PD’s leadership instability may also be explained by the challenges of managing a heavily factionalized party with few organizational incentives to keep the intraparty conflict under control (Floridia 2019). For instance, Walter Veltroni resigned after a defeat at the regional election in Sardinia, which however came in a row of negative results, including the 2008 parliamentary election. Pier Luigi Bersani also gave up after the disappointing result in the 2013 general election (Seddone and Venturino 2015). Matteo Renzi - then also Prime Minister - first resigned in 2017 when a referendum rejected the constitutional reform he was strongly committed to, and then in 2018 after another unsuccessful parliamentary election ⁶ . Similarly, in 2020 Nicola Zingaretti resigned in March 2021 due to intra-party factional conflicts and pressures.

Table 2 reports some descriptives on the leader selections organised by the PD so far. Beside the first race gained by Walter Veltroni in 2007, disputed under atypical rules, the role of party members has been negligible in some cases. In fact, in 2009 and 2017 only three candidates were fielded from the first stage, therefore no screening took place, even though in principle it could occur that a candidate fails to reach the 5 per cent threshold, and therefore only two are then enabled to run the second round. Instead, in 2013 and 2017 respectively one and three aspirants have been barred by the members’ vote. The last column of Table 2 shows that the front-runner has always won a majority of votes in the open primaries. This evidence seems to support the claim by Kenig (2008) that inclusive selectorates are correlated with a low level of competitiveness. However, there are remarkable differences between the very large percentage gained by Veltroni in 2007 and the lesser margin won by Bersani in 2009. Moreover, in three out the four comparable cases the percentage of the elected leader is larger in the open than in the closed primaries stage. The 2009 selection won by Pier Luigi Bersani is once more the exception, as the candidate has been voted by the party members more than by electors at large.

By adopting a longitudinal perspective, Figure 3 details the size of the PD’s party membership and open primaries’ participation. In the space of approximately twelve years, the PD has reduced its number of affiliates by half. In 2007, with nearly one million members, the PD was one of the largest organizations in Western Europe. This figure accounted for the two founding parties’ registered members - Democratici di Sinistra (DS) and La Margherita. However, at the end of the first recruitment campaign promoted by the new-born party, the number of affiliates has diminished by about 15%. Furthermore, since then, figures present a progressive (and not that slow) drop. To be precise, between 2013 and 2014, the party suffers from a severe decline in the number of enrolments, with a loss of about 30% of members. After 2014, data seem to stabilize but still point to a dramatic failure in recruiting new affiliates.

Figure 4 reports data on national turnout in closed primaries. The number of registered members participating in the first phase of the leadership selection has steadily decreased since 2009. In particular, the number dropped by more than half. Furthermore, the participation rate ⁷ confirms that the loss of mobilization capacity in closed primaries is not just a mechanical effect of the general crisis in recruiting (and maintaining) affiliates. Besides the drop in the absolute number of registered members, figures suggest a growing disaffection among militants: while in 2009 about 57% of the party members cast their ballot in closed primaries, after ten years this quota falls to about 42%.

Figure 5 focuses on open primaries and confirms a similar pattern. In 2007 the unexpected success in the open primaries - with about 3.5 million participants - has been interpreted as a support for this inclusive tool, at that time a true novelty in the Italian political environment (Sandri and Seddone 2015). Nonetheless, the initial enthusiasm faded away quite rapidly. While in 2009 we can see minor signals of a decrease in turnout, the declining pattern became extremely evident in 2013 - when for the first time the total number of selectors fell below three million - and is confirmed in 2017 with a further loss of about 1 million voters. The decrease in participation continued in 2019, with only 1.5 million participants. Looking at the participation rate ⁸ , the picture changes a bit. While the primary event is losing some attractiveness among members, sympathizers - despite some fluctuations - still maintain a certain interest in participating in leadership selection. In 2007, slightly fewer than 30% of the party electorate participated in open primaries. After only two years the turnout dropped, but more than a quarter of the PD voters took part in the 2009 open primary election. Participation rates peak in 2013 when almost one-third of the electors were involved, but a few years later, in 2017, participation rates decrease to 21.4%. A positive signal comes in 2019, when the participation increases back to 2009 levels.

In general, the data suggests a declining mobilization response among both registered members and sympathizers. Figures, indeed, show that the number of participants in both the primaries’ stages are constantly decreasing, and the same applies to the rate of participation. In closed primaries, the quota of registered affiliates participating in the first step of the leadership selection is below half of enrolled members. It should, furthermore, be noticed that this ratio is calculated on a smaller base - given the crisis in membership recruiting. Likewise, the participation rate in open primaries has to be interpreted in the light of a massive loss of voters in general elections.

The level of competitiveness is usually used as an estimate of the divisiveness brought by a leadership selection within the party organization. Competitive races are said to harm intra-party cohesion. When the voting shares of the two most voted candidate are close, a large number of the loser’s supporters may develop a sentiment of disgruntlement, leading them to voice their disappointment or even to exit the party by supporting other parties (e.g., Venturino and Seddone 2020; Wichowsky and Niebler 2010; Kenney and Rice 1987; Southwell 1986; Hacker 1965). However, here we look at competitiveness from a different perspective. In the context of the multi-stage procedure involving different selectorates portrayed above, competitiveness allows the exploration of the congruence between members and sympathizers. When these two selectorates are aligned, and their voting patterns supporting the leader are congruent, the party cohesion will be strengthened. As closed primaries are the first step of the leadership selection procedure, open primaries determine the result to be acknowledged by the party national assembly. This implies that, in case of divergence between the two main selectorates, the open primary outcome prevails while party members lose their voice. Likewise, different competitive patterns suggest that registered members and sympathizers are not aligned in their view on party leadership with implications at the organizational level. In closed primaries ending with a small margin between candidates and plebiscitary outcomes in open primaries, there is the risk of organizational divisions and internal conflicts among affiliates.

As mentioned above, within the range of the many indexes provided in literature for measuring the competitiveness, we opted for the closeness ⁹ , calculated in each region as the difference between the vote share obtained by the national winner and the regional runner-up. Figures 6 to 9 present the degree of closeness in closed and open primaries held, respectively, in 2009, 2013, 2017, and 2019 ¹⁰ . Starting with Figure 6, it is easy to see that in 2009, with few exceptions, the margins between the winning candidate and the runner-up are quite similar in both closed and open primaries. In 2009, both party members and sympathizers aligned in supporting the same candidate - Pier Luigi Bersani - at the party’s head.

This pattern is totally reversed when looking at closeness referring to the 2013 leadership selection (see Figure 7). The plebiscitary consensus obtained by the winning candidate from the inclusive selectorate does not correspond to closed primaries’ competitive scenario. Indeed, the closeness in open primaries ranges between 24.6 percentage points (Basilicata) and 65.5 percentage points (Marche). Instead, party members’ backing is by far less cohesive and supportive of the winning candidate. Even if Matteo Renzi is confirmed as the winner in all regions (with the only exceptions of Valle d’Aosta, Molise, and Basilicata), the comparison between closed and open primaries reveals that the two selectorates did not align on a common pattern. Interestingly, in 2017 - when Renzi ran for a second time as the incumbent party leader - the picture is (again) changed (see Figure 8). The two selectorates converged, displaying quite similar patterns of competition - Puglia being the only exception in this case ¹¹ . Moreover, the closeness level is similar in all regions, with large margins between the winner and the runner-up. The landslide success obtained by the incumbent hints at the re-alignment between party members and sympathizers and suggests that the leader could rely on a large legitimacy from both inside and outside the party.

Nonetheless, this was quite a precarious conjuncture. As mentioned above, Renzi’s second term at the head of the party ended with his resignation. In 2019 a new leadership race was organized. Nicola Zingaretti was appointed as the head of the party. Figure 9 shows a more nuanced picture, where affiliates and sympathizers present quite different voting patterns. The levels of competitiveness in closed and open primaries suggests that the two selectorates did not converge. The large margins between the winner and the runner-up confirm robust support for the winning candidate among sympathizers. With regards to closed primaries, party members, instead, diverge as demonstrated by the lower scores of the closeness indicator in most regions. The plebiscitary drive characterizing the inclusive selectorate is not congruent with the voting patterns among affiliates.